Dylan Patel: Jensen's exit threatens NVIDIA's supply chain dominance, open source ecosystem, and AI deal pipeline

Jensen Huang's exit threatens Nvidia on three fronts

Dylan Patel argues that Jensen Huang's departure from Nvidia creates three distinct and compounding vulnerabilities — in technical decision-making, supply chain leverage, and the open-source AI ecosystem — and that no obvious internal or external successor can cover all three.

Decision speed

Jensen's value was never in chip architecture. It was in arbitration. When Nvidia had three viable options simultaneously in play, Jensen would commit hard behind one at the last second, rally suppliers around it, and switch quickly if it went wrong. That is what Patel calls "founder mode" — the ability to make a unilateral call, absorb the risk, and move the entire supply chain behind it.

He points to two examples. Nvidia added Tensor Cores to Volta three months before tape-out — a decision that arguably created Nvidia's AI dominance. Blackwell required a last-minute pivot that, had it gone the other way, would have cost Nvidia significant market share. Without Jensen, those calls go to committee, which means they slow down, and the losers on options B and C stay bitter longer.

Supply chain

Patel is direct about what Jensen provided to Nvidia's manufacturing partners: the personal credibility and charisma to convince suppliers to commit 60–70% of capacity to Nvidia at below-market rates. A committee-led Nvidia cannot replicate that. Whoever succeeds Jensen is almost certainly less persuasive on that front, and suppliers know it.

“Open source in America would not exist without Jensen. Both Reflection and Thinking Machines are pushing the open source frontier in America, and both of them have massive investments in backing from NVIDIA. All these inference providers, which are multiple of the 10 plus billion dollar companies, all rely on open models from China. If we're slamming the hammer down on China way harder because the biggest defender against restrictions was Jensen and his team, the biggest loser is the open source ecosystem.”

Open-source ecosystem

Jensen was arguably the single biggest backstop for open-source AI in the United States. Nvidia backs both Reflection AI and Thinking Machines, the two companies Patel identifies as pushing the American open-source frontier. Downstream, the inference providers (several of which are valued above $10 billion) and application companies like Harvey, Instinct, and Sierra all rely on open models — many of which originate in China.

Jensen was also the most credible voice against tightening chip export restrictions into China. With his credibility now gone — and Nvidia under investigation and subpoena — Patel expects restrictions to tighten significantly. That creates a chain reaction: if Chinese labs face harder hardware constraints and pull back on open-sourcing their models, the American open-source ecosystem loses its most dynamic external contributor.

Successor shortlist

Internally, Patel names Ian Buck and Jonah Alben as the two most plausible candidates — Buck for charisma and deal-making, Alben for technical depth and org coverage. Neither covers the full Jensen surface area, and promoting either one likely weakens the function they currently own.

Externally, Patel's picks are Charlie Kwas (number two at Broadcom) and Johnny Srouji (SVP Hardware at Apple, the architect of Apple's chip strategy). He calls them the only two people in the world who can navigate both supply chain complexity and technical decision-making at Nvidia's level. Hock Tan and Lisa Su would be better fits in his view, but he thinks neither would take the role. Su's track record, Patel notes, was built beating an incompetent Intel — Nvidia's competitive field is categorically harder.

VC and inference cloud exposure

The venture community's exposure here is real. Patel says OpenAI has probably completed its last private round and Anthropic is heading toward an IPO, so VCs are largely out of the lab trade. The more vulnerable layer is the inference cloud companies — Base Ten, Fireworks, and others — that need compute credit and financing backstops to scale from $1 billion in revenue to $10 billion. Jensen and Nvidia's investment arm were providing that backstop. A committee-led Nvidia is likely to be far less active on that front.

Anthropic as an unintentional winner

Patel identifies Anthropic as structurally better positioned than OpenAI through this period. OpenAI is more than 90% dependent on Nvidia compute; Anthropic is at one-third or below. A disrupted Nvidia supply chain hits OpenAI harder. And if export restrictions tighten and Chinese open-source models slow or stop, Anthropic's closed-source model position becomes more valuable to every application company that currently leans on open weights.

Takeaway: No single successor candidate combines Jensen's supply chain influence, technical arbitration instinct, and deal-making reach. The gap is not just an org chart problem — it shows up in supplier confidence, open-source model availability, and the compute financing that keeps the inference cloud layer solvent.