Your signal. Your price.
The Commerce Department is reportedly investigating Moonshot AI and other Chinese labs for circumventing export controls to access NVIDIA GPUs.
Investors previously worried about circular financing, such as NVIDIA's $30 billion investment in OpenAI, fearing it artificially boosted chip revenue.
Wayve recently secured investments from major chip companies (Nvidia, Qualcomm, AMD) and automotive/tech giants (Uber, Nissan, Mercedes, Stellantis, Microsoft), signaling industry validation for their once-contrarian end-to-end AI approach.
Google signed a three-year deal with SpaceX for compute access, agreeing to pay $920 million monthly for at least 110,000 Nvidia GPUs from October through June 2029.
Jensen Huang secured Nvidia's memory supply with a multi-year deal with SK Hynix, designating them as Nvidia's largest memory supplier and a design partner for next-generation physical AI and personal AI chips.
John suggests Kimi K3's emergence challenges the notion that US frontier labs have an insurmountable lead due to China's restricted access to NVIDIA chips, raising questions about alternative training methods or shadow GPU clusters.
OpenAI is in advanced negotiations to lease a 10-gigawatt data center campus in Ohio, projected to cost $500 billion. Nvidia is a financial backer, with 800 megawatts expected online by 2028 under a potential 20-year lease.
Matt Odell suggests the AI investment space is bubbly, noting that 25% of Nvidia's year-to-date revenue comes from a single client in Singapore, likely China attempting to bypass export restrictions.
Alex notes Inkling is stronger than NVIDIA's Nemotron but weaker than top Chinese open-weight models and closed-weight Western models, suggesting the West has had poor incentives for strong open-weight releases.