NVIDIA has been on a roll lately. It recently acquired Hugging Face, a well-known hub for open-source models, and is now set to invest between $2.5 billion and $3 billion in the latest funding round for Thinking Machines Lab. This lab was established by Mira Murati, the former CTO of OpenAI. Despite being founded just 19 months ago, Thinking Machines Lab boasts a pre-investment valuation of up to $40 billion, even though its annualized revenue barely surpasses $100 million. Murati launched this venture after departing from OpenAI, and the majority of the team members are also ex-OpenAI employees. The company has sequentially rolled out the fine-tuning platform Tinker and its self-developed large model Inkling. Moreover, it has forged a strategic partnership with NVIDIA to deploy the 1GW Vera Rubin system. By investing in and partnering with cutting-edge technology firms, and complementing this with the acquisition of Hugging Face, NVIDIA has constructed a closed ecological loop within the open-weight domain. This loop spans from entry points, models, and fine-tuning platforms all the way to chips, marking NVIDIA's transition from an AI infrastructure provider to a company that pre-secures computing power demands at the model layer.