Last call: Apply for the Autumn 2026 Baiguan China Tour before this Sunday (Sep 13) to claim early-bird discounts.

Check out the last remaining available trips here.

Recently, we have invited several industry experts to talk about robotics, new energy, and supply chains in China. Today, we have invited another real industry heavyweight, who goes by “Dr. Joe,” to talk about one of the most strategically important industries of today: semiconductors.

Dr. Joe was an early co-founder of perhaps the most consequential semiconductor foundry business in China, and is still actively involved in the industry.

In this conversation, Dr. Joe will share an insider account of exactly how far China’s semiconductor industry still lags behind the Western world and what a self-sufficient Chinese semiconductor industry would mean for the rest of the world.

Baiguan: Most discussions of China’s semiconductor industry begin from the premise that its foundries still lag behind TSMC. But people disagree about the size and cause of that gap. Is the main constraint capital, talent, equipment, process know-how, or simply the time required to accumulate experience?

Dr. Joe: There is no simple answer because there are many kinds of semiconductor products. When people speak of “the gap,” they are usually referring to leading-edge CMOS technology, particularly at the five-nanometer and three-nanometer nodes. In that specific area, the single biggest and perhaps the only constraint is access to EUV lithography equipment, which is not available in China.

But most semiconductor products do not require EUV. In those areas, China is moving very quickly, and in some cases it is already leading. CMOS image sensors are one example. The major global suppliers include Sony, Samsung, and OmniVision, while Chinese companies have also become highly competitive in specialized applications. In areas such as security cameras and satellite imaging, some Chinese CIS products are already among the best in the world.

The picture is therefore very different from the common perception that China is behind across the board. In many specialized segments, the gap has already reversed. China’s semiconductor exports have also reached record levels over the past three years. The real lag is concentrated in leading-edge CMOS manufacturing, not in the semiconductor industry as a whole.

Baiguan: So your argument is that, wherever EUV is not essential, China is either already competitive or likely to become competitive relatively soon. The remaining gap is concentrated at the most advanced logic nodes.

Dr. Joe: That is correct. The clearest limitation today is advanced CMOS at five nanometers, three nanometers, and beyond. Even there, however, alternative architectures are becoming more important. Huawei’s Tau Scaling Law, 3D integration, and heterogeneous integration are all attempts to improve system performance without relying exclusively on conventional transistor scaling.

In the near term, these approaches can produce chips with performance comparable to products made with more advanced process nodes. They do not mean that EUV will never be needed. But they can reduce the importance of EUV for a period of time and create other routes to comparable system-level performance.

Baiguan: When you say China will eventually need EUV, how far away is “eventually”?

Dr. Joe: My prediction is that within five years, and possibly sooner, China will have an EUV solution or an alternative technology that serves the same purpose.

Baiguan: That’s a strong statement.

Dr. Joe: It is important to remember that the current ASML architecture is not the only possible way to generate EUV light. Elon Musk has recently signaled interest in free-electron laser, or FEL, technology for the proposed Terafab in Texas. This would still use EUV wavelengths, but the source architecture would be different. Instead of pairing an individual light source with every lithography machine, a very large accelerator-based or synchrotron-like source could potentially serve an entire fab, or even several fabs.

Both China and the United States are exploring this general direction. There is no law of nature that says advanced lithography must follow ASML’s present technical path. ASML’s system is the result of decades of successful engineering and iteration, but other engineering solutions are possible. ASML itself is a Dutch company, not an American or Chinese one, which is another reminder that this remains a global technological competition. That is why I am optimistic that EUV will become less of an absolute bottleneck within five years.

Baiguan: In other words, public debate may be too fixated on one company and one present-day architecture.

Dr. Joe: Exactly. Technology is not static. The semiconductor industry works with roadmaps that extend five to ten years into the future, and there are usually several proposed solutions to each major technical problem. The market ultimately decides which approach wins. Outsiders often take the most impressive technology of the present moment and assume it will remain the only viable path indefinitely. That is rarely how this industry develops.

AI chips, the memory wall, and 3D integration

Baiguan: Is this progress in alternative chip architectures one reason Chinese AI models such as Kimi, GLM, and DeepSeek have advanced so quickly?

Dr. Joe: That is a separate issue.

An AI model is primarily an achievement in algorithms and software. Hardware is another part of the system. Both the United States and China combine computing infrastructure with model development, but they do not necessarily have the same balance between the two.

If access to the strongest chips is limited, developers can compensate by simplifying models and reducing the amount of computing power required. That is part of what Chinese model developers have been doing, up until now.

If more advanced 3D and heterogeneous-integration chips become available, however, they will give those developers a stronger hardware platform and allow them to build more capable models.

Baiguan: So the potential of these alternative hardware technologies has not yet been fully reflected in the performance of Chinese AI systems?

Dr. Joe: Not yet. I expect the first important results to appear next year. I am directly involved in the development of a chip based on 3D integration, so I have been following this very closely.

NVIDIA’s GPUs are extremely powerful, and the memory capacity around them is very large. But the movement of data between memory and the processor remains constrained by what the industry calls the “memory wall.” A GPU may spend a large share of its time waiting for data rather than performing calculations. My own estimate is that this waiting time can reach roughly 70 percent in some workloads.

If 3D integration substantially shortens those data paths, much more of the GPU’s theoretical computing capacity can be utilized. In my view, a well-designed system based on this approach could match or exceed the performance of NVIDIA’s highest-end chips, even if the underlying logic is not manufactured on the world’s most advanced process node.

Baiguan: Jensen Huang has made the point that the rest of the industry is not standing still either. TSMC and other leading companies are also investing heavily in advanced packaging and 3D integration. Does China really have an advantage here, or is everyone moving in the same direction?

Dr. Joe: Everyone is moving. The question is not whether other companies are progressing, but who is ahead in a particular architecture and who controls the relevant intellectual property.

For instance, today’s HBM systems are generally based on 2.5D integration and chip-to-chip connections. The next step is wafer-scale 3D integration using wafer-to-wafer bonding. That is a more advanced form of integration. I believe the industry will need wafer-to-wafer technology by the HBM5 generation.

There is at least one Chinese company with important intellectual property in wafer-to-wafer stacking. I do not want to name the company here, but it is an example of a field in which a Chinese company may be ahead rather than behind. If the rest of the industry eventually adopts the same route, major international memory companies may need access to Chinese intellectual property.

Same thing is actually happening in memory. Historically, DRAM cell area moved from roughly 12F² to 9F² and then to 6F². The next generation is 4F², which is close to the theoretical lower limit for a conventional DRAM cell. Samsung, SK hynix, Micron, CXMT, and other companies are all working on this.

One company in China announced a 4F² product in 2024 and began delivering products in 2025. Samsung, by comparison, is targeting production around 2028. If that timetable holds, the Chinese company will be two or three years ahead. Combine 4F² cells with wafer-to-wafer integration, and China could become a leader in some next-generation DRAM technologies without first solving every problem in EUV lithography.

Baiguan: It doesn’t seem this type of information gets reflected in today’s market consensus at all.

Dr. Joe: Indeed. But engineers who attend the relevant technical conferences can see these developments, but many people in finance and the broader public still use an older picture of the industry. I expect that picture to change substantially over the next two years.

Baiguan: How much does China’s wider manufacturing ecosystem contribute to this progress?

Dr. Joe: China is already the world’s largest manufacturing center, and semiconductors are the most technologically advanced extension of that base. Different provinces have their own semiconductor programs, while the national Fifteenth Five-Year Plan identifies three especially important four-character concepts: 存算一体computing in memory, 三维堆叠3D integration, and 光电融合optoelectronic integration.

These are not uniquely Chinese ideas. The entire industry is moving in these directions. What matters is that they have been elevated into a national technology strategy, and Chinese companies are moving quickly in all three.

Computing in memory and 3D integration are intended to address the memory wall. HBM is extremely important today, but technically it is an awkward and transitional solution. The industry needs a more direct way to place memory and computing together. I expect samples based on a new approach next year and volume production the year after that.

Samsung, SK hynix, and Micron are also pursuing these technologies. My point is not that they will stop innovating, but that one Chinese company is currently ahead in a particular implementation. The same kind of intellectual-property dynamic has already appeared around YMTC’s Xtacking NAND architecture. In the future, some global leaders may need to license or otherwise work with technology developed in China.

Baiguan: Does the same optimism apply to semiconductor materials, including photoresists and silicon wafers?

Dr. Joe: Japan remains the leader in many semiconductor materials, and this is still an area of Chinese dependence. But there are now a very large number of Chinese companies conducting materials research. I expect another fundamental change within several years.

My broader prediction is that around 2032, the overall picture will flip. That is only six years away. In some categories China will still be catching up, but in others the direction of the gap will reverse.

In the paid sections below, Dr. Joe will continue to answer questions on:

• How global talent flows have shaped and will shape the semiconductor industry

• Why Dr. Joe returned to China

• The triumphs and mistakes of SMIC, China’s largest domestic foundry

• What a self-sufficient Chinese semiconductor supply chain would mean for the rest of the world.

• Will China bring “over-capacity” for semiconductors as well?

Read more

The catalyst was 《后西游记》(”Journey to the West: After”)— a 30-episode, 40-minutes-per-episode mythological series that aired August 31, 2026, simultaneously on Mango TV and Hunan Satellite TV’s prime-time slot. There are no human actors, no film crew, no on-location shoots; every character, set, and frame is generated by AIGC. Per the production team, the cost came in at roughly one-tenth of a traditional drama of comparable length [*].

Hunan TV—commonly nicknamed “the Mango TV” because its logo resembles a mango—holds a very special place in the childhood memories of Chinese people born in the ‘80s and ‘90s. It has always been a pioneer, introducing reality television, variety entertainment, and talent competitions while most traditional Chinese networks stuck to old-fashioned programming. And now, it is placing a new bet: AI-generated dramas.

Subscribe now

Seedance, and a regulatory green light

The show was built on ByteDance’s Seedance 2.0 / 2.5 models, delivered through Mango TV’s in-house AIGC Innovation Content Center [*]. That alone is a notable vote of confidence in a domestic video model from a tier-one broadcaster.

But the more interesting unlock is regulatory. The show is the first project to run under the National Radio and Television Administration’s “边审边播” (review-while-broadcasting) mechanism, rolled out under the NRTA’s so-called the “21 Articles” [*].

Issued in August 2025, the 21 Articles are the clearest signal yet of Beijing’s shifting posture on audiovisual content:

• The 40-episode cap on TV dramas is gone.

• Overseas IP adaptation is explicitly encouraged.

• “Review-while-broadcasting” is piloted — but only for series, seasonal, anthology, and sitcom formats (the structurally modular kinds that don’t need a fixed ending locked before air).

In other words, the regulator is simultaneously loosening format restrictions and embedding AI-native production into the compliant mainstream. For a year that I earlier called 2026: Year One of China’s AI Film Industry, this is the moment the establishment stopped treating AI video as a fringe experiment.

While people debated “does it look real,” China started getting paid

AI animation is among the first categories to reliably generate cash flow. Prior to Mango TV airing the show, Chinese netizens were already watching AI shorts online. For instance, ‘AI anime drama 漫剧’—anime series created purely by AI—has become the addictive phenomenon known as ‘mealtime shows’ due to its ability to pack addictive hooks and memes into short episodes

While the discourse is still stuck at “but does the AI hand look right?” or “can AI writing really replace a human?”, in China, AI video has quietly become a cash-flow business. As I wrote back in my AI film industry piece, we’ve moved well past the “toy project” phase. The monetization isn’t theoretical anymore — it’s showing up in platform payouts, in recharge numbers, and in overseas app-store receipts.

The most important mental model: AI video models are engineered to spike dopamine, not to be philosophically profound. A vertical short-drama viewer is on a subway scroll looking for a 20-second emotional hit — the slap, the reveal, the underdog comeback. They don’t care if the glass-physics is slightly off. It’s the visual equivalent of junk food: cheap to make, irresistible to watch, but highly profitable.

Let me show you the receipts.

Where the money actually is: three working models

1) AI 漫剧 (AI “anime dramas”)

AI anime-dramas are the sleeper hit of 2026. Production is startlingly cheap: a 3D AI anime runs about ¥2,000–2,500 per minute versus six figures for traditional animation [*]. Platforms have built explicit monetization rails — Douyin, 红果 (Hongguo, micro drama platform owned by Bytedance), and Tencent’s 火龙漫剧 (Huolong) have launched dedicated anime revenue-share plans.

Per-episode payouts can reach thousands of yuan for shows crossing 10 million views [*]. Some verified accounts report personal AI-content incomes above ¥200,000/month without selling courses — purely from platform share and novel affiliates [*].

2) AI live-action short dramas — already topping the charts.

Since the start of 2026, multiple AI live-action short dramas have become breakout hits: 《斩仙台 AI 真人版》 (Zhan Xian Tai AI Live-Action) crossed 100 million views in six days; 《嫡女泣血》 (The Legitimate Daughter Weeps Blood) surpassed 230 million views to top industry charts [*].

Period-costume and Chinese fantasy-adventure are already very popular formats of micro-dramas in China and have become the “guilty pleasure” of many. Even before the emergence of AI, an entire “assembly line” already existed for producing these fast-paced, addictive series. They bypass A-list stars, instead licensing novel scripts from the internet and sign internet influencers or lesser-known actors. Now, AI has simply driven production costs—which were already squeezed—down to rock bottom.

3) The “B2B” model: “legitimate” film and TV integration.

Early this year, the acclaimed Chinese director Jia Zhangke released an AI film built on Seedance 2.0, starring an AI version of himself — a signal moment of a top-tier professional embracing the stack [*]. Similarly this year, Kuaishou’s Kling was integrated into the popular historical drama 《太平年》 (Swords Into Plowshares), where AI-assisted storm-and-naval-war sequences improved efficiency 8–10x while holding cinematic quality [*].

That’s what I call the “B2B” model: sell the model and workflow into professional pipelines, not just to consumers.

The micro drama export: almost every AI short drama abroad is Chinese

If you’ve been watching the overseas short-drama charts, you’ve probably noticed something: the AI video short plays flooding foreign app stores are almost entirely made by Chinese companies. This is the direct sequel to the Chinese-made short-drama boom abroad I covered when ReelShort first topped the US App Store — except now the marginal cost of a new title has fallen off a cliff.

According to Sensor Tower’s Q1 2026 statistics, out of the $750 million global in-app purchase (IAP) market for that quarter, DramaBox (Dianzhong Technology) and ReelShort (Crazy Maple Studio) each generated nearly $140 million in IAP revenue, tying as global dual oligopolies. Together, these two China-backed leaders captured nearly 37% of total direct consumer spending worldwide for the quarter.

It is, in a sense, the video equivalent of the “token export” wave I wrote about earlier: Chinese AI capability leaving the country as finished, monetizable product.

Investment implications

Read more

Yijing also writes at

Baiguan: China won enormous global market share in solar panels, batteries and EVs. If that was phase one, what is phase two?

Yijing: Phase one was hardware scale. Phase two is system intelligence.

China has already demonstrated its ability to manufacture clean-energy technologies at extraordinary scale and drive down their cost. The next challenge is to make renewable power, storage, electric mobility, industrial demand, and computing infrastructure operate as one increasingly complex system.

