Alibaba and Tencent pivot to domestic chips amid U.S. export curbs
Chinese tech giants are relying on internal chip development to bypass U.S. restrictions, though they warn that training new AI models will face delays.
The short version
- Tencent and Alibaba say domestic chips can support current AI inference but will delay training new models.
- New U.S. rules require licenses for exporting advanced chips to several Asian and South American countries.
- Alibaba's T-Head chip division is expected to ramp up production by 2027 or 2028.
- Tencent reported $31 billion in revenue for Q3 2026, while Alibaba reported $40.5 billion.
- Shareholders have sued Alibaba over alleged ties to a company suspected of funneling chips to China.
Why it matters
This shift signals a move toward a bifurcated tech landscape where Chinese firms must build independent hardware stacks. For users, this may mean slower updates to Chinese AI models but more resilient domestic cloud services.
Navigating new export rules
The U.S. Department of Commerce recently released an interim rule that expands export controls on advanced AI processors. These measures now require licenses for chipmakers to sell advanced hardware to data centers in countries like Malaysia, Singapore, Thailand, Indonesia, and Brazil. The rule also targets large data centers that offer significant computing power to Chinese customers.
These restrictions are designed to limit the amount of high-end computing power available to Chinese tech giants. Companies like Alibaba and Tencent are among the first to provide corporate responses to these specific measures. The new rules also narrow the exemptions that manufacturers can use to ship advanced chips without obtaining a license.
The impact of these controls extends beyond just the sale of finished products. They also limit how Chinese firms can produce their own semiconductors using facilities located in other countries. As a result, major manufacturers like Samsung Electronics and TSMC have reportedly suspended some chip shipments to Chinese clients.
Tencent's inference strategy
Tencent executives highlighted a distinction between the power needed to train new models and the power required to run existing ones. They noted that their current cloud business is primarily focused on inference, which is the process of running an AI model rather than building it. Because of this, the company believes it is not seeing a major impact on its current operations.
Tencent's chief strategy officer, James Mitchell, stated that the company has reached a point where inference can run on a "surprisingly limited number" of domestically produced chips. This suggests that the current domestic supply chain may be sufficient to handle smaller and more efficient AI models. However, the company acknowledges that these domestic chips are not yet a complete replacement for high-end foreign hardware.
Despite the current stability, Tencent's president, Martin Lau, admitted that the restrictions are "unfortunate" for long-term growth. He noted that it will now take more time for the company to train the next version of its Hunyuan model. Lau predicted that these U.S. controls will eventually speed up the development of China's own domestic chip industry by forcing all players to rely on local supply chains.
Alibaba's hardware roadmap
Alibaba reported that its progress in developing domestic computing power is already ahead of its original expectations. The company has invested heavily in its in-house chip division, known as T-Head, and plans to list this division in Hong Kong. CEO Eddie Wu stated that the current T-Head hardware is already ready to handle the training of large-scale and ultra-large-scale models.
While the company remains confident in its 2030 targets, it warned that the new export rules will affect products and services starting next year. Wu suggested that cloud revenue growth could slow to a range of 30% to 45% before potentially rebounding to 50% or higher by 2028. He noted that there is still significant uncertainty regarding the supply chain for these advanced components.
The company expects T-Head production to ramp up significantly in 2027 or 2028. Once that production scale is reached, Wu believes the company will be in a better position to control the pace at which it iterates its foundation models. This move toward self-sufficiency is intended to provide more autonomy over their AI development cycle.
Legal and market pressures
Alibaba is currently facing scrutiny following a scandal involving Nvidia and a Singapore-based cloud computing company called Megaspeed. Megaspeed is suspected of moving high-end AI chips into China in a way that breaks U.S. export rules, though the company denies these allegations. This situation has put Alibaba in the spotlight regarding its historical ties to certain executives and entities.
A group of Alibaba shareholders recently filed a lawsuit against the company over these concerns. The lawsuit accuses the company of concealing its ties to a former Megaspeed executive and understating the risks of export-control violations. The shareholders claimed that media reports of these ties contributed to a 14% drop in Alibaba's share price over a four-day period.
The broader market is also feeling the pressure as other Chinese developers scramble to adapt. For example, Moonshot AI reportedly scrapped a $150 million deal to access Nvidia's Blackwell chips through an Indonesian data center. These events underscore the difficulty Chinese firms face in securing the high-end hardware necessary for the next generation of AI.
The state of the market
The U.S. Commerce Department is facing increased scrutiny over the effectiveness of its chip controls. Data released by the House Foreign Affairs Committee showed that the department had granted permission for nearly 880,000 AI chips to be sent to Chinese companies. This includes 75,000 chips for Alibaba and 150,000 for Tencent out of more than 3.4 million chips that Nvidia and AMD sought to export.
The financial results for both companies show significant growth despite these hurdles. Tencent reported 207.3 billion yuan in revenue for the third quarter of 2026, with its fintech and business services revenue rising 9.4%. Alibaba reported 271.5 billion yuan in revenue for the same period, with its AI and cloud revenue jumping 53% to reach 61.0 billion yuan.
Both companies are significantly increasing their capital expenditures to build out their infrastructure. Tencent's capital expenditure rose 527% to 81.4 billion yuan, while Alibaba's rose 60% to 50.5 billion yuan. These investments reflect a massive push to build out the domestic hardware capacity needed to sustain their AI ambitions in a restricted environment.
What to watch
- Whether Alibaba's T-Head production hits its 2027 targets
- The outcome of the shareholder lawsuit against Alibaba
- How other Chinese AI startups adapt to the lack of Nvidia hardware
Sources and links
- OfficialTencent reported static.www.tencent.com
- ReportingFortune: the article we sourced
First reported by Fortune. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.