When the Central Bank of AI Changes Chairs
| Author | Tomasz Tunguz |
|---|---|
| Published | November 12, 2026 |
| Format | Article |
| Original | Read the original |
Jensen Huang ran Nvidia as the central bank of AI, funding the buildout through M&A, equity stakes, lease guarantees and stretched payment terms; with a CFO now in charge, any tightening of those slows the AI economy the way Fed tightening slows the broader one.
When the Central Bank of AI Changes Chairs
Saved via This is a great framing… (tweet), Will Quist’s quote: “Nvidia as the central bank of AI. Hard to imagine that any successor to Jensen will have as much absolute access to liquidity…” It is the same-day companion to Yoni Rechtman’s The End of Nvidia as the Central Bank of AI (Note to the Slow Team), which uses the identical metaphor.
Key claims
- The news. Jensen Huang resigned as Nvidia CEO; CFO Colette Kress is interim CEO and Harvey Jones becomes chairman.
- Nvidia deployed 112% of operating cash flow into equity investments and M&A last quarter.
- If the central bank tightens, the AI economy slows. Over two years Nvidia committed $42B to M&A and technology buyouts, invested $61B (cost basis) in equity stakes across 229 startups and 13 public companies, extended $164.5B in lease guarantees and cloud capacity backstops, and carries $63.1B in customer receivables with terms stretching up to a year. A CFO-led regime under board scrutiny that tightens any of those slows AI buildouts (see AI Capex, Neocloud).
- Nvidia fused the startup and chip ecosystems. It was chip supplier, lead venture investor, anchor cloud customer and lease guarantor across 200+ startups at once, and those deals “ran on Huang’s personal conviction & relationships.”
- Open-source AI loses its patron. Nvidia committed $22B to its Nemotron stack across Hugging Face, Poolside and Essential AI; Meta, Google and at least eight $100M+ US startups will have to carry American open-weight models without it.
- Expect tighter export controls, as the episode raises pressure in Washington.
- An opening for custom silicon. Custom ASICs are already 28% of AI server shipments and 15–20% of accelerator revenue in 2026, led by Google’s TPUs (65% of custom silicon) alongside AWS Trainium3, Meta MTIA and Broadcom XPUs; turbulence at a single-source supplier speeds the shift.
Notable quotes
“For the past four years, Jensen Huang ran the Central Bank of AI.”
“Under a CFO-led regime facing heightened board scrutiny, any tightening of those equity checks, payment terms or lease backstops will slow AI buildouts just as the broader economy slows when the Federal Reserve tightens credit.”
“Under Huang, Nvidia acted simultaneously as chip supplier, lead venture investor, anchor cloud customer & lease guarantor across more than 200 startups.”
“Leadership turbulence at a single-source supplier accelerates that shift.”
“It is hard to overstate how central Nvidia has become to startup financing, M&A exits & policy across the fastest-growing segment of the US economy.”
How it connects
- The End of Nvidia as the Central Bank of AI (Note to the Slow Team): the investor’s what-to-do-now memo; Tunguz supplies the balance-sheet numbers behind the metaphor.
- Nvidia, Jensen Huang, AI Capex, Neocloud.
- Tomasz Tunguz: his other writing in the wiki.