The scale of that achievement also explains why the bottleneck is changing. In 2025, China added 438 GW of grid-connected wind and solar capacity, taking cumulative wind and solar capacity to 1.84 TW, or about 47% of the country’s total generating capacity. Wind and solar generation accounted for about 22% of total electricity consumption, while renewables as a whole supplied 38.5%. The gap between the share of installed capacity and the share of electricity actually generated illustrates the next challenge: converting abundant but variable assets into power that is available when and where it is most valuable.

Location and timing are increasingly important also because China’s renewable resources, industrial demand and population centers are not evenly distributed. In 2025, cross-provincial and cross-regional electricity trading reached 1.59 trillion kWh, while continuous provincial spot-market operation achieved near-national coverage. These mechanisms allow prices to reflect periods of surplus and scarcity more accurately and give generators, storage operators and users stronger incentives to change their behavior.

For investors, the question is no longer simply: Who can manufacture at scale? It is: Who controls the critical points that make the whole system work?

Those critical points may include forecasting renewable output, routing electricity across regions, shifting demand between hours, aggregating distributed assets and responding to real-time prices. As these functions become more important, the industry’s profit pools can migrate away from the largest physical assets and toward the companies that coordinate them.

That moves value toward grid flexibility, power electronics, energy management, storage monetization, and the integration of energy with AI and advanced manufacturing.

Baiguan: In which areas can we find next clean-tech winners?

Yijing: When hardware becomes abundant, value often moves toward the control points. The next major winner may be less recognizable than a solar-panel or EV manufacturer, but more deeply embedded in how the entire system functions.

Cooling determines how efficiently AI infrastructure can operate. Power electronics control how electricity is converted and managed. Energy trading and virtual power plant platforms connect distributed assets to electricity markets. Thermal management systems influence the performance of EVs, batteries, and data centers.

These companies may be less visible to consumers, but they can control performance, reliability, and operating economics.

China is beginning to formalize markets for these coordination functions. National policy now defines virtual power plants as market participants that can aggregate distributed generation, storage and flexible loads, with nationwide regulation-capability targets of more than 20 GW by 2027 and more than 50 GW by 2030. Subject to local rules, they can participate in medium- and long-term electricity markets, spot markets and ancillary-service markets.

This is particularly relevant to the tour’s NIO Power visit: charging and battery-swap infrastructure can begin as mobility infrastructure and gradually acquire a second role as a distributed power-system asset.

Moreover, frontier technologies such as perovskite solar cells and controlled nuclear fusion may also give rise to new winners. We will also cover them in the upcoming tour as well.

Baiguan: Does AI infrastructure create a burden for the energy transition, or a new investment opportunity?

Yijing: AI is not simply creating another source of electricity demand. It is forcing energy and computing infrastructure to be designed together.

Data centers require reliable power, grid connections, cooling, storage and backup capacity. These constraints are creating opportunities across the infrastructure stack.

The expected increase is substantial. In the International Energy Agency’s base case, electricity consumption by Chinese data centers rises by around 175 TWh between 2024 and 2030, an increase of about 170%. Data centers are also geographically concentrated loads, so their pressure on a particular grid connection, substation, or local power market can be much greater than their share of national electricity use would suggest.

The most interesting companies will not merely sell more electricity or more computing equipment. They will solve the points where the two systems meet: power availability, thermal efficiency, intelligent load management and the integration of generation and storage.

AI could become one of the greatest new demands on the power system—and one of the strongest catalysts for modernizing it. We will explore these questions during our visits to AliCloud and Sanhua.

Baiguan: If many Chinese companies can scale, where can an industrial company still build a durable moat?

Yijing: Scale is a capability. It is not necessarily a moat.

In China, one of the most difficult—and most critical—advantages is the ability to survive and compound across cycles. Many companies can grow during an industry upswing. Far fewer can continue investing, retaining customers, and protecting their economics amid price competition, policy changes, technology shifts, and demand cycles.

That resilience may come from proprietary technology, manufacturing quality, customer qualification, capital discipline, system integration capability, or the ability to identify the next growth curve before the existing one matures.

In many industrial markets, the moat is created by a combination rather than a single protected technology. A mission-critical component may be relatively inexpensive compared with the final product, but replacing it can require renewed testing, system redesign, reliability validation and changes to the customer’s production process. Companies that combine product performance with manufacturing yield, application engineering, customer certification, service networks and the ability to support clients globally can therefore retain pricing power even in markets characterized by intense headline competition.

Another increasingly important advantage is the ability to move from selling an individual product to solving a system-level problem. A company that understands how its component affects energy consumption, operating reliability, and the economics of the customer’s full system may be able to capture more value than a competitor that sells largely on unit price.

The real test is not simply whether a company can become large. It is whether it can remain strategically relevant, financially healthy, and commercially competitive across multiple cycles. In the Chinese market, that long-term adaptability is exceptionally difficult to build—and absolutely critical to survival. One case we will examine during the tour is GCL, a first-generation Chinese solar company that has remained strategically relevant through multiple industry cycles.

Baiguan: What is most often misunderstood about China’s industrial policy?

Yijing: China is often viewed from outside as a predominantly top-down system. Policy direction matters, but it explains only part of how industries actually develop.

Central policy can establish strategic priorities. State capital can support infrastructure and absorb early risk. Local governments compete to attract companies, talent and supply chains. Private companies then compete intensely on technology, cost, customers, financing and execution.

Research mapping more than 768,000 Chinese industrial-policy documents issued between 2000 and 2022 illustrates this division of labor. Only 13% were issued by the central government, compared with 45% at the provincial level and 39% at the city level. Central authorities were more likely to set strategic priorities and use regulatory tools, while local governments relied more heavily on infrastructure, subsidies, labor support, and industrial clustering to attract companies.

If you truly understand this market, you also see its highly market-driven—and in many sectors intensely capitalist—side. Companies must commercialize quickly, win customers, raise capital, recruit talent and defend market share. Policy may help create the playing field, but it does not determine which company ultimately wins.

China’s industrial system is thus best understood as a combination of strategic direction, local experimentation, private entrepreneurship and intense market competition. That interaction—not policy alone—is what international investors need to understand.

One of the best places to visit in order to have a good grasp of these central-local-business dynamics is Suzhou Industrial Park, one of China’s very first industrial parks, which we will also visit thoroughly this time.

Baiguan: What can investors understand during five days in Shanghai, Suzhou and Hangzhou that they cannot learn from reports or conferences?

Yijing: Reports can tell you what China has built. Being on the ground shows you what it looks like at scale.

Participants will enter real operating environments—not simply conference rooms—and see technologies, production facilities, energy assets and infrastructure in context. The physical scale, speed of deployment and integration across supply chains are difficult to understand from data alone.

More importantly, site visits make it possible to test the operating assumptions behind the data. Is a storage asset actually being dispatched, and how is it paid? Is a manufacturing advantage based on automation, yield, supplier density or simply lower labour and capital costs? Is a data center constrained by chips, grid access, power quality or cooling? Is a startup’s demonstration connected to a repeatable customer need? These questions are difficult to answer from a corporate presentation alone.

The second difference is access. Through 2060 Advisory’s long-standing relationships, participants will engage directly with senior leaders and decision-makers behind the selected institutions—the people responsible for strategy, technology, investment and operations. These conversations are difficult to access independently, and even harder to assemble across multiple leading institutions in five days.

Together, the site access and senior-level dialogue allow investors to understand not only what companies have achieved, but how their leaders think, what they are building next, where the real constraints lie and where future value may emerge.

If you want to know more details about the upcoming energy tour in collaboration with Yijing, RSVP and get full details here:·

To read a previous Q&A with Mr. Gao Erji of Caixin, our partner in the AI & Robotics leg, read here:

You can also check out the Q&A of regularly asked questions below:

Please feel free to contact us at Baiguan_ChinaTour@bigonelab.com with any questions.

Last week, the 2026 World Robot Conference (WRC) concluded in Beijing. ICYMI, Amber published her own field notes about the event yesterday.

Today, we are delighted to invite someone who was deeply involved with the WRC as well as the broader China’s tech industry, Mr. GAO Erji, executive president of Caixin Media and managing director of Caixin Data, to pick his brain on his takeaways about WRC, his own views about AI and robotics in general, and what to expect in the upcoming AI & Robotics tour in Beijing that Caixin Data and Baiguan collaborate on in October.

Mr. Gao (Linkedin) also writes a personal newsletter at Gao’s Substack.

Q: Mr. Gao, please introduce a bit about yourself and Caixin Data.

A: I am Gao Erji. Before joining Caixin, I spent a decade as an investment banker and led the execution of more than 50 cross-border M&A transactions. For the past twelve years, I have been with Caixin, where my original mission was to upgrade our financial databases with AI. Through five strategic acquisitions, we built Caixin Data into what it is today. In 2017, Caixin founded its AI Lab to drive the technological infrastructure underpinning our intelligence services. I also serve as a strategic advisor to several institutions, and I have written some two hundred novels.

Caixin has been ranked ninth worldwide in the newly released 2026 Global Digital Subscription Report, with 1.2 million paid subscribers, confirming its position as the largest subscription-based news organization outside the U.S. and U.K. Our journalism is defined by depth, rigor, and professionalism. Beyond news, we host more than twenty high-end global forums each year, attended by heads of state, central bankers, and global business leaders. Caixin Data sits at the intersection of this editorial heritage and artificial intelligence, transforming decades of proprietary content and structured financial data into AI-ready assets.

That combination has become increasingly important with the rise of generative AI. General-purpose models can absorb enormous quantities of public information, but professional financial applications require a different layer of infrastructure: structured company and industry data, traceable sources, consistent taxonomies, and information whose provenance can be verified. In this sense, the development of financial AI is not simply a race to build larger models. It is also a race to connect those models to reliable professional knowledge.

Q: Would you want to share some takeaways about the WRC?

A: August 2026 has cemented Beijing as the global hub for robotics. Hot on the heels of the annual World Robot Conference, the 2nd World Humanoid Robot Games unfolded in quick succession. I attended both events in full. I have several takeaways to share.

First, both the conference and the Games have set new records in overall scale.

The expansion is visible in the numbers. The second World Humanoid Robot Games brought together 666 teams and 2,056 robots from 16 countries, compared with 280 teams and more than 500 robots at the inaugural event in 2025. Chinese participation alone included 641 teams and 1,975 robots representing 157 companies and 200 universities and research institutions.

Second, we are witnessing tangible leaps in robotic performance. Moves such as spinning back kicks have become standard capabilities for humanoid robots. Robots have comfortably shattered human world records in sprinting, high‑jump, long‑jump, and long‑distance running, while demonstrating competence in tennis, badminton, and other sports.

But athletic records are arguably less important than the increasing emphasis on practical tasks. This year’s competition expanded scenario-based events covering factories, hotels, homes and logistics, including assembly, housekeeping and emergency-response tasks. More than 40% of the tests require full autonomy. In other words, the benchmark is gradually shifting from whether a robot can perform an impressive movement to whether it can perceive, decide and complete useful tasks reliably without human intervention.

Third, World Action Models are gaining broader adoption across industries. This signals rapid, mutual iteration across algorithms, data architectures, generalization capabilities, robot hardware‑software stacks, and supply chains.

The significance is that robotics development is increasingly becoming a full-stack optimization problem. Improvements in models can generate better training data and control policies; better actuators and sensors make more complex behaviors possible; deployment then generates new real-world data that can be fed back into the models. The speed of progress therefore depends not on any single breakthrough but on how quickly these layers can iterate together.

Fourth, traditional industries are rapidly integrating into robotics value chains and flagship robotics events. Incumbent enterprises are rolling out robotic‑related use‑cases for exhibition. For instance, telecom operators have launched their own robotics offerings; automotive OEMs and component suppliers are releasing either complete humanoid robot units or robotic hardware components. For our part, Caixin brought our AI-powered financial tool suite to demonstrate at the event.

Lastly, participation by 16 nations in the Robot Games underscores China’s use of AI and robotics as instruments of international outreach. Indeed, AI and robotics are emerging as pivotal pillars for China’s global technological cooperation.

Q: Let’s wind back a bit. DeepSeek last year changed the international perception of Chinese AI almost overnight. But was DeepSeek really an exceptional breakthrough, or was it simply the moment outsiders finally noticed an ecosystem that had already become much stronger?

A: DeepSeek was not an overnight miracle but the inevitable result of years of foundational investment. Liang Wenfeng and his team at High-Flyer Quant had spent years building underlying algorithms and talent density long before the world took notice. The international sense of surprise simply reflects how little outsiders had been tracking China’s AI ecosystem. DeepSeek was the moment the world finally noticed a system that had already grown formidable.

This is not hindsight. Even if you just read us, for example, Caixin was tracking DeepSeek well before the global spotlight arrived. On August 31, 2022, Caixin published a report on DeepSeek titled DeepSeek’s Rise Sparks Arms Race Among Major AI Model Developers. In May 2024, we reported on the release of DeepSeek-V2 and its industry-shaking pricing strategy—dubbed the “Pinduoduo of AI”—which forced major tech firms into a price war. In late December 2024, weeks before the R1 sensation, we covered the launch of DeepSeek-V3, noting its training cost of merely $5.58 million and its performance on par with GPT-4o.

Q: China appears to be producing an extraordinary number of AI models, robotics companies and embodied-AI startups. How much of this represents genuine technological differentiation, and how much is simply capital and policy creating too many similar companies?

A: In the early stages of any technological revolution, dense capital and policy support are prerequisites for breakthroughs. After the initial “hundred-model war,” the foundation-layer LLM landscape has consolidated significantly. Survivors are now accelerating into vertical scenarios, while robotics firms are differentiating along world models, joint modules, and force-control algorithms.

For LLMs, the important change is not simply that fewer companies remain at the foundation-model layer, but that competition is moving to different layers of the stack. Some companies are competing on reasoning and multimodality; others on inference cost, long-context processing or open-source ecosystems.

As for robotics, China’s dense manufacturing ecosystem reinforces this process. Robotics companies can draw on existing supply chains in motors, reducers, sensors, batteries, automotive electronics and precision manufacturing rather than building every component ecosystem from scratch. That lowers the cost and time required to experiment with different hardware configurations. But it does not eliminate the fundamental problem: many companies can build a robot; far fewer have yet demonstrated a repeatable, economically valuable application.

Q: In the upcoming tour that we collaborate on together, we are going to hold meetings with C-level and even founding-level people of Moonshot (Kimi) and Zhipu (GLM). What do you expect visitors can seek answers about?

A: With Kimi, visitors should press on two fronts: first, how Moonshot converts its long-context advantage into a sustainable business model in a Chinese consumer market that remains cautious about paying for software; second, how that capability will integrate with serious professional use cases beyond chat.

With Zhipu, the critical questions are how its aggressive open-source strategy translates into commercial returns, and how it balances technological ambition with the immediate demands of enterprise and financial clients in on-premise settings.

These are deliberately different questions because the two companies represent different routes toward commercialization. The interesting comparison for visitors is therefore not simply which company has the stronger model, but how different technical strengths, product strategies and customer bases translate into durable businesses.

