Will saving 4 billion yuan still trigger regulation?

Save 4 billion yuan or trigger regulation? Microsoft's plan to introduce domestic AI models to Copilot is in a dilemma. In order to reduce high AI operating costs, Microsoft is currently evaluating the introduction of the Kimi K3 model developed by Chinese artificial intelligence company Dark Side of the Moon into its Copilot service. Previously, the service mainly relied on the technologies of OpenAI and Anthropic, and the addition of new models is expected to reshape its underlying architecture. With outstanding performance and cost reduction advantages, it can save up to 600 million US dollars per year. As the open source model with the largest parameter scale in the world, Kimi K3 has demonstrated excellent performance comparable to top models in many authoritative evaluations such as programming. Microsoft's internal calculations show that switching to this model can reduce the inference cost of a single token by up to 60%, saving the company approximately US$602 million per year. Huge profits collide with political risks, and policy red lines become the biggest concern. However, this highly commercially attractive decision is facing severe external political considerations and policy risks. The U.S. government is currently planning to further restrict local companies from using Chinese AI models, and may even introduce targeted bans. If Microsoft deeply integrates Chinese open source technology into its flagship products, it is very likely that Microsoft will attract strong dissatisfaction and scrutiny from the White House. Faced with huge economic interests and complex political situations, Microsoft is still in a stage of careful weighing and making decisions. via AI News (author: AI Base)