Open Models Just Lost Their Bodyguard, Says Interconnects
Open models didn't just lose a sponsor this week, they lost the person who picked up the phone in a crisis. That is the core of Nathan Lambert's essay on Interconnects, published Nov 14 at 12:23 UTC, which calls Jensen Huang's resignation "a major setback for the US development of an open model ecosystem": https://www.interconnects.ai/p/open-source-ai-without-jensen
Lambert's logic goes like this. Nvidia still has a business reason to back open weights: it keeps power from concentrating in a few giant customers who would rather build their own chips. But the company's number one job now is to stabilize, deliver to customers over the next 24 months and avoid implosion. Open models were one of Jensen's "founder-style bets," and Lambert expects the flood of funding to "very likely dry up." Worse, Jensen was more than a cheerleader in Lambert's telling; Jensen used real power to stop problems early, before they grew.
The timing is bad because the open-model economy is still a baby. The upside Lambert sees is that it has correctly latched onto "ownership of intelligence" and enterprise post-training. The downside is compute. Inference startups can't get enough chips because Anthropic and OpenAI are buying so much capacity, so open weights get served at fat margins and are nowhere near a commodity. In that world, closed labs with scale and chip-to-datacenter integration win. Add a political worry: sloppy state-level regulation, and the claim that the ecosystem has been "one, strong Chinese open-weight model away from a crisis" since the summer.
Money alone won't fix it. Lambert quotes SemiAnalysis CEO Dylan Patel: "It's not money, it's talent." Nemotron had the right ambition and strong people but hasn't delivered "mission accomplished." Thinking Machines' Inkling 2 Small is, in Lambert's view, the best Western open-weight model in a while, yet still far behind Kimi K3.1 Pro. The likeliest new allies are hyperscaler chiefs like Zuckerberg, Jassy or Nadella. Meanwhile Altimeter's Brad Gerstner and Atreides' Gavin Baker sent the Nvidia board a letter on Nov 14 urging it to consider Jensen's return (CNBC): https://www.cnbc.com/2026/11/14/nvidia-investors-jensen-huang-return-board-letter.html
The uncomfortable read: an ecosystem that collapses when one patron leaves was running on patronage, not on a business model. Lambert's essay is a call for coalition building, but the deeper fix is the one the post itself points at. Open weights need a revenue engine that pays for the next training run without anyone's personal protection.
← Back to all articles
Lambert's logic goes like this. Nvidia still has a business reason to back open weights: it keeps power from concentrating in a few giant customers who would rather build their own chips. But the company's number one job now is to stabilize, deliver to customers over the next 24 months and avoid implosion. Open models were one of Jensen's "founder-style bets," and Lambert expects the flood of funding to "very likely dry up." Worse, Jensen was more than a cheerleader in Lambert's telling; Jensen used real power to stop problems early, before they grew.
The timing is bad because the open-model economy is still a baby. The upside Lambert sees is that it has correctly latched onto "ownership of intelligence" and enterprise post-training. The downside is compute. Inference startups can't get enough chips because Anthropic and OpenAI are buying so much capacity, so open weights get served at fat margins and are nowhere near a commodity. In that world, closed labs with scale and chip-to-datacenter integration win. Add a political worry: sloppy state-level regulation, and the claim that the ecosystem has been "one, strong Chinese open-weight model away from a crisis" since the summer.
Money alone won't fix it. Lambert quotes SemiAnalysis CEO Dylan Patel: "It's not money, it's talent." Nemotron had the right ambition and strong people but hasn't delivered "mission accomplished." Thinking Machines' Inkling 2 Small is, in Lambert's view, the best Western open-weight model in a while, yet still far behind Kimi K3.1 Pro. The likeliest new allies are hyperscaler chiefs like Zuckerberg, Jassy or Nadella. Meanwhile Altimeter's Brad Gerstner and Atreides' Gavin Baker sent the Nvidia board a letter on Nov 14 urging it to consider Jensen's return (CNBC): https://www.cnbc.com/2026/11/14/nvidia-investors-jensen-huang-return-board-letter.html
The uncomfortable read: an ecosystem that collapses when one patron leaves was running on patronage, not on a business model. Lambert's essay is a call for coalition building, but the deeper fix is the one the post itself points at. Open weights need a revenue engine that pays for the next training run without anyone's personal protection.
Comments