Q: China seems particularly strong at taking technologies out of the laboratory and turning them into cheap, manufacturable products. Does that give China an unusual advantage in embodied AI compared with purely software-based AI?

A: Manufacturing scale is an important part of China’s DNA, but we believe the current wave of AI and embodied intelligence is driven less by supply-chain cost advantages than by sustained R&D investment and the research density created by a massive engineering talent pool.

A mature supply chain can reduce the cost of actuators, sensors, batteries, controllers and mechanical components, and it allows prototypes to be redesigned and manufactured quickly. China’s existing strength in automotive electronics, industrial automation and precision manufacturing clearly provides a favorable environment for robotics; the country’s dense and vertically integrated supply chains are widely regarded as one reason Chinese humanoid companies can iterate rapidly and reduce hardware costs.

But cheap hardware alone does not produce embodied intelligence. The harder frontier increasingly lies in perception, planning, manipulation, generalization and the accumulation of high-quality interaction data. That requires research organizations, algorithms and engineering talent capable of making the hardware progressively more intelligent. China’s real advantage, therefore, may lie in the combination: a large research and engineering workforce operating inside a manufacturing system where ideas can move unusually quickly from model to prototype to physical deployment.

The upcoming tour will allow you to take a close look at how exactly this unique combination creates synergy.

Q: Robotics seems to develop really rapidly in China. Are you afraid this could be a huge bubble? What would the upcoming tour help answer this question?

A: Some valuation inflation is normal in the early phase of any technology wave. Caixin has recently launched an Embodied Intelligence Long-Term Value Index precisely to warn against applying traditional valuation frameworks to this sector.

The difficulty is that technological progress and investment excess can coexist. The existence of speculative valuations does not mean the underlying technological transition is unreal; conversely, rapid improvements in robot performance do not guarantee that every company participating in the boom will generate attractive returns.

That is why commercialization matters. Visitors should distinguish between demonstrations and repeatable deployments: Who is actually paying for the robot? What task does it replace or augment? How reliable must it become before the economics work? How much human supervision is still required? And does each additional deployment generate data that improves the product?

This distinction is becoming particularly visible at the Robot Games themselves. Athletic performances attract attention, but scenario-based tests such as factory assembly, restaurant service and EV charging expose much harder problems involving perception, precise manipulation and recovery from errors.

The upcoming tour will allow visitors to verify underlying technical strengths firsthand, and assess for yourself the level of froth in current valuation levels.

Q: China’s technology policy frequently creates intense competition rather than protecting national champions. Is this deliberate? And does the resulting overcompetition ultimately accelerate innovation or destroy returns?

A: This reflects a deliberate preference for market competition over industrial policy protectionism.

In practice, this creates an unusual combination of policy direction and decentralized competition. The central government may identify a strategic sector, but numerous cities, funds, universities and private companies then pursue competing technical routes and commercial models. Local governments and enterprises frequently collaborate through “tournament” mechanisms such as technology challenges and competitive grants. (One part of our upcoming tour will take us to Yizhuang, a key high-tech zone in Beijing, and see how that central-local relationship works in practice.)

The result can look inefficient from the outside: duplicated factories, overlapping research programs and periods of severe price competition. Yet similar competitive dynamics have appeared in industries ranging from solar and batteries to electric vehicles. Competition forces companies to reduce costs, shorten development cycles and continuously improve products.

For venture investors, high uncertainty is the price of entry. In Chinese tech, one must accept that the winner may take all, yet identifying that winner early is exceptionally difficult.

On the other hand, this intense competition may produce interesting “picks and shovels” plays. Just as in the EV industry, the better investment may not be EVs but batteries, overcompetition in robotics in general does not rule out the possibilities of bottlenecks in the supply chains. And you can only find those bottlenecks by seeing it with your own eyes.

Q: If you had to choose one part of China’s AI or robotics ecosystem that international investors are currently overestimating, and one that they are underestimating, what would they be?

A: Overestimated: Vertical applications of foundation models. Many investors assume LLMs can rapidly penetrate vertical industries, yet acquiring high-quality proprietary data, mastering complex domain knowledge, and closing commercial loops are far harder than the narratives suggest. A number of vertical-application stories are overblown.

Underestimated: The upside from combining traditional industry assets with AI. Take Caixin as an example. Decades of accumulated content assets, financial databases, and analytical expertise, when augmented by AI, can unlock value far beyond market expectations. The “traditional professional asset + AI” model—in content verification, financial analysis, and industrial research—offers substantial headroom that investors have largely ignored.

The key distinction is between possessing information and possessing information that has already been accumulated, structured, and repeatedly tested inside a professional workflow. A financial database assembled over decades contains not only raw records but classifications, entity relationships, historical series and institutional knowledge about how professionals actually use the information. Similar advantages exist in law, medicine, engineering and other knowledge-intensive industries.

AI can lower the marginal cost of interrogating those assets dramatically. Tasks that once required an analyst to search multiple databases, read dozens of documents and manually reconcile information can increasingly be performed through natural-language interfaces and agentic workflows. That does not make the underlying professional asset less important. It can make it more valuable, because AI dramatically expands the number of ways in which that asset can be queried and recombined.

Q: If we return to Beijing five years from now, which of today’s seemingly futuristic industries do you think will have become completely ordinary, and which ones will probably still be mostly demonstrations?

A: AI-powered automation will become as ordinary as water and electricity—ubiquitous infrastructure across business and government. Special-purpose robots for specific industrial and service scenarios will also be deployed at scale. General-purpose humanoid robots in households, however, will likely remain constrained by safety, ethics, and technical feasibility, still largely confined to showrooms. The same applies to high-level autonomous driving in complex urban environments.

The important distinction is between generality and economic usefulness. A robot does not need human-level general intelligence to create significant economic value. Machines optimized for warehouses, factories, inspection, hospitality or other bounded environments can become commercially important much earlier precisely because the range of situations they must handle can be constrained.

Another key area to watch is commercial space, which our upcoming tour will also touch on: Low-earth-orbit satellite constellations and their downstream applications—satellite broadband, remote-sensing analytics, and IoT connectivity—will likely have shifted from experimental to operational. However, commercial human spaceflight and space tourism will probably remain in the demonstration phase, constrained by cost, regulation, and crewed-safety requirements.

Q: China has severe constraints on access to cutting-edge AI chips. Yet Chinese AI development has continued remarkably quickly. Has the semiconductor constraint actually changed the trajectory of Chinese AI, and where are its effects most visible?

A: The constraints have indeed altered the trajectory. On one hand, they have accelerated domestic AI chip R&D; the gap with global leaders persists, but iteration continues. On the other, investment has structurally shifted upstream—toward chip design, advanced packaging, and compute leasing—to build supply-chain redundancy and resilience.

It’s not just GPUs. Constraints alter engineering decisions throughout the AI stack. Developers have stronger incentives to improve model efficiency, optimize inference, use mixture-of-experts architectures, make better use of available accelerators and reduce the amount of compute required for a given level of performance. DeepSeek became an internationally visible example of this broader pressure toward efficiency.

At the same time, substitution is incomplete. The performance gap at the frontier remains important, particularly for training the largest models and for workloads dependent on mature software ecosystems. The result is therefore neither “export controls have stopped Chinese AI” nor “export controls have failed completely.” They have raised costs and changed investment priorities while simultaneously strengthening the incentive to develop domestic alternatives.

Q: AI is increasingly becoming an information problem rather than merely a model problem: who owns reliable data, who can structure it, and who can let agents use it safely. Does China have an underappreciated advantage here because of the enormous digitization of its economy, or does fragmented and restricted data remain a weakness?

A: China possesses abundant data resources, yet vertical-domain data remains fragmented across institutions. Systematic integration is the next critical hurdle. For example, Caixin Data has invested heavily in proprietary economic and financial databases, and we currently provide RAG (Retrieval-Augmented Generation) services to major LLM companies. At the same time, we observe that some model builders rely on crude, low-cost data acquisition, which is degrading output quality. Professional-grade verification and quality-assessment services for model outputs represent a significant frontier—one that Caixin is actively exploring.

The distinction between quantity and quality will become increasingly important as AI moves into professional applications. The open internet provides enormous amounts of training material, but finance, law, medicine and industrial research depend on information that is accurate, current, structured and attributable to a reliable source. In those environments, a plausible but unverifiable answer is often not useful at all.

RAG is one way of addressing this problem by allowing a model to retrieve relevant information from controlled knowledge bases at the time a question is asked rather than relying exclusively on what was embedded during model training. But retrieval alone does not solve the problem if the underlying database is incomplete, poorly structured or unreliable.

This creates a new value chain around professional AI: acquiring and structuring proprietary data; maintaining entity relationships and historical records; retrieving the appropriate evidence; and finally verifying whether the model’s output is supported by that evidence. For institutions that have spent decades building professional information assets, AI represents not merely a threat to the traditional information business but also a potential new distribution layer for those assets.

If you want to know more details about the upcoming Robotics & AI Executive Tour in collaboration with Mr. Gao’s team at Caixin, RSVP and get full details here:

To ensure the best experience, space is limited to 20 seats. Early-bird pricing ends September 13.

You can also check out the Q&A of regularly asked questions below:

Please feel free to contact us at Baiguan_ChinaTour@bigonelab.com with any questions.

Subscribe now

_{Data: Tushare / Wind | Data Through: Jul 24, 2026 close}

China’s two technology boards -- the ChiNext and the STAR Market -- have ranked among the top-performing equity indices globally since the start of 2025. We have tracked this run in multiple prior notes. As in every other major equity market, AI has been the core engine of the upside.

Since late June, however, the tech indices have suffered a sharp correction. The defensive CSI Dividend index has begun to rebound. This move is broadly synchronous with the pullbacks in the US SOX and Korea’s KOSPI.

STAR nearly doubled while ChiNext surged -- then the peak arrived and the market started to reverse

Two forces, one China-specific factor

The global tech correction shares plenty of common causes. Both trading technicals and industry fundamentals are working in tandem.

On the trading-technical side: the prior run-up was simply too far, too fast. We have built a two-dimensional framework to track China’s AI bubble across various sub-themes, and found many of them have surpassed the peak expected growth rates and valuations of the last renewable energy bubble.

On the fundamental side, which is more central: the entire semiconductor chain -- i.e., upstream AI hardware -- has captured a disproportionate share of AI profits, cannibalizing downstream earnings. The long-term ROI remains uncertain, and the market is growing more anxious about this.

But the pressure facing the Chinese market has one factor that is its own: the imminent listing of ChangXin Memory Technologies (CXMT).

Monday: CXMT begins trading

July 27, Monday -- CXMT formally lists on the A-share STAR Market.

A widely circulated belief in the Chinese market holds that the IPO of a super-mega-cap inevitably marks the end of a bull run -- corresponding to an interim high in the index. The chart below, making the rounds across social media over the weekend, captures this sentiment.

Every major mega-IPO has clustered near an interim market top

I tend to look at it the other way around. Companies themselves prefer to IPO and issue stock at the peak of market sentiment, at a good price. That is what makes the pattern appear so consistently. The mega-IPO doesn’t cause the top -- the top, more precisely the extreme sentiment, enables the mega-IPO. But historical regularity is enough on its own to make investors de-risk ahead of time, and that may well be one reason for the magnitude of the recent sell-off -- even though China’s own leverage problem is nowhere near as severe as Korea’s.

Most investors are choosing to watch and wait. You can see it in the volume. The market is, literally, holding its breath for CXMT.

Volume has collapsed -- the lowest turnover of the entire drawdown

Why CXMT matters beyond China

The listing of this memory giant on the A-share market means China finally has a proper domestic DRAM player -- one with global scale and, potentially, global spillover effects.

Once CXMT has the capital markets’ backing, capacity expansion is a given. Historically, China’s pattern of pouring massive capital into manufacturing capacity expansion has earned the reputation of a “developed-country meat grinder” -- a process that can restructure profit pools across the entire supply chain.

That is not good news for competitors. Think autos and new energy vehicles. But it is potentially good news for consumers, and for companies unwilling to tolerate upstream price gouging -- a consumer-electronics giant like Apple, for instance, which is already reported to be sourcing memory from new Chinese suppliers.

ChiNext vs STAR: the split tells you where the pain is

Drilling into the A-share market, the first thing that stands out: ChiNext has fallen more than STAR.

The ranking inverts: dividends are the only positive line since the peak, and ChiNext suffered a bigger loss than STAR

Both are China’s tech boards, but the composition differs in a way that matters here. In short: the core optical module and memory companies sit on ChiNext before CXMT’s listing, while semiconductor -- including equipment, materials, and manufacturing -- is concentrated on STAR. The STAR Market was established later than ChiNext, and its listing rules tolerate pre-profit companies, specifically to support capital-intensive early-stage semiconductor businesses. As a result, semiconductor density on STAR is far higher.

Let’s first look at the ChiNext market.

Chinese investors habitually sort the optical-module names into the “overseas chain” -- they follow US AI capex, and many supply directly to NVIDIA or American hyperscalers. Most memory names, by contrast, were until CXMT’s arrival peripheral players -- companies that profited from the appreciation of memory inventory without being the primary beneficiary. Their valuations have been cut recently as the real big guy is coming.

For example, Techwinsemi (001309.SZ) -- a storage controller chip designer -- has hit consecutive limit-downs and is now treated as a sentiment bellwether for the entire speculative memory trade.

Storage has been cut nearly 40% -- far worse than optical modules

Six consecutive limit-downs: Techwinsemi as the market’s sentiment bellwether

STAR’s resilience: self-reliance premium, and the picks-and-shovels bid

STAR’s drawdown is light compared to ChiNext, and close to CSI 300 -- the latter stuffed with high-dividend names that have been relatively defensive in this sell-off. STAR also carries something CSI 300 does not: a semi self-reliance premium, the narrative of China’s domestic semiconductor ascent.

Here is the interesting part. If you crack open the semiconductor sector itself, you find that semiconductor equipment and manufacturing are actually up during this drawdown.

Equipment and manufacturing held firm

What is the market thinking? Part of the answer, in my view, is that CXMT’s listing, expansion, and the broader memory capacity build-out all depend on massive upstream semiconductor investment. These equipment and materials companies are the direct beneficiaries.

This logic lines up with a recent Bernstein note, dated 13 July 2026: Global Semis -- Can semi cap work if memory doesn’t? Their answer is yes. Bernstein shows that the historical correlation between memory and semiconductor equipment (WFE) share prices is only 0.4-0.6. There have been multiple two-year stretches where WFE outperformed memory by 38-49 percentage points while memory went sideways or down. Their argument: memory capex is accelerating globally, and that capex is the WFE order book. Equipment can work even when memory doesn’t.

Semiconductor manufacturing and equipment in Chinese market are especially resilient for the same reason: The companies in those two sub-indices are, in many cases, directly linked to CXMT -- they are its suppliers. The market may well be pre-trading the benefit of CXMT’s capex ramp before the listing even happens.

The pricing map: huge divergence, freezing sentiment

As for what happens after CXMT lists, the market itself is deeply divided. One thing that is different this time: in the month or so before the offering, the market has already absorbed a violent drawdown. Sentiment has reached a relative freezing point -- the ¥1.94tn turnover print on July 24 is the lowest of the entire correction.

