The United States surpasses China by 23 times in private AI investments, but Kimi K3 questions the effectiveness of bans.
The gap in artificial intelligence funding volumes between the US and China continues to widen, with the latest data from the Stanford AI Index 2026 report clearly confirming this. In 2025, private investments by American companies in the AI sector reached a staggering $285.9 billion, while China's figure stood at only $12.4 billion. Thus, the US surpasses China by more than 23 times.
However, despite such an impressive financial advantage, it is the Chinese model Kimi K3, developed by Moonshot AI, that has sparked another wave of panic in Washington. This model, distributed with open weights, is available for download and deployment on any server. Its sudden success in programming tests led to a collapse in the stock prices of American chip manufacturers, instantly reviving long-standing debates about the need for strict restrictions.
Why Kimi K3 is reigniting debates about bans
Last year, the US Department of Commerce already considered adding Chinese labs to the Entity List — the same blacklist that Huawei was placed on in 2019. At that time, supporting innovation was named a priority, and radical measures were abandoned. But the success of Kimi K3 has given hardliners new arguments. White House AI advisor David Sacks directly stated that this is a key moment for regulation, and that closed market leaders want authorities to eliminate their open-source competitors.
The hype around Kimi K3 is growing rapidly. Moonshot AI suspended accepting new subscriptions just 48 hours after launch, and is now preparing for an IPO in Hong Kong.
What lies behind the funding gap
The numbers look impressive, but the report's authors acknowledge: official figures do not account for direct government injections from Beijing. From 2000 to 2023, Chinese government funds directed approximately $184 billion into the local AI sector. Additionally, financial benefits often outweigh patriotic considerations: DeepSeek V4 Pro charges $0.87 per 1 million output tokens, while a similar volume from Anthropic's Claude Fable 5 costs $50. Coinbase CEO Brian Armstrong reported that switching the exchange to GLM 5.2 and Kimi K2.7 Code cut corporate AI expenses in half.
In Washington, they have encountered this before. In January 2025, the release of DeepSeek cost Nvidia a record $589 billion in market capitalization in a single day. It was after this collapse that the first discussions of a potential ban emerged.
The problem is that a ban might not work at all. The weight coefficients of Kimi K3 have already been placed in open repositories, and completely removing them is virtually impossible. Restrictions will only raise costs, and the launch of Alibaba's Qwen3.8-Max shows that new competitors will appear regardless of bans. While financial indicators favor the US, the decisive factor is becoming not the volume of investment, but the actual effectiveness of the technologies being created.
Expert commentary: The story of Kimi K3 clearly demonstrates that in the AI race, money is just one factor. China, using an open-source model and aggressive pricing, creates technologies that businesses adopt voluntarily, not by decree. The US faces a difficult choice: either try to block the inevitable, or adapt to a new reality where efficiency matters more than budget.