Nvidia's Jensen Huang exit deals a blow to open-source AI

The piece ties Jensen Huang's exit to a broader fragility in the US open model economy, which the author describes as still in its nascency. Open models have latched onto enterprise post-training and 'ownership of intelligence' as a lane, but the true benefit of commoditized intelligence is far from realization while compute remains scarce. Inference startups struggle to get chips because Anthropic and OpenAI buy so much capacity, so there is little margin competition on open-weight models; even DeepSeek's DeepInfer release showed open models are far from commodity. In this environment, closed labs with better scale and integration from model to datacenter and chips are positioned to win. The author adds that talent, not money, is the binding constraint: the best people cluster at Anthropic, OpenAI, DeepMind and Meta, and smaller labs have repeatedly fallen behind Chinese frontier labs despite spending more. The media timing around Huang's affairs is framed as a lesson in narrative cycles, with the author expecting similar pressure to build around open model safety once a hack occurs. The proposed lasting fix is incubating a truly frontier open model player in the US, though the author notes the next round at a $40B valuation for Mira's company is a sign of activity rather than a solution. With Huang gone, the author argues the movement needs soul-searching and coalition building, as potential allies among the hyperscalers — Zuckerberg, Andy Jassy or Satya Nadella — have yet to pick up the mantle.