The most optimistic prediction earlier was that CXMT would open straight to ¥5 trillion in market cap. That number is now scoffed at by most investors.

Below is a widely circulated scenario analysis of CXMT’s Day-1 market cap and its market impact.

Accordingly, the best outcome for the market is a mild open -- a Day-1 print near or below the ¥3tn fair-value anchor, letting the re-rating of related names unfold over a month rather than cashing it in on day one. On the prediction markets, fewer than 35% of participants now believe CXMT will open above ¥3tn. If that is the case, it would be a healthy de-expectation. We will see what happens tomorrow.

Subscribe now

Share

Related read

About Baiguan Pro

We publish more deep dives like this in for our Baiguan Pro subscribers. In the article, you’ve seen a few examples, including our recent take on the framework proposed to evaluate China’s AI bubble:

and our latest check up after the market correction:

The aforementioned rebound of the dividend stocks is actually a much long-term themes we think that worth investors' attention, which we discussed in a previous Baiguan Pro deep dive:

We also share our team views toward Chinese assets every month, and our proprietary data at BigOne Lab to help you stay updated on the performance of key names and sectors.

For investors who wish to receive more targeted content on Chinese secondary market investments, we recommend that you get a paid subscription with us, as we are offering free upgrade to Baiguan Pro for all existing paid subscription tier. The offer ends on Jul 31.

If you are already a paid subscriber with us, feel free to reply and let us know if you would like to be upgraded to the Baiguan Pro tier.

Subscribe now

China is entering the 15th Five-Year Plan (2026–2030) with a rare explicit number on consumption. In July, the State Council approved the Expansion of Consumption 15th Five-Year Plan, setting a quantitative target: social retail sales of consumer goods around RMB 60 trillion by 2030. For a country that has spent post-COVID years trying to reflate domestic demand, that kind of headline number matters. It reads like a clear, unambiguous priority.

So many investors are asking the same thing: Is this the window where Beijing finally puts serious stimulus behind domestic consumption? A numeric target in a five-year plan can feel like the setup for a beta trade — especially when consumer equities already look depressed.

1/4 China’s consumption names are cheap but still unloved

The consumer staples and discretionary sectors remain deeply discounted at a 25–27% five-year percentile after a mild rebound, standing in sharp contrast to the CSI 300’s elevated 92.7% percentile. This stark valuation divergence clearly reflects how capital markets are pricing a K-shaped economic reality—buoyed by strong momentum in the “new economy”, specifically AI and data centers, while domestic consumption remains sluggish.

In a past Baiguan Pro issue, we highlighted how computing infrastructure and new energy leaders have been replacing traditional enterprises within the CSI 300. Consequently, the index itself has transformed into a much more accurate proxy for China’s emerging 'new economy.'

2/4 Does the RMB 60 trillion target mean massive stimulus is coming?

From 2025’s RMB 50.1 trillion (+3.7% YoY), reaching ~RMB 60 trillion by 2030 requires only about 3.7% CAGR. That is almost exactly the 2021–2025 post-COVID trend (~3.3%). The current H1 2026 run-rate of +1.3% would miss the target badly.

The target in the plan just asks China to keep its post-COVID trend: some improvement from today’s soft goods print is implied — but nothing like a boom. Cheap ≠ buyable beta.

3/4 Households are still spending on services

Although consumption of goods and staples is likely to remain sluggish (not a broad beta rebound), China is not in a total spending freeze. Service retail has maintained a mid-5% YoY growth band for a year. (As of 2025, per capita services consumption made up 46.1% of China’s total consumption.)

4/4 “Services” is not one trade: where is organic growth inside services — and how to express it?

Organic growth in the services sector is heavily concentrated in experiences, led by tourism consulting and leasing (+11.3%), as well as culture, sports, and leisure (+10.4%). Conversely, catering (+2.8%) is actually dragging the overall average down.

By “experiences,” I refer to high-engagement, in-person activities. This momentum is vividly captured by the surge in live performances (+28.3%) and museum tours (+24.6%) [*]. Inbound tourism has also rebounded strongly, driven by an expansion of visa-free arrivals that catalyzed a 30.6% year-over-year growth in visitor trips.

Other categories, such as education, are notably absent from the highlights. Given China’s demographic headwind of a declining population, I personally think education is unlikely to drive significant organic growth going forward.

This means that secondary-market expressions need to be selective. Frankly speaking, exact equity proxies for China’s “experience economy” are still quite elusive. To bridge this gap, we are taking a first cut at constructing a basket that best captures this interest-driven, experiential growth. Our basket spans OTAs, live-experience ticketing platforms, hotels, and digital discretionary categories like gaming and entertainment. More on this in our Baiguan Pro tier.

Ending notes: possible 2H catalysts for consumer names

Apart from longer-term themes, HK internet and e-commerce giants could see several positive catalysts in H2 this year as delivery wars quiet down and profit margins improve. Specifically, the commercial landing of AI and agentic applications could provide strong catalysts for upside given current depressed valuations.

For instance, the expansion of cloud services, Kuaishou’s Kling, Tencent’s WorkBuddy (which the company claims is China’s most used productivity AI agent by DAU), and the launch of agents within the WeChat ecosystem could all serve as positive catalysts for H2.

Disclaimer: For informational purposes only. Not investment advice.

About Baiguan Pro

On April 1, we officially launched Baiguan Pro, our professional-tier subscription designed for investors seeking deeper, more specific coverage of Chinese equities. Baiguan Pro delivers the Baiguan team’s monthly outlook on overall Chinese equity assets, our quarterly focus on themes and sectors, richer company-level analysis, and follow-ups on the ideas and companies previously presented in the newsletter.

As a reminder, existing paid subscribers can claim a complimentary upgrade to Baiguan Pro (offer runs through July 31, 2026) by simply replying to this email. You could also consider signing up for a paid subscription and emailing us for a free upgrade! )

Related Reads

Amid the backdrop of Chinese companies going abroad, a new phrase has been coined in China to describe this phenomenon: “token 出海” (tokens going abroad). In today’s newsletter, I want to share some of my recent reflections: how real this trend is, why it’s happening now, who’s selling, and the investment implications behind it.

Subscribe now

The “token export” trend is growing faster than anyone expected

The reason to trust the OpenRouter numbers is that OpenRouter doesn’t run any models itself. It’s a switchboard — it takes your request and sends it to whichever provider fits your price and latency, so its data captures genuine demand rather than marketing. And that data is stark. Chinese models surpassed US models on the platform for the first time in the week of February 9, 2026, and accounted for 61% of consumption among the top ten models, holding the top three spots.

Although the 61% figure represents the top ten models rather than all 400-plus models hosted on the platform, the overall direction is undeniable.

The market itself has also exploded. OpenRouter has scaled nearly 5× in the past six months, from roughly 5 trillion tokens per week to over 25 trillion in May 2026. This means China didn’t just gain market share; it took a bigger slice of a much bigger pie. Furthermore, a16z’s Martin Casado also estimated that roughly 80% of startups utilizing open-source AI stacks are running Chinese models.

Inside China, the curve is even steeper. Average daily token calls went from 100 billion in early 2024 to 140 trillion by March 2026 — more than a thousandfold in two years, per the National Data Administration (NDA). And Beijing has started treating tokens less like a technical artifact and more like a commodity you can meter, price, and export.

In March 2026, the NDA formally standardized the Chinese word for “token” (词元, cí yuán) and its director described tokens as measurable, priceable, and tradable. And in April, the port city of Shantou completed what officials called China’s first city-level “closed loop” for token exports: servers in China generate the tokens, an API call comes in from Southeast Asia, the answer goes back out, all under a compliant cross-border data arrangement. Nothing physical crosses the border — just the inference result.

Regulatory compliance still needs to be fleshed out, but the trajectory is clear—China views tokens as its next goldmine amid the global race for compute dominance.

Why it’s happening now: the frontier gap closed, and tokens got expensive

Two things happened in the last year that made this trend possible.

First, the performance gap between frontier models and other open-weight models is narrowing.

Stanford’s 2026 AI Index — the most-cited scorecard in the field — put the top US model just 2.7% (39 points) ahead of the top Chinese model as of March, down from a 17.5~31.6 point difference in mid-2023. Stanford’s own word for it was “nearly closed.“

In academia, the returns on all model training spending are also visibly shrinking: annual gains on the MMLU benchmark fell from 16 points in 2021 to under 4 in 2025 even as R&D budgets ballooned, and roughly half of the major benchmarks are now saturated — everyone clusters near the top, so the scores stop meaning much.

For the ordinary user tackling daily tasks, the difference between the number-one and number-five models is becoming nearly impossible to distinguish. After all, not every job is rocket science requiring a frontier model; for the majority of personal and business scenarios, a much cheaper model suffices.

My empirical observations are similar. In my own coding and software development projects, I don’t find a large enough difference between, say, GLM 5.2 and Claude Fable to justify the price premium. I’ve also tested these models on non-coding tasks—such as suggesting product improvements, brainstorming design solutions, or architecting system-level engineering robustness—and in most cases, they yield identical alternatives.

The second thing is that ”tokenmaxxing“ is ending. Enterprises are reevaluating ROI.

Management went from going all-in on agents to introducing monthly limits on tokens and model tiers in less than a year. As frontier model costs rise, enterprises are now exploring lower-cost, open-weight, or in-house options that ensure proprietary data ownership.

For instance, Palantir’s Karp has openly criticized “tokenmaxxing” as a viable business model, noting that several of its US clients supporting critical infrastructure are already running open models (NVIDIA’s open model) [*].

These exact shifts give China’s open-weight models a distinct competitive edge.

Subscribe now

China’s token export playbook

There are two distinct export routes.

The first route is to move only the tokens. The power stays in China, the servers stay in China, the model runs in China, and what crosses the border is the answer, delivered by API. That’s the Shantou model, and it’s where most of the OpenRouter volume actually lives. It turns cheap Chinese electricity into a services export with no containers and no tariffs.

Among China’s frontier labs, DeepSeek is closest to a “pure technology export” and acts as the industry’s price anchor. Operating with no prioritized B2B or B2C application strategies and no overseas offices, its global influence spreads purely through its open-weight models adopted by tens of millions of developers worldwide.

Zhipu AI (2513 HK), deeply Tsinghua-rooted, bases its strategy on three distinct pillars: MIT-licensed open weights, agentic coding, and—most distinctively—sovereign compute. Its GLM architecture is trained and served heavily on Chinese domestic accelerators (including Huawei Ascend, Moore Threads, Cambricon, and Kunlunxin). Its revenue is heavy on B2B and B2G, with 85% of its H1 2025 revenue coming from on-premise deployments to Chinese SOEs and financial institutions. However, it is racing to catch up with Western frontier labs—recently reported to match Anthropic’s Claude Fable 5 on specialized cybersecurity tasks.

Moonshot AI’s Kimi-K2.6 is arguably the premier open-weight native multimodal model on agentic benchmarks, capable of orchestrating up to 300 parallel sub-agents for autonomous coding loops. Commercially, it is the fastest-inflecting lab in the cohort: its Annual Recurring Revenue (ARR) surged to $200 million by April 2026, and its models are reportedly feeding Cursor’s training — meaning it’s already inside the Western developer-tool supply chain.

The second route is to export tokens, applications, and ecosystems simultaneously. Among Chinese labs, consumer-facing applications mostly dominate.

MiniMax is the consumer exporter, drawing about 70% of its 2025 revenue from overseas, driven historically by consumer apps (Talkie companion app, Hailuo video) and now increasingly by the multimodal capabilities and M-series API, whose M2.5 briefly led all of OpenRouter. Although Moonshot also derives revenue from its B2C applications, its user base remains mostly domestic as of now. MiniMax is, by a wide margin, the company earning the most overseas C-end revenue among the Chinese AI players.

Alibaba (Qwen) is China’s leading player in exporting a full-stack B2B ecosystem. It consolidates its models, cloud infrastructure, and chips under the Alibaba Token Hub (ATH) business group, and is rapidly expanding its data center footprint globally across South Korea, Malaysia, the Philippines, Thailand, and Mexico.
A massive driver of this global B2B strategy is the high volume of international developers downloading Qwen’s open-weight models from Hugging Face to host them locally. Ultimately, Alibaba’s play is to commoditize these open-weight models so that global clients become structurally locked into renting Alibaba Cloud’s compute ecosystem to deploy them.

What holds it back

The most important constraints are trust and jurisdiction, not capability.

Under China’s 2017 National Intelligence Law, companies must cooperate with state intelligence work, so every API call landing on a China-hosted endpoint carries an exposure no price cut can fix. But because the weights are open, Western hosts like Fireworks and Together serve the exact same models without any data touching China, at roughly double the first-party price — still cheap.

Nevertheless, enterprise adoption of Chinese models remains overwhelmingly low compared with that of smaller startups and individual users. Regulated, high-stakes workloads in finance, healthcare, and government stay with Western proprietary models.

Large enterprises still overwhelmingly buy from Anthropic, OpenAI, Azure, and Google Cloud, and US labs still capture most of the industry’s actual revenue even while losing volume. For instance, Anthropic holds ~12% of token volume on OpenRouter but captures ~46% of revenue [*].

Apart from the obvious geopolitics and regulations, localized sales execution is a real challenge faced by Chinese AI labs that wish to go B2B abroad.

One interesting story I heard from a friend in the industry shows the level of localization needed: one of their clients ended up choosing MiniMax’s voice model because it was among the only models that could provide a voice solution with the specific regional accents the client was looking for in that market. Despite the hype about AGI, the reality is that the sheer amount of detail required for localization still demands significant on-site, attentive human effort.

B2B sales require a strong local sales and marketing team to build client relationships —this means considering the nuances of language, business practices, and even local culture. This is the area where many Chinese companies still lack, despite their strong technical capabilities and excellent price-to-value.

Investment implications

Subscribe now

Read more

In 2025, Hong Kong roared back to the top of the global leaderboards, hosting 119 IPOs that raised a massive US$37.4 billion, a blistering 231% surge over 2024.

2026 is even hotter. In Q1, HKEX just wrapped up its strongest first quarter in a long while, printing 40 new listings that raised a staggering HK$110 billion, nearly six times the capital raised in Q1 of last year.

The structural drivers are real and should not be dismissed. HKEX’s Chapter 18C framework for specialist technology companies has given AI leaders such as Minimax and Knowledge Atlas a fast track to the capital market. A+H dual-listing momentum is running hot, helping a long line of established A-share giants, from CATL to Montage, find an overseas fundraising platform in Hong Kong. Foreign capital that had been avoiding Chinese equities is dipping back in, looking for a regulated, liquid, internationally accessible entry point. Hong Kong is, structurally, better positioned as a listing venue than it has ever been.

But our memories still beg us to consider this: when the market runs this hot, what does it typically mean for the investors already sitting in the secondary market? Is the IPO boom a celebration of a rising market, or is it the last act before the curtain falls?

Where Does the Money Come From?

The intuition is simple. When a company lists on the Hong Kong Stock Exchange, it issues new shares to the public. The public pays cash. That cash goes to the company (or to selling shareholders). It does not stay in the secondary market.

