Nathan Lambert warns that Jensen Huang’s reported exit from Nvidia could leave the US open-model ecosystem short of two things: money and political muscle. His Interconnects essay argues that Huang supplied both, and that Nvidia’s commercial incentives don’t automatically replace his influence. Source
The South China Morning Post reported on November 13 that Huang had resigned the previous day and that chief financial officer Colette Kress would serve as interim CEO while the board searches for a permanent successor. Source
Open-weight models make their trained parameters, or weights, available so developers can download, run and modify the models. Fully open-source AI also provides materials such as source code, training data and instructions for reproducing how the models were developed. Source
The business case survives the boss
In a November 14 X post, Lambert summarized his concerns as “Political exposure” and “Investment.” Open-model builders, he argued, would need to find profitable products sooner without Huang’s backstop. Source
In his essay, he predicts that the flood of funding driven by Huang “will very likely dry up.” The essay identifies no announced post-departure funding cuts. Source
Lambert also explains why Nvidia still has reasons to support open models. A broader ecosystem can help prevent power from concentrating among a few giant customers that might develop their own chips and reduce Nvidia’s profits. SCMP’s reporting similarly describes strong commercial incentives that would likely keep Nvidia committed to the broader ecosystem, even if individual investments face closer scrutiny. Open models aren’t a charity project in this account; they’re good for the chip business. SourceSource
A US lab and political allies
“My biggest near-term worry is sloppy regulation at the state level, followed by national regulation,” Lambert writes. He credits Huang with using his influence to protect the ecosystem while its businesses and use cases developed. A business case doesn’t answer the phone in Washington. Source
Lambert also points to scarce computing capacity and the difficulty of assembling top research talent. Billions of dollars and a promising team don’t guarantee a leading model. Source
Lambert calls for a US open-model lab capable of competing with the most advanced models, and urges “soul-searching and coalition building” to support the ecosystem. Source
Comments