China may still be chasing the US at the frontier model layer, but it is moving faster where ordinary users actually feel AI: inside apps, services and small businesses.
The AI race is often described as a contest over who has the biggest model, the most expensive chips or the most impressive benchmark score. China is making a different argument. Its advantage may be less about building the single smartest system and more about getting useful AI into the hands of millions of people who are trying to sell, buy, code, trade, design, translate and run daily work faster.
That was the point made by Chi Zhang, general manager of finance industry at Alibaba Cloud Intelligence Group, during a panel discussion on Thursday at the 2026 HKEX Future Tech Summit in Shenzhen. Zhang said China is catching up with the US in frontier AI model capabilities at a tremendous pace, but its bigger opportunity sits in applications, helped by a deep pool of entrepreneurs and engineers and an economy still hungry for productivity gains.
This matters because most users do not experience AI as a benchmark. They experience it as a button inside a shopping app, a workflow inside an office tool, a customer service assistant, a travel planner, a coding helper or an agent that can handle routine tasks without needing a specialist to sit beside it. On that measure, China has become a serious test case for how AI leaves the lab and enters ordinary commerce.
The strongest signal is coming from Chinese platforms that already sit close to everyday behavior. Alibaba has been connecting Qwen more deeply into its consumer and business ecosystem, starting with food delivery and expanding toward services such as Taobao and Fliggy. Baidu, Tencent and other large platforms are also pushing AI into search, messaging, cloud tools and enterprise software.
DeepSeek changed the conversation last year by showing that a Chinese model could win global attention without matching the spending patterns of Silicon Valley. Its lower-cost approach pushed US investors and founders to ask whether the AI race was only about capital intensity, or whether efficiency and distribution could matter just as much. That question has not gone away.
According to a report from Business Insider, Alibaba.com president Kuo Zhang said 30% to 40% of the platform’s customers are now solo entrepreneurs, with AI agents helping them handle tasks such as product listings, customer service and tax compliance. That is not a small detail. It shows how quickly AI becomes valuable when it is attached to work people are already doing.
In the US, many of the best AI tools still feel like separate destinations. Users open ChatGPT, Claude, Gemini or a coding assistant, then bring the output back into their workflow. That is powerful, but it creates friction. China’s large platforms are pushing harder toward AI as an embedded function, where the user may not even think of it as a separate product.
The US still owns much of the frontier
None of this means China has overtaken the US across the whole AI stack. American companies still lead many frontier model discussions, especially around advanced reasoning, enterprise deployment and high-end infrastructure. OpenAI, Anthropic, Google and Meta continue to set much of the global pace in model development, product packaging and capital formation.
But frontier leadership is not the same thing as adoption leadership. A model that is slightly better on hard tests may matter less to a merchant, factory owner or travel agent than a cheaper tool that is already available inside the software they use every day. This is where China’s market structure helps. Dense mobile ecosystems, fierce app competition and thin margins push companies to apply AI quickly rather than present it as a premium layer for later.
There is also a policy angle. Beijing wants AI to spread through the economy, especially as growth becomes harder to generate from property, exports and old infrastructure spending. For entrepreneurs, that creates a practical incentive. If AI can reduce headcount needs, speed up operations or make small firms look larger than they are, adoption becomes less about curiosity and more about survival.
The risks are real. Chinese AI companies face US chip controls, global distrust over data governance and growing scrutiny from Washington. The Guardian reported this week that the Pentagon added major Chinese firms including Alibaba and Baidu to a list of companies it says are linked to China’s military, a designation that does not impose direct sanctions but deepens the political pressure around Chinese technology abroad.
That pressure may limit how far Chinese AI apps can travel in sensitive markets. It may also make US companies and governments more cautious about relying on Chinese models, even when they are cheap and capable. For China, the domestic market is large enough to keep adoption moving, but international trust will be harder to win than local usage.
The more interesting question is what the US learns from this. Silicon Valley has built extraordinary AI systems, but the next phase will reward companies that make those systems feel boringly useful. The winning products will not be the ones that merely impress users in a demo. They will be the ones that sit inside daily work and quietly remove effort.
China’s edge in everyday AI is not guaranteed to last. But for now, it is showing that adoption can be its own form of power. The market should watch less for the next leaderboard result and more for where AI becomes an invisible part of normal economic life.
Also read: Xpeng puts its founder in charge of the humanoid robot race • ZincFive is taking its nickel-zinc data center batteries public through a $600 million SPAC deal that signals a quieter revival in blank-check financing • Plaud is putting Singapore at the center of its Asia expansion