In other words: a large IPO market is, by definition, a large liquidity extraction machine. Every dollar parked in a subscription form is a dollar not buying Tencent, CNOOC, or Alibaba. Every HKD 100 billion raised in new listings is HKD 100 billion that has, at least temporarily, left the pool of capital competing for existing shares.

Furthermore, behind every single one of those blockbuster IPOs sits a ticking clock. As the standard 6-month and 12-month lockup periods on these multi-billion-dollar tech, AI, and biotech giants begin to lapse over the coming months, a massive wall of secondary paper will suddenly become freely tradable and is about to hit the market.

The counter-argument is that IPO activity reflects investor confidence, which is itself a symptom of a rising market. Busy IPO calendars happen during bull runs, not bear markets. Correlation, if it exists, might run the other way: good markets attract listings, not the reverse.

The data, however, tells a more interesting story. The relationship is not contemporaneous, but forward-looking. Heavy IPO years tend to be followed by weak secondary market performance, sometimes severely weak.

Read more

On days like this, we think this is the perfect time to share with you a wonderful guest piece from that will give you a systematic review of risk, reward, and the right allocation strategy to China from the perspective of someone who has been working and investing in this part of the world for almost 2 decades.

Pandawatch‘s real name is Enrique Becerra, a Hong Kong–based investor who has managed his own capital full-time since 2017, focused on China A-shares and Hong Kong-listed equities. He is the author of the monthly letter “Investing in China”. He also tweets @pandawatch88. Based in Hong Kong since 2007, he spent 16 years as an M&A investment banker at Citi and Bank of America Merrill Lynch, most recently as Managing Director and Head of Asia Financial Sponsors, executing deals across Greater China, Korea, and Southeast Asia.

Enjoy!

Reconsidering Risk, Reward, and Allocation to China

By Enrique Becerra a.k.a

Why this article exists.

Back in January I published a deck called A Hitchhiker’s Guide to the China Stock Galaxy: 2026–2030. I made it because in conversations with fund managers I kept hearing the same story: it has become really hard to have a meaningful discussion with US and European allocators about China.

The moment someone in the room says “yeah, and what about Jack Ma”, “tutoring stocks”, or “Taiwan”, the conversation ends. Zero nuance. This piece puts these one-liners under scrutiny.

Because of the above, many have missed a 15% China rally in 2024, then a 30% rally in 2025, and then the doubling of the onshore tech index Chinext over the last 12 months.

Or at best, enjoyed a 3% allocation to it. Wait what, 3%?!

Intro: According to MSCI and FTSE, the China market is the size of Canada.

China is the second largest equity market on the planet, at about USD 22 trillion of market cap across Hong Kong, Shanghai and Shenzhen. It is also the second largest market in terms of overall liquidity and number of Fortune 500 companies.

However, China weight in the global index benchmarks MSCI ACWI and FTSE All-World is 3%. The same as Canada. Half of Japan.

If we were to re-do the MSCI ACWI by actual market cap, China weighting would be 15%. Similar to its share of global GDP, which makes more sense.

There are a few reasons why both MSCI and FTSE do the weights the way they do. It doesn’t mean you should too, as it may not give you a balanced global exposure.

I don’t expect this massive underweight to close anytime soon. Changing the methodology to more closely align to market cap would mean reducing US from 63% to 52% and corresponding outflows. Good luck to MSCI CEO in that endeavor. But the gap may eventually narrow, and this is something to keep in mind.

What is the right number? We come back to why this article exists: many institutions operate in an environment not conducive to having a nuanced discussion about China investment risks or rewards. And without an informed view on risk and reward, you can’t size a bet.

The article is divided in three parts. Part 1 talks about five risks. Part 2 talks about three rewards. Part 3 talks about strategies to allocate money.

The point of this piece is to help allocators re-examine their China assumptions. Do not treat it as a comprehensive review of risk or a scientific paper, but a source of ideas and ammunition to discuss with peers. The answer may still be a zero China allocation, but at least it’s an informed zero.

Part 1: A better understanding of Risk in China.

Some investors (not you of course, dear reader) approach China risk in the same way they approach the country weightings: “if MSCI says it is 3%, that is probably the right number.” Therefore “if well-regarded guru says that a conflict in Taiwan means China equities are a doughnut, that’s probably the right pastry.” Investors love narratives, heuristics, and outsourcing due diligence to others.

Below I examine five common China risk narratives:

• China is ultracompetitive, nobody makes any money

• Your upside is capped, the government doesn’t let companies become too big

• The core mission of Chinese companies is different from US companies

• Regulations move fast in China

• Taiwan

The first two sound good in the echo chambers, but are light in substance. The last three are more serious.

Risk 1: “China is ultracompetitive, nobody makes any money.”

Saying this has become very trendy recently. People will reference a book author that makes this point in every podcast, or a post on X from someone that just did a factory tour in China. It sounds smart, fits many priors, and if you look at the actual numbers, it turns out to be inaccurate.

As with every good-sounding story, it must have a piece of truth. And for sure: certain industries, like solar, are incinerating money.

The missing truth however is that in a majority of very competitive industries, like EVs, consumer electronics, retail, industrial automation, etc, you have market leaders making good money: BYD was loss making in 2019 and made USD 4bn in profits in 2025, while its competitor Geely doubled profit this year to USD 2bn. Xiaomi was making USD 1bn in profit in 2019 and made USD 5bn in profits in 2025. Midea and Haier, competing white goods manufacturers, or Anta and Lining, the two largest Chinese sportswear brands, have all doubled profits in the 2019 – 2025 period, while navigating a very tough and deflationary consumer macro environment. Among industrial manufacturing names, heavyweights like Nari or Inovance have doubled and tripled their profits respectively in the same period.

Survivorship bias in the sample? Maybe. What about the generic widget manufacturer with 1% market share? Yes, that one is probably struggling. It should be noted here that 40% of the companies in the Russell 2000 have negative earnings. Losses are not a uniquely Chinese export. Takeaway for risk management? Do not predict, observe, invest in the ones growing profits. Another takeaway: question hot takes from podcasts, numbers please.

Profits do not always go in a straight line up and to the right. There are investment cycles and there are collecting cycles. The combined profit pool of the top 10 retailers in China doubled from USD 29bn in 2019 to USD 62bn in 2024 despite the industry experiencing intense ecommerce rivalry in 2021–22. In contrast, in the same period, the top 10 US retailers, despite a milder competitive environment and much better consumption macro with stimmy checks, just grew their profit pool by 50%.

Back to China, 2025 saw another investment cycle begin among ecommerce players aiming to expand the pie of quick commerce and fight for the largest piece. We’ll see how that shapes out in 2026-27, and if we come out the other side with usd100bn combined EBIT.

China Top 10 retailers EBIT (USD bn).

Competition? Big. Killing industry profits? Not necessarily.

Risk 2: “Your upside is capped, the government doesn’t let companies become too big.”

China currently has 23 listed companies with market cap above USD 100bn, and 8 of them are controlled by entrepreneurs. Ten years ago, it was basically just one. Entrepreneurs and their shareholders have made money.

At the start of 2026, Hurun counted 1,000 billionaires in China vs. 550 in 2016, doubled. The top 10 today are worth USD 490bn vs. USD 190bn in 2016, more than doubled.

If the system was set up to chop tall trees, one would expect the counts above to be stagnant. What is stagnant is the number of billionaires publicly criticizing SOE banks.

Let’s move on to what I consider more serious risks. The first is more philosophical, the next two more tangible.

Risk 3: The core mission of Chinese companies is different from US companies.

Borrowing some reasoning here from Midea’s chairman: the history of the Chinese equity market is too short. Private companies and the stock market have existed in the US for 200 years. In China, for less than 40 years.

In the US there is a well-established understanding: companies are not expected to shoulder excessive social responsibilities; their core mission is to maximize profit and shareholder value.

In China such consensus has yet to be formed. The experience is too brief. Right now, the stock market exists to serve “the real economy”, and companies exist to serve “national development” goals. An investor in China must be aware of what these goals are.

For example, from a recent article in Qiushi outlining economic priorities: “…to foster a group of new growth engines, including new-generation information technology, artificial intelligence, biotechnology, new energy, new materials, high-end equipment and green industry during China’s 15th Five-Year Plan period (2026-2030).”

If you don’t pay attention to the shift in goals, you risk missing the action. Or worse, you risk being offside, which takes me to our next risk.

Risk 4: Regulations can move fast in China.

For better and for worse. We are going to focus on the “for worse” here, with a few examples. There’s nothing too esoteric behind regulatory actions and the tea leaves come in capital letters. The issue is execution, which is faster than in other jurisdictions and can lead to wild price moves.

First, in many cases regulatory action is actually not that fast, but still somehow manages to surprise investors:

• The afterschool tutoring ban came after 2 years of warnings (in NPC statements and state media) about the pressure this product was putting on families. Investors ignored and kept buying, chasing price. When the regulatory hammer came, EDU and TAL were trading at 80x LTM PE

• Last week, after three years of warnings, FUTU and two other online brokers got fined by CSRC around 15-20% of annual profit for servicing mainland residents without a securities license (people were wondering if CSRC had forgotten about them, nope they had not). Investors learn though, and this time when the penalty arrived the valuation was not 80x PE but 12x PE, already carrying the regulatory uncertainty

Other times regulation does come very fast as a response to specific company actions like:

• Didi rushing a US IPO in June 2021 ahead of a new data law coming into force in September 2021 (risk factor #6 in the prospectus). Two days after the IPO, CAC launched a cybersecurity probe, banning new user registrations and pulling the app from domestic stores

• The well-documented drama around Ant Group, which needs no recap

And in other cases, regulatory action comes to companies where users complain about market dominance, but the trigger and timing can be random and abrupt. For example:

• Antimonopoly investigation into Trip.com, where small hoteliers had been complaining to local regulators that the platform was forcing them into exclusivity and to give up control over room rates. Trip got a few small fines over the past couple of years, but eventually SAMR decided to jump in. Meituan went through a similar process in 2021 and came out with a USD 500m fine but higher revenue and margins (they have bigger problems now, I’m sure they miss SAMR)

• The caps on pricing for online consumer lenders like QFIN and FINV, an industry where in the risk factors of smaller players you can read gems like “we indirectly charge our borrowers in the manner our borrowers are not aware of.” Turns out borrowers found out

• Regulatory action aimed at companies extracting rents is not a uniquely Chinese export, see FICO and UnitedHealth investors for details

Finally, regulatory action can provide opportunity. Moutai was a 12x bagger in 5 years after regulations to curb government spending on banquets and gifts took the stock down 50%. It is important to distinguish between regulations targeting the company, the product, or the behavior.

Practical implications for risk management: there are 8,000 listed companies in China, minimize your exposure to those working against the direction of policy (see risk 3). If you want to be involved because the risk reward set-up makes sense, size appropriately.

Risk 5: Taiwan.

Everyone has their own very thoughtful take on CCP stated goals, DPP stated goals, the outlook for the new KMT president, and the latest DJT commentary. Fine.

As an investor looking to measure this risk, I just look at the issue in two ways. First, the incentives for something to happen, and second what then if something happens.

In the short and mid-term, the incentives for all three parties point against escalation. US does not want to lose access to TSMC fabs or the supply chain for DC infrastructure components. It would slow down AI progress. Intel is not going to make cool Rubins anytime soon. Further, without access to TSMC, Nvidia, Microsoft, Google, and Apple share prices could be cut in half. The other two parties would suffer tragic human losses. No one wins much.

But accidents can happen. And here investors do a quick “reasoning by analogy” and assume China stocks are a zero, because of what happened to Russia. The analogy doesn’t work because China is a different global economic actor than Russia. In a Taiwan conflict, with the global tech supply chain in disarray, are you so sure that US and Europe will go down the massive sanctions route, and further spread the pain to industries that rely on access to China supply chains or sell to Chinese buyers? Maybe yes. But maybe not.

Before assuming this scenario is a zero for China stocks, a bit more work should be conducted on incentives, dependencies, and resilience. Once the work is done, maybe we find out that one way to diversify this risk, which would hit global equities hard, is to actually own some China. Counterintuitive? Huawei chips are not using TSMC fabs, and China has its own domestic datacenter supply chain.

Hope we don’t have to test any of this. People on both sides just want to keep enjoying life, eating good food, and having fun with friends. Let’s keep it that way.

[Baiguan’s note: Robert recently wrote about why it is highly unlikely for China to take military action against Taiwan]

Wrapping up the risk section.

Other risks not discussed here, but that you should think about, include:

• “The GDP has doubled, but the stock index has not, means massive dilution” which if you are versed in basic arithmetic you can quickly dismiss by looking at the expansion in the number of index components

• Known-unknowns such as deflationary cycle and population contraction

• Unknown-knowns such as what comes after the Xi era

• And unknown-unknowns which, you see, I don’t know

Human systems live in a critical state. Never underestimate what can come next.

Part 2: The Reward side of the equation.

Like the risk side, the reward side in China often suffers from the same: too much story and little substance. Back in 2010 we IPOed restaurant companies in China with “80 locations but TAM of 1.2bn people”.

Let’s tackle the reward side in a more serious and tangible way. I’ll cover three:

• At the portfolio level, China offers diversification of economic outcomes

• At the company level, geopolitics is paradoxically expanding TAM and accelerating growth

• At the market level, domestic capital markets are going through a step change that is bringing new ideas to investors

Read more

The China Securities Regulatory Commission (CSRC) and seven other government agencies announced that these firms violated domestic laws by facilitating unauthorized overseas stock trades for mainland retail clients without the necessary onshore licenses. In response to the breaches, regulators are imposing severe financial penalties, including a proposed $271 million fine for Futu and over $50 million in combined fines and asset confiscations for Tiger Brokers.

As part of a two-year campaign to aggressively enforce the country’s strict capital controls, the targeted brokerages are now barred from accepting new investments from mainland clients and have been ordered to systematically wind down existing non-compliant accounts.

Is this a surprise?

While the sheer size of the fines and the strict two-year mandate to entirely shut down operations for mainland clients made headlines, the writing has been on the wall for years. This recent announcement represents the final “dropping of the shoe” in a 5-year-long regulatory campaign rather than a sudden policy pivot.

Here is a timeline of key events leading up to today:

• October 15, 2021: The CSRC first expressed its displeasure with cross-border online brokers’ “illegal activities” through media reports.

• December 30, 2022: The CSRC issued an official warning that Futu and Tiger Brokers were conducting cross-border securities business for mainland investors without the necessary approvals. The regulators explicitly labeled these activities as illegal and ordered the firms to stop soliciting new mainland clients, while leaving open the question about how to treat existing clients.

• January 13, 2023: The CSRC released the Securities Brokerage Business Management Measures, which formally established that “domestic securities business must be licensed” and expressly prohibited foreign institutions from conducting unauthorized marketing, account opening, or trading services.

• May 2023: In a major step toward compliance, both Futu and Tiger Brokers voluntarily removed their trading apps from online app stores in mainland China, effectively cutting off the easiest access point for new domestic retail investors.

