China doesn’t need to build better AI than the US to win the tech race

Cheaper Chinese open-weight models may just need to prove themselves good enough for businesses to adopt

Vivien Shiao

Published

UP ONE day, down the other. That pretty much sums up the US stock market – and my corresponding blood pressure – these days as investors try to work out whether the artificial intelligence boom is a bubble waiting to burst.

One development that made waves was the unveiling of Chinese AI startup Moonshot’s Kimi K3 open-weight model on Jul 17, which reportedly rivals frontier models from Anthropic and OpenAI at a fraction of the cost.

Much of the AI race has been framed as a contest over who can build the smartest model. But that may no longer be the right question to ask.

Instead, it now becomes who can build AI that is good enough, cheap enough and open enough to become the foundation businesses choose to build on.

China’s DeepSeek breakthrough earlier last year was initially seen as a one-off moment. It now looks like the start of a broader shift.

Findings from the UK’s AI Security Institute, released a day after Kimi K3’s release, showed that leading Chinese open-weight models now trail US frontier models by just four to seven months, down from six to 10 months in 2025.

That matters because businesses rarely choose technology on benchmark scores alone. They weigh capability against cost, security, reliability, compliance and how easily a model fits into existing systems.

As the performance gap narrows, these other considerations start to matter more.

Unlike proprietary models that are accessed through paid cloud services, China’s open-weight models have parameters that are publicly available.

This allows businesses to download, customise and run AI on their own infrastructure, potentially lowering costs drastically and giving more control over sensitive data.

So why are Chinese AI companies giving away technology that could otherwise command premium prices?

Earlier in July, President Xi Jinping called for more equitable global access to AI at the World AI Conference in Shanghai, urging countries to seize the “historic opportunity” presented by open-source AI.

The message casts China as a champion of accessible AI. But there is logic behind this. By making high-quality models widely available, the Asian country challenges the business models of US AI leaders while encouraging businesses and developers around the world to build on Chinese technology.

Increased adoption strengthens China’s technological influence, regardless of whether the model itself generates significant revenue.

Scott Galloway, a professor at New York University and serial entrepreneur, has described this as AI “dumping”, likening it to China’s expansion in industries such as steel, solar panels and electric vehicles through low-cost competition.

Software is obviously different from steel, and AI models are not manufactured goods. But his broader point stands.

If businesses increasingly swap expensive proprietary US models for cheaper Chinese alternatives that are “good enough”, investors may question whether lofty valuations of today’s AI leaders remain justified.

That could also increase pressure on Washington to restrict Chinese AI models. Yet technology companies including Nvidia, Microsoft and Meta have urged policymakers to avoid “premature restrictions” on open-weight AI, arguing that such models strengthen competition and ensure the benefits of AI are broadly shared.

Open-weight advocates have long maintained that wider access improves software quality and safety. Greater adoption also expands the ecosystem, driving demand for computing infrastructure while reducing the risk of a few proprietary models dominating the market.

What already seems clear, however, is that China is changing the economics of the AI race.

China may not need to build better AI than the US to emerge as a winner. It may simply need to build AI that the rest of the world decides is good enough to use.

source: The Business Times https://www.businesstimes.com.sg/opinion-features/china-doesnt-need-build-better-ai-us-win-tech-race?ref=global-opinion