• July 2025: The China Securities Journal exposed how Tiger Brokers was still attempting to circumvent the regulatory ban by using “edge-pushing” (擦边) tactics, such as using WeChat customer service agents to supply mainland investors with fake overseas stock trading records, to help them qualify as “existing overseas investors” and bypass the compliance blocks.

• May 22, 2026: Friday’s final regulatory decision was made.

What’s the impact on online brokers like Futu?

According to various sources, about 13% of Futu’s accounts are held by mainland investors, which represent around 20% of Futu’s client assets and 25% of revenue. Theoretically, if all of those mainland investors sell their shares, that would be about a 20-25% hit to Futu’s financials.

The actual impact could be smaller because the government is also making it clear that mainland investors who live overseas and trade overseas will not be affected, limiting the actual impact.

Moreover, in practice, determining if someone is “living overseas” is more nuanced than it sounds. It is certain that Futu will close all its mainland-based operations and will likely block visits from mainland China. However, in practice, it is entirely conceivable that someone with a VPN and some form of proof of a foreign address could establish themselves as a foreign-based investor and thus continue trading.

What’s the impact on China ADRs

Futu and Tiger Securities' stocks collapsed due to this news. Along with it, the entire China ADR and internet space also dropped on Friday. There were fears that the “forced selling” by these brokers would exert significant selling pressure on the ADRs.

But these fears do not stand up to scrutiny.

Of the mainland investors investing through Futu and Tiger, it is likely that purely US stocks like Nvidia and Tesla make up the majority of their portfolios. This is because many dual-listed ADRs (e.g., Alibaba) and most large to mid-cap Hong Kong stocks (Tencent, Meituan, etc.) are already accessible through the Stock Connect program, allowing mainland investors to invest directly from onshore. Consequently, there is no particular reason these investors would go all the way overseas to invest heavily in China ADRs.

Let’s take Tencent, a popular Chinese stock listed overseas, as an example. According to the Hong Kong Stock Exchange’s own data, investors who bought Tencent shares through Futu accounted for only ~0.8% of Tencent’s total share base. Assuming 20% of those are held by mainland investors, Tencent's actual stake “at risk” would be less than 0.2%.

For the bigger picture, at the end of 2025, Futu held around $160 billion in total client assets. At 20%, around $32 billion of those belong to “mainland investors”. If we assume 30% of those were held in overseas-listed Chinese stocks, that would amount to only $10 billion. On the other hand, all of the companies in the KraneShares CSI China Internet ETF (KWEB) alone are worth more than $1 trillion. So, at most, it will have a net selling impact of just 1%. And if we take into account all non-KWEB Chinese companies and the aforementioned fact that some “mainland” money would still remain, the final impact will be far less than 1%.

Is this the beginning of a market-wide, 2021-style regulatory crackdown?

I suspect the actual reason for the broader sell-off was the PTSD that investors still had from the terrible memories of the 2021-2022 period of government crackdown, such as the infamous one on after-school tutoring. At the time, there was a consensus (which I didn’t agree with) that there was a systematic and concerted effort to crack down on the private economy.

Is this regulatory action against overseas online brokers a sign of the bigger crackdown? I don’t think so. This is clearly an industry-specific regulatory action that has been lurking around for years.

The regulator, drawing on lessons from the past, seems also to be trying to compartmentalize the issue. The actual fines imposed on Futu and Tiger for the illegal offenses are lower than many expected, and a two-year window for asset liquidation seems very reasonable. Also, as mentioned previously, it took them about 5 years of piecemeal regulation and cat-and-mouse games to get to this point.

Overall, the authorities did not seem as heavy-handed as they were in 2021 during the “double-reduction” episode, and the market has now developed a normalized view of the regulatory landscape.

Any other stones unturned?

By now, a natural question is whether there are still similar “slow-burn” regulatory stories lurking in the corner that we should be aware of.

In the paywall section below, I’ll give a full review of 2 potential flashpoints and 2 areas that are seemingly flashpoints but aren’t.

Read more

Every AI boom needs infrastructure. In the US, that conversation centers on hyperscalers racing to build data centers. In China, the same logic applies — but the players, incentives, and market structure look very different. Understanding this landscape is increasingly relevant as capital continues to flow into China’s AI infrastructure story.

China’s AI data center (AIDC) market is in the middle of a massive buildout. Goldman Sachs projects China’s top internet firms to invest $70 billion in data centers in 2026, with power demand from the sector rising 25% that same year.

For context, the US still leads on accumulated computing power: a Federal Reserve analysis from late 2025 estimated the US controls roughly 74% of global high-end AI compute, with China at 14%. The spending gap reflects this: the four largest US hyperscalers (Alphabet, Amazon, Meta, Microsoft) have collectively committed around $650 billion in AI infrastructure for 2025–2026, versus China’s $70 billion. While China lacks an accumulated share of computing, it is racing to close through a faster power infrastructure buildout.

Disclaimer: This is not investment advice. Please conduct your own research before making any financial decisions.

The China data center market is not one market. It splits along two axes that matter for investors. The first: who owns the infrastructure — third-party neutral operators, state-owned telecom giants, or the hyperscalers themselves? The second: what does the infrastructure do — massive GPU training clusters for frontier AI, or edge inference for production deployment?

The picks-and-shovels layer: who actually builds the GPU warehouses

China’s cloud and data center market has split into two.

Subscribe now

Read more

Five years ago, mid-sized companies (100–999 employees) contributed a significant portion of the jobs. If we take online recruitment platforms as a proxy to analyze China’s labor market, six out of every ten jobs came from a mid-sized employer in 2020. But today, fewer than four out of ten do.

These postings haven’t migrated up to large enterprises; instead, they’ve migrated down to small companies with fewer than 100 employees.

Two possible stories could explain this:

1. Companies are becoming more entrepreneurial, potentially benefiting from productivity gains driven by AI and automation.

2. Mid-sized firms are laying off staff and posting fewer jobs, which mechanically lifts the small-firm share without any sector actually ‘winning.’

There could certainly be a combination of these two factors—as well as other, more complex dynamics at play. However, by analyzing the relationship between changes in job posting volume and average salaries, we can gain a clearer sense of the underlying reality. (For instance, a sector benefiting from productivity gains would likely see both job growth and rising salaries, even if those roles are concentrated in smaller companies. )

Crucially, this overall shift reveals significant nuances across different industries. We analyzed 13 major industries that provide the bulk of China’s online job opportunities, examining 75 months of data since 2020 to determine which narrative holds true for each.

Specifically, for each sector, we looked at the following angles:

• The share of all job postings by company size: To see whether the broad structural shift holds true for the industry.

• Change in total job postings and average salary by company size: Is the pie growing or shrinking?

• Employment quality: We use the share of part-time/gig jobs over total jobs as a proxy. While not a 100% accurate representation, an increasing share of part-time employment could indicate that companies are prioritizing cost flexibility, and that employment stability is declining.

Note: Online recruitment platform data may carry a bias toward white-collar roles.

“Charts of the Week” is Baiguan’s series that features key data points to help you quickly grasp the general state of affairs in China in just a few minutes. We handpick the highlights of the data charts from a variety of sources, analyzing and delivering insights trusted by 100+ top institutional and corporate clients worldwide at BigOne Lab. Don’t forget to subscribe before you continue reading!

Subscribe now

Read more

Subscribe now

Last month, we asked: Is China’s Jan–Feb strength organic, or just stimulus? Two pillars are important for answering this question: the wealth effect (real estate) and income (salary). Through the week ending April 11, both still point up.

Real estate: activity broadens after the Chinese Spring Festival

Property viewings and transaction volumes are both tracking materially above their 2025 levels following the Spring Festival. Specifically, the transaction volume of second-hand homes has surpassed the peak following the 2023 Spring Festival, when a surge was driven by pent-up demand from COVID.

The recovery of the second-hand home transaction also seems even across city tiers:

The inventory level also improved materially across all city tiers, albeit at different paces. Using total listings / weekly transaction volume * 4.33 as a proxy (where 4.33 is just the average weeks per month), the average “months of supply” now trends down for every city tier post Spring Festival.

Share

Bottom line: the real estate recovery is widening, not narrowing. Although signs of a strong rebound are not in the near future (we see the listing price still trending down slightly or holding steady), real estate’s weight on the economy is decreasing.

In the previous post, we also observed from job posting data that the real estate and related industries (the orange area) saw their share of total recruitment cut in half, dropping from 15% in 2020 to less than 8% by the end of 2025.

Salaries: healthy growth in March

Average posted salaries on Chinese online recruitment platforms bottomed in November 2024 and have accelerated almost every month since.

March 2026 printed +5.7% YoY — the strongest reading since mid-2022 — with MoM momentum also picking up.

Two caveats: the data sample can skew toward while-collar roles. Additionally, the salary growth may not be even across industries. As we’ve previously pointed out (see Old to New), higher-paying “new productivity” jobs in tech and AI-adjacent roles are crowding out lower-paying ones.

But even after adjusting for both, a full year of accelerating wage growth is a meaningful break from 2023–24.

A note to readers: We recently launched Baiguan Pro, a professional tier for investors seeking more in-depth analysis directly related to Chinese equities. There, we share our team’s monthly view on Chinese assets, built on macro data points like these, along with other proprietary data and market inputs. If the Charts of the Week help you calibrate the China investment, Baiguan Pro takes it a step further.

Interested readers can read the inaugural post here.

Related Reads

As early as August 2024, right before POP MART’s shares really took off, we first commented on its stellar growth based on our own data.

In June 2025, took a deep dive into the consumer psychology around this company, including her own personal conversion as a consumer.

The stock price of POP MART recently had a major setback. After its annual earnings announcement, the stock has tanked by more than 30% over five sessions, wiping out around HK$100 billion in market value.

Three main reasons explained this market’s sudden lack of confidence. First, the market is concerned that the overseas business performed worse than expected. Second, there are concerns that POP MART relies too much on a single IP, namely Labubu (The Monsters). Third, Wang Ning, the founder and CEO, gave an ultra-conservative guidance of “no less than 20% growth” during the earnings release.

In the first half of 2025, overseas revenue skyrocketed 440% year over year. However, in the second half of the year, the overseas revenue growth rate “slowed down” to 243% year over year.

240%+ growth is certainly slower than 440%+ growth, but we can’t help but feel that calling this a “deceleration” runs the risk of nitpicking.

POP MART’s phenomenally explosive global growth story was a huge surprise for everyone in 2025. The market was naive to build a linear extrapolation of future growth rates from this level in the first place, and it is only natural to expect that, after such triple-digit hyper-growth, the company is bound to see some deceleration.

On the other hand, it’s important to note that China's sales have been remarkably robust and resilient. In the first half of 2025, Greater China revenue grew 135% year over year. In the second half, the growth rate remained virtually unchanged at 134% year-over-year. This demonstrates that China remains the main growth engine for POP MART, with strong and stable growth momentum that has not lost steam.

Our own proprietary data, tracking in-store transactions in China, also shows the growth rate in Q1 will be substantially higher than “no less than 20%”. (You can find the exact data behind the paywall.)

As for over-reliance on IP. It’s true that Labubu now accounts for 38.1% of total revenue, in 2025, up from 23.3% in 2024. However, the growth picture is more nuanced than it first appears. Labubu grew by 365.7% year-over-year, which is indeed the fastest growth rate among POP MART’s established IPs. Yet the company’s portfolio is diversifying rapidly. DIMOO grew at 205.3%, SKULLPANDA at 170.6%, and CRYBABY at 151.4%. More significantly, Twinkle Twinkle, positioned as the next major IP, grew at an extraordinary 1,601.6% year-over-year, from just 120.8 million yuan in 2024 to 2.06 billion yuan in 2025, now capturing 5.5% of total revenue.

If Labubu had not existed, you would still consider this company a high-growth one, if only with these IPs. An investor simply should not blame something for growing even faster than its fast peers.

We believe the real trigger for the big share price slump is the highly conservative “guidance” that Wang gave to the market.

Overall, POP MART grew 185% in 2025, reaching 37 billion yuan in total revenue. But now the management slapped a mere “20%” on the market’s face. A violent re-rating is only natural afterward.

According to Wang Ning’s explanation. It was a voluntary decision to decelerate and regroup internally after a “F-1-style” helluva ride. But the market is usually skeptical of such a statement. Is it a genuine choice to decelerate and regroup, or a just cover-up for a company that’s losing steam?

Regardless of the reason, it is certain that the company won’t pursue rapid growth in the near future. In a hot stock that has been dominated by long-short hedge funds, this is enough to scare many investors away.

At the same time, we start to see signs of super-long money coming into the scene. Duan Yongping, often dubbed as “Buffet of China” and a mentor of PDD’s Colin Huang, very publicly retracted his previous comment that he would not invest in POP MART, and praised POP MART as “right business, right people, and right price” and a “real precursor of the internalization of Chinese brands”.

It seems a rotation of investor types is underway right now.

Is POP MART valued correctly?

POP MART inspires many analogies. Back when it was a hot name, analysts often imagined POP MART would become the future Disney of China. While a closer analogy might be Japan’s Sanrio (8136.JP), creator of the Hello Kitty franchise.

We don’t believe POP MART is Disney or Sanrio, yet.

The crucial difference is that POP MART is still too young. By simple definition, when someone is young, they can’t prove that they are evergreen.

If you're looking for an analogy to understand POP MART, we think it's more like a video game company right now.

The situation is similar to Tencent Games back in 2017, when one super franchise, Honor of Kings, accounted for ~50% of its mobile game revenue. At the time, investors were also concerned about the sustainability of the situation.

Just like Tencent Games in 2017, POP MART has not yet demonstrated the longevity of the super franchise. On the other hand, the success and runway of each franchise are far from predictable. Even their own management cannot predict them. They also don’t have a long enough track record to prove that they can keep producing hit after hit.

Adding to the problem is that, so far, POP MART’s IPs lack narrative power. This is a key difference between POP Mart and Disney at this moment. Disney’s IPs have narratives and stories, while Labubus are storyless, making it even harder for investors to look ahead.

All of these mean that the current valuation of POP MART, at a low double-digit TTM P/E and even single-digit forward PE, is exactly where it should be.

Investors love linear extrapolation. Investors hate it when there is too much uncertainty, and they can’t perform linear extrapolation due to low visibility. POP MART is thus rapidly, violently gearing down from a high-growth, high-multiple story to a stable-growth, low-multiple value play that many institutional investors would not want to participate in.

Looking ahead over the next few years, we should focus only on whether the share price can be driven by actual EPS growth, and assume that re-rating will occur only after years, and even decades, of solid, healthy growth.

Even if you are a POP MART bull, this is the reality you have to be clear about.

So, should we expect POP MART to sustain long-term growth?

We think we should.

In fact, we believe if there is any company that is able to deliver a long-lasting enterprise value capitalizing on China’s (and the world’s) growing demand for high-emotional-value consumption, it’s got to be POP MART.

There are at least three reasons to support our conviction: people, platform, and attitude towards investors.

First, let’s look at it from a first-principle perspective: for a business as ephemeral and as unpredictable as that of POP MART, what’s the single most important factor to consider? Technology? Marketing capability? Cash?

None of that.

The single most important factor is people and culture.

This is not the first time that POP MART has faced a crisis. In fact, the “crisis” of today is far less significant than previous ones.

When POP MART IPOed in 2020, investors questioned whether the company relied too much on its main IP, Molly. There was the same debate over whether the business model worked and if the company could keep delivering long-standing IPs. Following that, the share price fell by more than 80% from its IPO price by 2023, before rising 30-fold over the next two years.

And even further back, POP MART had been building this unique and lesser-known business for more than a decade before its IPO. There weren’t many overseas models to copy. Many people do not understand the model. It’s not a red-hot sector with an existing roadmap to show them the way. They have always been lonely.

It’s also when Wang Ning developed the core tenet in his management culture: “Respect time, respect operations” (尊重时间, 尊重经营), which sounds very much like benfen of Duan Yongping and Pinduoduo.

Over the last 16 years, the company has proven it can survive cycles, find its own audience, and keep innovating on its own timeline and pace. They seem able to consistently make the right choices and are unafraid of being alone or singled out, which is a remarkable characteristic in Chinese business culture and society.

This is why we believe this current voluntary deceleration of growth in exchange for operational improvement is real, and not a cover-up for slow growth. It’s yet another sign that this company does not care about short-term performance but is squarely focused on the long term.

It’s also important to understand that “no less than 20%” is not “guidance” in traditional Wall Street speak. The “20%” number is the least Wang Ning thinks they can deliver with 100% certainty. Anything above is not planned for and will not be the focus of this year. It’s the equivalent of saying there is no quantifiable business target for this year at all.

It’s healthy that they are sober about where they are, despite the immense market expectations. This is the right way to do it, especially in a business that nurtures and manages intangible assets. Focusing too much on short-term financial metrics would only kill its soul.

The second reason for our confidence is that we should not forget that POP MART has already achieved incredible success, leaving it with a formidable moat that won’t be easily surpassed by copycats in the short term.

The most formidable asset is its brand. POP MART now has significant branding power, not just at the individual IP level but also at the company level, both in China and, very uniquely, worldwide.

This immense brand equity gives them a significant edge over copycats and competitors in their dealings with all stakeholders, not just consumers. For example, they have moved into prime real estate locations across China and in many parts of the world, ensuring that each of its new product lines and IPs will be the first for their consumers to encounter. They have also built a great reputation with designers and a strong connection with their core community, which gives them a real edge in continuing to produce new product lines.

Last but not least, despite POP MART’s long-term mindset, they also care about shareholder interests.

This is unlike PDD, which is also a very long-term-focused company but has never shown a commitment to boosting shareholder returns, the main factor driving PDD’s extremely low valuation now. Most people do not doubt that Pinduoduo is a superbly run company. But investors are concerned that if a company never cares about minority shareholders' interests, the very foundation of any financial modeling based on distributable discounted cash flows will be pointless.

POP MART is different: it has also proven itself to be an investor-friendly company. Whenever the share price dropped dramatically, it bought back shares aggressively. In the several trading sessions since its share price tanked, the company has already repurchased close to 1% of its total capital. An aggressive share buyback will also be part of POP MART’s path toward stronger EPS growth, alongside sheer profit growth.

The latest data

Behind the paywall, you will see the latest data from BigOne Lab, our parent company, about POP MART’s latest performance in China.

The data below are currently accessible to all paid subscribers. But in the next few months or quarters, we will provide regular updates of this data (and similar data) for Baiguan Pro subscribers. One of the many key differences between a paid subscription and a Pro subscription is that Pro subscribers will receive much more ongoing data updates.

Once again, for existing Baiguan paid subscribers: We are offering an upgrade to the Professional tier, giving you full access to all Baiguan Pro content. To upgrade, please REPLY to this email with the email address associated with your paid subscription. We are processing these upgrades manually through the end of the second quarter (June 30). (Please note that if you opt-in to the Baiguan Pro tier, your subscription will renew at the new rate if auto-renewal is enabled.)

Read more

Subscribe now

China’s Jan & Feb prints were improving across the board

Inflation: In January–February 2026, China’s CPI rose 0.8% year-on-year, and Core CPI hit 1.3% — the strongest reading since January 2023. PPI also showed encouraging signs at the factory gate, with industrial producer prices rising for five consecutive months on a month-on-month basis.

Spring Festival spending held up well, with services standing out. We covered this in detail in a previous post.

The key question: Is this organic or policy-driven?

This is where the honest read gets more nuanced. Much of the consumption uplift to date has been policy-assisted — trade-in subsidies, vouchers, and a record-long nine-day holiday that pulled spending forward.

The 2026 Government Work Report set a GDP target of “4.5% - 5%”—realistic but unexciting. The report prioritizes technology and new productive forces. The report prioritizes technology and new productive forces. Among its measures, China plans to issue RMB 250 billion in ultra-long-term special government bonds to support trade-in subsidies for consumer goods and to establish RMB 100 billion in fiscal-financial coordinated funds to boost domestic demand. To put that in perspective, these measures amount to less than 0.2% of GDP — hardly a massive fiscal stimulus. There is also no large-scale direct transfer to households on the table.

The elephant in the room: Is the strong Jan & Feb spending organic, and will it continue?

To put this into perspective, in today’s Charts of the Week, we return to two key pillars of domestic consumption: the wealth effect (real estate) and income expectations (salary).

Our January–February proprietary data on both fronts points in a positive direction.

Real estate: Our tracking of China’s online real estate platforms shows the second-hand housing market stabilizing under policy support, with transaction volumes in 90+ cities up 15% YoY. Viewing activity also picked up, with week-on-week viewings across 80+ sample cities rising 15%, reflecting strong buyer interest and a recovering market.

Salary: Average salaries growth across online recruitment platforms are gradually recovering, rising from near-zero in 2024 to around 4–5% in early 2026. This steady wage growth supports household purchasing power, suggesting consumption could continue at a measured pace.

Importantly, the worst of the real estate bubble’s impact on the workforce appears to be behind us, as we covered earlier:

Additionally, in select discretionary categories, we are already seeing what looks like organic spending growth that isn’t purely subsidy-driven. That is the signal worth watching. For example, we wrote about the Sports & Health sector yesterday and will follow up with more examples in future posts.

One additional tailwind: China sits in a different risk bucket than the rest of Asia

The escalation of the Iran conflict has rattled Asian markets, but not all of Asia is equally exposed. China actually sits in a more defensive position, holding large strategic and commercial oil reserves that help cushion short-term disruptions. It has also spent years diversifying its supply toward Russia and Central Asia.

In contrast, Japan relies on the Middle East for roughly 90% of its crude oil imports. South Korea sources about 70% of its crude from the region and routes over 95% of that through the Strait of Hormuz.

Market reactions reflect this. Compared to other Asian markets, China has seen a comparatively modest decline since the conflict began.

A-shares also carry a structural cushion that most markets lack: the “national team” actively intervenes when needed, reducing correlation to external shocks in ways that are hard to replicate elsewhere in the region.

Therefore, I am more constructive on Chinese markets, especially the on-shore market at this particular time.

We will continue tracking real estate and job market data — crucial macro datapoints for our paid subscribers in future posts.

Share

In this newsletter, I’ll walk you through some of my observations from this year’s festivities and share where the potential investment opportunities lie.

Overall trend: Strong demand; the rise of “Chinese aesthetic” in tourism

This year’s tourism numbers are impressive: Domestic tourism during the 2026 Chinese New Year holiday grew 19% year-over-year, reaching 596 million tourists, with tourism revenue increasing by 19% to RMB 803 billion. This means per-tourist spending, averaging around RMB 1,348, remained steady compared to last year. Outbound travel also saw a strong rebound, with a 24% year-over-year increase to 4.8 million travelers.

However, beyond these headline figures, the tourism boom is becoming more structural. The year 2026 may very well be the year when the “Chinese Aesthetic” takes full control of the tourism market.

Subscribe now

What the Spring Festival Gala tells

To understand the social climate in China today, one can’t overlook what happened at the Spring Festival Gala, an event watched by almost every Chinese person, or at least used as background noise. This year, the Gala featured more AI, robotics, and other high-tech elements, which have already been widely discussed. However, what often goes underreported is the increasing presence of foreign-language programs—something that very rarely appeared in recent years’ Gala. This shift signals a change in the social climate, moving away from political focus toward a more open-minded, economy-driven, and everyday-life-oriented vibe.

Most notably, rather than hosting the entire performance in the central television studio in Beijing, the Gala continued its format of holding satellite events across various cities. These events highlighted the unique cultural heritage and innovative economies of these locations, further showcasing China’s diverse regional identities.

During the Chinese New Year, famous tourist cities tend to get crowded, but this year, the Gala program selected cities that are less congested yet still rich in tourism resources or unique culture. Four cities were selected to host the satellite events: Harbin, Yiwu, Hefei, and Yibin, which are expected to drive the tourism sector forward.

The Spring Festival Gala serves as a microcosm of this broader trend: China has been increasingly promoting smaller cities and their tourism offerings, incorporating more intangible cultural heritage, and catering to the growing interest in unique local craftsmanship and cultural heritage. We have been covering this trend since 2024, which explains why smaller and lower-tier cities have outpaced big cities, provincial capitals, and traditional tourist hubs in tourist growth in recent years.

Beyond obvious factors like crowding and rising transportation costs, tourists have developed a genuine interest in “in-depth travel”—exploring the rich cultural and historical diversity of China’s provinces. (I often feel that visiting different provinces and regions in China could feel like stepping into a completely different “country,” each offering countless unique historical and cultural experiences developed and fused over a long history. This vast diversity will continue to fuel curiosity and growth in the tourism sector.)

Investment Implications: Overall, this trend is positive for hotels and airlines. The tourism boom is unlikely to be short-lived. People are genuinely exploring various cities, towns, and cultures, not just indulging in post-pandemic “revenge consumption.”

Subscribe now

Trend 1/4: The rebound of luxury spending

Hainan duty-free sales have seen growth since the official establishment of the Hainan Free Trade Port (FTP) last December. In January, Hainan duty-free sales grew by 45%. During the 9-day holiday period, the sales growth slowed to 15% year-over-year, but shopper counts still maintained a steady +20% year-on-year increase.

Outbound travel to Hong Kong also rebounded during the 9-day holidays, largely driven by duty-free shopping and high-end retail.

Investment Implications: The revival of luxury spending could continue this year, partially supported by the rally in the stock market, which could help improve sentiment among middle-class and affluent families. This should also benefit Hong Kong tourism and commercial properties, especially in high-end retail.

There could also be idiosyncratic opportunities, such as Laopu Gold, which benefits from the increase in jewelry spending, the cultural preference for gold in China, and the perceived value retention of gold amid rising gold prices.

Trend 2/4: “Experiential” over physical products

This is a structural trend that I previously explored in our inaugural post this year: 2026 China Equity Outlook. During this Spring Festival, live events, shows, concerts, and “emotional value” consumption continued to trend high.

During the 9-day holiday, Damai (HKG: 1060), China’s leading ticket booking platform, saw the number of performance projects grow 27% YoY. Specifically, Livehouse booking surged 65%, and immersive shows grew over 40%. Live events have become a lifestyle for the Chinese consumer.

“Mood boost” consumption, such as pop toys (潮玩), continues to trend high. Take Pop Mart, for example: without going into detail in today’s post, we still see triple-digit sales growth in January leading up to the 9-day holiday, extending the growth and IP rotation trend explored in our deep-dive previously.

Investment Implications: The rising demand for live events and “emotional” retail should be beneficial to booking platforms and IP-heavy brands.

Trend 3/4: Surge in agentic bookings

Subscribe now

Read more

“Charts of the Week” is Baiguan’s series that features key data points to help you quickly grasp the general state of affairs in China in just a few minutes. We handpick the highlights of the data charts from a variety of sources, analyzing and delivering insights trusted by 100+ top institutional and corporate clients worldwide at BigOne Lab.

Understanding China’s true job market is complex. On the one hand, official data suggests the overall unemployment rate remains steady and healthy. On the other hand, if you live in China, you hear constant conversations about how difficult the job market has become and how salary growth has stalled.

This is a crucial question: gauging the health of Chinese households is essential to determining whether China can successfully transition into a consumption-driven economy, which depends on healthy, organic income growth.

So, what is really going on?

China’s economic structure has been undergoing a quiet transition from “old” to “new” over the past few years. Much of this shift was muted by pandemic news. While headlines consistently mention “industrial upgrading” (产业升级) or “New Quality Productive Forces” (新质生产力), it can be difficult to decipher whether this is actually happening on the ground. Today, we dive into online job postings data to find the answer.

Methodology Note: To map this transition, I classified millions of job postings from major recruitment platforms into six categories: New Productivity & Industrial Upgrade, Real Estate & Related, “Old Tech” & Traditional Manufacturing, Finance, Education, and Others.

The “New Productivity” category tracks the strategic pivot toward sectors like EVs, semiconductors, AI, green energy, biotechnology, etc. “Old Tech” includes industries that grew rapidly over the past decade—such as e-commerce, software, and internet platforms—which have increasingly matured into traditional industries rather than representing the current “cutting edge.” Other categories are self-explanatory.

While some cross-functional roles defy perfect classification, the massive scale of data since 2020 provides a definitive view of the structural transformation occurring over the last five years. (See the footnote for full definitions.)

Subscribe now

Q1: From “old” real estate to “new” productivity: has the transformation happened?

News headlines are swamped with news of China’s struggling real estate market (rightfully so, as listing prices for existing homes have yet to find a bottom). However, a quiet transformation has taken place within the job market. Since 2020:

• The real estate and related industries (the orange area) have seen their share of total recruitment cut in half, dropping from 15% in 2020 to less than 8% by the end of 2025.

• Simultaneously, the “New Productivity” sectors (the blue area) have shown powerful counter-cyclical growth, with their share of job postings climbing from 10% in 2020 to over 15.4% in 2025.

It is no coincidence that the added value of China’s real estate industry fell to only ~6.3% of GDP in 2024 [*]. (In my opinion, this partly explains why Beijing feels less urgency to launch a massive stimulus; real estate no longer influences overall employment as heavily as it did just five years ago.)

Despite the “structural pain” society is currently enduring, this shift proves the underlying logic of the Chinese economy has fundamentally changed. The absolute number of jobs created by the ‘New Productivity’ sectors (an increase of about 300,000 positions) has effectively offset the bulk of the losses in the real estate sector (a loss of about 350,000 positions).

The transformation of the “New” replacing the “Old” real estate is now firmly established in the trend data.

Q2: Why do on-the-ground sentiments feel so cold?

If the overall number of jobs has remained relatively stable and “Hard Tech” is booming, why is public sentiment so gloomy?

The answer lies in structural friction and a perceived “quality downgrade” for the middle class. Real Estate, Finance, and Education—industries that traditionally recruited a significant and highly educated portion of the population—have seen a consistent shrinkage in job postings since 2022, never recovering to pre-pandemic levels. These happen to be the three high-paying industries most favored by the university-educated, urban middle class (the demographic with the strongest voice on social media). These traditional sectors also still serve as a massive reservoir for the labor force, maintaining a share of roughly 32%.

For those who managed to keep their jobs in these industries, salary growth has been optimistic for them.

The new jobs replacing the old ones offer no easy transition. You cannot simply retrain a real estate marketing manager to design semiconductors or engineer hydrogen fuel cells. The booming “New Quality Productive Forces” require highly specific STEM backgrounds (often Master’s or PhDs). Meanwhile, the other side of the job growth spectrum—such as wholesale, retail, and local services—often represents a downgrade in social prestige and stability for university graduates.

This “talent incompatibility” is what creates the profound sense of a “job crisis” on the ground.

Q3: Are there other interesting insights and key takeaways for investors?

If you want to find the next structural growth opportunities in China, follow the hiring.

Trend 1: Consumer internet could soon become a “utility”

The era of hyper-growth for tech platforms and e-commerce—sectors that once defined the ‘cutting edge,’ provided the most sought-after roles for China’s top-tier talent, and made investors filthy rich—is over. When we examine the absolute job gains and losses across sub-industries from early 2020 to early 2026, the data paints a vivid picture of an economy executing a ruthless, structural pivot.

Subscribe now

Read more

First, Alibaba’s Qwen triggered a nationwide frenzy by offering “tea freedom” through a massive milk tea subsidy, sending its Qwen AI app to the top of the charts and creating messes at some milk tea shops as delivery people scrambled for orders.

Not to be outdone, Yuanbao, Tencent’s own AI chatbot, launched a billion-yuan red packet campaign, hoping to recreate the success of WeChat Pay. However, the plot twisted when WeChat, the centerpiece of the Tencent family of products, abruptly restricted the sharing of these promotional links to curb spam and “induced sharing.”

ByteDance’s Seedance 2.0, on the other hand, stunned the whole world by its mesmerizing audio-video co-generation technology, strong character consistency, and director-level scene transition, effectively giving every social media user their own Hollywood-quality production studio.

And Spring Festival has not even started!

At this point, I want to put down some preliminary observations here, if only to make a mental mark for the record amid these rapid, crazy changes that are unfolding at this very moment.

Observation 1: The market rewards the cutting edge, but penalizes the old playbook

If it were a decade ago, Tencent’s Yuanbao red packet campaign would have excited the market. After all, the whole red packet playbook was exactly what ushered WeChat Pay into success a decade ago. However, this time, the campaign has caused much more doubt than excitement. As a result, Tencent’s share price has already dropped more than 15% year-to-date at the time of writing this piece.

Nor is the market very excited about BABA’s milk tea giveaways either. Its stock price has traded sideways despite its Qwen app almost becoming the most actively used AI app in China within an extremely short period of time.

Contrast that with Zhipu AI, believed to be behind Pony Alpha, an open-source coding model that many developers compare to Claude Code in capability. Zhipu also announced its GLM-5 model just now, which sent its shares almost quadrupling only a month after its IPO.

The divergence is striking. In this decade, the old playbook no longer inspires. The market is clearly looking for genuine frontier capability, not user growth engineered by subsidies. Merely having more users, if those users are acquired through incentives rather than through technological excitement, is heavily discounted.

Observation 2: The real market leader is not tradable

It’s also a pity that the company that is arguably leading China’s AI race is not listed.

ByteDance, which has Doubao, currently has the most popular AI chatbot in China and arguably the highest user stickiness. Many users already treat Doubao as a daily companion, almost in the sense portrayed in the film Her. While Qwen may be close in daily active users, much of that still appears subsidy-driven. Doubao’s user base feels more organic and durable, at least at this stage.

At the same time, ByteDance’s Seedance 2.0 has become the global leader in video generation since its recent debut. ByteDance’s trial with AI-native smartphone also inspires imagination. (You can check out a product test of Doubao phone here made by , our partner newsletter.)

Yet ByteDance is not listed. Neither is Huawei, another important AI force.

As a result, even if one believes strongly in the long-term future of Chinese AI, the Hang Seng Tech Index may unfortunately not be the most direct way to express that view at this moment. Core constituents such as Alibaba and Tencent have been weak, and the index has traded sideways.

Observation 3: The great divergence of how AI would impact other industries

As much as there is a domestic divergence in how old and new forces tackle AI, internationally, a significant divergence is emerging in how markets interpret AI’s impact on other industries.

In the US, when Anthropic or Google launched a new model, software and business service stocks sold off sharply. Hundreds of billions in market value could be erased in days. The dominant narrative is that the old industries are at risk of decimation.

In China, sentiment looks different. When major breakthroughs in video generation like SeeDance appear, online literature, film-making, and content platform stocks rally. Even if China’s software sector has not been exceptionally strong, it has not collapsed in the same way as in the West either. And milk tea brands like Guming and Mixue are emerging as the biggest net beneficiaries of Alibaba’s generous subsidies.

Why?

One reason is industry maturity. China’s SaaS industry never truly developed into a large, standardized, high-margin sector comparable to the US. Much of the market remains project-based and customized. Ironically, that lack of maturity could prove resilient in an AI era. Companies built around implementation and consulting may still be needed, while benefiting from lower AI-driven costs.

As for the entertainment industry, China does not yet have a Hollywood-sized creative market to disrupt. When AI improves content creation, investors focus on new possibilities rather than on the elimination of entrenched incumbents.

There may also be a deeper cultural and political layer. In the current Chinese context, it is difficult to imagine technology wiping out an entire industry overnight without at least paying some price. The concept of common prosperity has shaped expectations around these things.

I do not claim to have a definitive explanation here. But this divergence itself is worth watching.

Observation 4: Trust first, commerce later?

The next major milestone in AI, after coding, may be agentic commerce: how AI changes how people transact and live.

Alibaba is positioning itself as the leading global experimenter in this space. As the operator of one of the largest e-commerce ecosystems in the world, it is pushing consumers to shop in an AI-native way earlier than anyone else.

The results so far are mixed, however.

When I tried shopping through Qwen’s agent, the experience proved that AI-mediated shopping is technically feasible. But beyond that proof of concept, it did not feel revolutionary. Which leads me to think that chatting back and forth with a bot to find something to buy may not be the final form of commerce.

By classic internet logic, this is not a problem. “Move fast and break things” has been the motto for the industry since its very beginning. Early user experience can be rough as long as feedback loops are strong and iteration speed is fast.

But it may be too premature to push agentic commerce too aggressively, which leads to a more fundamental question: what do people actually look for when they shop?

For purely utilitarian purchases, such as recurring household goods, automated agents make sense. But much of consumption is not utilitarian. For many high-value purchases, the experience itself is part of the product: walking through a store, browsing shelves, and discovering items serendipitously. I don’t believe these are inefficiencies to be eliminated. For many, shopping is not just a means, but an end in itself.

A better model, intuitively and logically, would be to build trust first.

If users see their AI agents as friends, advisors, or companions, they will naturally ask them for shopping recommendations. Commerce then becomes an extension of trust, not vice versa.

If this theory is true, then agentic commerce does become mainstream, ByteDance, again, may still capture the largest upside. Doubao already has user trust and stickiness. Combined with Douyin’s e-commerce ecosystem, all the essential pieces are in place.

Tencent may also be underestimated. Even if it currently lags in model leadership, it possesses vast behavioral data to eventually build a trustworthy AI virtual assistant as well as a deeply integrated WeChat mini-program commerce ecosystem within to build a complete e-commerce experience.

Alibaba has limited social advantages and does not control the most popular AI companion app right now. It is therefore logical that it plays to its core strength: commerce. Whether that will work is questionable, but it just does not have many alternative strategic choices.

Observation 5: 2026 as the year of mass adoption

The upcoming Spring Festival holiday is shaping up to be another holiday about AI. Chinese AI companies are clearly timing product releases around this period, when users are at home, traveling, and experimenting with new apps.

It is not yet clear who will ultimately win. What is increasingly clear, however, is that 2026 is shaping up to be a year of mass AI adoption in China.

This adoption will go far beyond chatbot novelty. By the end of 2026, it is highly likely that a majority of urban consumers will have completed at least one transaction through an AI agent. Coders will routinely use domestic Chinese coding agents. Video creators will rely on AI-native tools as a standard workflow.

This unprecedented mass experimentation and mass adoption of AI applications, with all of the potential impact and consequences, will provide meaningful lessons for the whole world to see.

This should be our last post before the Spring Festival commences.

Finally, to all our readers, thank you for staying with us through another turbulent year in markets. May the Year of the Horse bring you momentum, milestones, and more exciting breakthroughs in your work and in your life!

Subscribe now

Since 2023, we have been frequently pointing out a clear trend: Chinese companies’ hiring for overseas positions has been outperforming the domestic job market, as firms actively expand abroad. “Going abroad” has been a prevailing theme among Chinese enterprises even before Trade War 2.0, and overseas hiring has only continued to gain momentum even after Trump’s second term commenced (as did China’s exports).

Now, the natural question for Baiguan readers and China investors is: what do we do with this trend?
In today’s Charts of the Week issue, let’s dig into that question with some interesting data insights.

Subscribe now

Chart of the Week: Growth in China’s overseas job recruitment slowed significantly in North America in 2025, but remained steady in almost every other region.

In 2025, exports remained one of the brightest spots in China’s economy, defying the headwinds from Trade War 2.0. Amid heightened volatility in US–China relations, hiring demand for jobs in North America cooled notably after the explosive growth seen in the prior year (though it still continued to grow).

Demand in other key regions, however, remained steady. Southeast Asia and Africa continued to be the largest overseas job destinations for Chinese enterprises, with no signs of slowdown. Meanwhile, demand in the Middle East — while still not a majority destination — has seen explosive growth over the past two years.

This suggests that China’s “going abroad” blueprint has become far more diversified over time, giving Chinese companies greater resilience — and more cards in hand — amid trade tensions.

When we break down overseas job recruitment by industry, we also see an interesting pattern of rotation and diversification.

In 2024, construction, business services, and information technology led the “going abroad” initiatives. (Internet and e-commerce giants such as Pinduoduo (Temu) were categorized under IT, which significantly drove the overseas hiring demand. Remember when we covered how China was exporting both merchandise and manufacturing capabilities back in 2023? Chinese companies going global: merchandise and manufacturing exports in the new era.)

In 2025, however, we are seeing a more mixed portfolio of industries ramping up overseas expansion — not just in physical products, but also in services such as financial services, research, and design. Notably, business services jobs saw explosive growth, largely driven by companies that support overseas expansion, including digital advertising, consulting, legal, and human capital services.

One interesting example: we observed that Michael Page’s China arm — a recruitment agency focused on senior and executive talent — experienced very rapid hiring itself. This suggests that Chinese companies are increasingly “settling down” abroad and investing in professional expertise and management talent. If anything, they are expanding their overseas footprint even further after Trade War 2.0 commenced.

Some companies in the spotlight

Looking ahead, 2026 is shaping up to maintain strong export momentum. One key takeaway here is that investors may want to pay close attention to companies with meaningful overseas exposure — or those actively ramping up efforts to expand international market share — as they have a higher likelihood of delivering positive earnings surprises.

This is especially relevant for industries weighed down by persistent pessimism around domestic “involution” and price wars, such as upstream raw materials, EVs, consumer electronics, and food delivery platforms. While the domestic market is unlikely to see a demand boom overnight, companies with strong overseas exposure could see overseas revenue surprises that are not yet fully priced in.

Let’s look at a few examples — companies that were hiring aggressively for overseas roles (or domestic roles targeting overseas markets) in 2H25. These are the firms positioning themselves for a much larger international footprint in 2026.

XPeng (XPEV) | Smart EVs & AI mobility

Overseas job hiring trend: +78.4% YoY

By the end of 2025, XPeng expanded its global footprint to 60 countries, with its overseas deliveries soaring 96% YoY to 45,008 units [*]. Entering 2026, XPeng will establish independent supply chain teams in Europe and Southeast Asia starting in 2026 to support localized operations [*]. One key upside is that Chinese EV makers can generate significantly higher margins in overseas markets. For instance, BYD reportedly generates three to four times more profit per vehicle overseas compared to its fiercely competitive domestic Chinese market as of late 2025. Overseas growth could therefore serve as a catalyst to lift pessimism surrounding China’s saturated EV market and shrinking margins amid ongoing price wars. (XPeng also comes with attractive autonomous driving optionality this year, which is shaping up to be an important investment theme to watch.)

Other EV makers have also ramped up overseas hiring, including Li Auto (+385%), Changan 000625.SHE (+385%), and Dongfeng 0489.HK (+180%). BYD’s overseas hiring did not expand further in 2025, largely because its hiring already surged in 2023, as BYD began building its overseas footprint well ahead of peers.

Yili Group (600887.SH) | Dairy products

Overseas job hiring trend: +408.3% YoY

This might come as a surprise to many. It’s hard to imagine a traditional dairy maker—dealing with fresh, perishable goods—successfully selling into overseas markets. Yet, Yili is doing exactly that by building actual factories on the ground. Their Indonesian production base is now the largest ever built by a Chinese dairy firm in Southeast Asia [*]. Instead of just exporting Chinese milk, they are localizing the brand; for instance, their Joyday ice cream is now the fastest-growing brand in Indonesia and has already reached over 10 countries across the Middle East and Africa. They are essentially exporting their manufacturing playbook and supply chain efficiency to high-growth emerging markets where dairy demand is still booming. I think this could bring a nice catalyst, especially given that the narrative is predominantly dominated by concerns over deflation and low birth rates in China.

Similar companies are taking the same strategy: Mengniu Dairy 2319.HK (280% YoY) and Anjoy Foods Group 603345.SHA (550% YoY) are both ramping up the hiring for the overseas market in 2025H2. Pleasant surprises may emerge from these “boring” businesses in 2026.

Continue for more examples and data points. To get a sense of what is offered, you are welcome to check out this older post in the same series: Charts of the Week. You can also get free access by sharing us. Paid subscribers enjoy a variety of benefits. Check out the benefits: exclusive benefits for Baiguan’s paid subscribers.

Subscribe now

Read more

Following a series of bearish reports, short interest in the stock has remained stubbornly high. Pop Mart’s price has since corrected more than 35% from its peak.

The core narrative driving this sell-off is straightforward and, admittedly, compelling. Many argue that Pop Mart’s explosive 2025 run was a temporary ‘supercycle’ driven by a single IP anomaly—Labubu—and a viral moment in the West that is now effectively over.

For instance, short sellers argue that Labubu is a ‘one-hit wonder’ that has already faded from its peak with no successor in sight. They cite plummeting search interest, cooling social media buzz, and falling resale prices on second-hand platforms as proof of demand destruction.

Share

While our own social media data confirms this trend—and I do acknowledge that the Labubu craze has decelerated—I remain constructive on the current setup for the company following this sharp correction.

In a nutshell, I believe the operational reality at Pop Mart points to normalization, not a total collapse. There is a massive difference between a company going to zero and a company returning to ‘normal’ (and still rapid) growth.

Behind the paywall, I will explain why Pop Mart doesn’t look particularly expensive at these levels, backed by BigOne Lab‘s recent sales and social media data, as well as some on-the-ground observations of new monetization mechanisms the company has recently rolled out.

Disclaimer: This newsletter is for informational purposes only and does not constitute investment advice. Please do your own research or consult your financial advisor.

Subscribe now

Read more