The trillion-dollar question isn't whether Nvidia's neocloud bet will pay off—it's what happens when the man willing to backstop it is gone.
The Summary
- Ben Thompson frames Nvidia's strategy as "risky business"—using cash flow and credit to fund the chip buyers themselves, a circular financing model only possible for a company controlled by one risk-taking founder
- Interim CEO Colette Kress faces her biggest test Tuesday when Q3 results drop, amid questions about whether the new leadership will continue Huang's patron model for neoclouds
- The math problem: the AI buildout needs to generate roughly a trillion dollars in revenue industry-wide to justify current spending, and if it falls short, hyperscalers will hoard compute while neocloud assets flood the market
- Without Huang's willingness to shore up the ecosystem, Nvidia's hedge could become a millstone just as supply chain ramps bring more chips to market
The Signal
Nvidia built the AI gold rush, then started lending picks and shovels on credit. Thompson's analysis cuts through the euphoria to name what everyone's been dancing around: Nvidia wasn't just selling chips, it was financing the entire neocloud ecosystem. CoreWeave, Lambda Labs, the whole constellation of GPU-as-a-service startups—they bought hardware with Nvidia's balance sheet backing them.
This worked because Jensen Huang could do it. A founder-controlled, multi-trillion-dollar company can make bets that would get a professional CEO fired. Huang spent a career pushing limits and winning. But that's past tense now, and Colette Kress, the interim CEO and longtime CFO, has to answer for it Tuesday.
"Bears called this circular financing; bulls called it a necessary bridge to a future where AI revenue paid down these companies' debt and drove massive upside."
The revenue has to materialize. Not eventually—soon. Thompson pegs it at around a trillion dollars industry-wide, the kind of number that makes you check your math twice. If AI applications don't generate cash at that scale, the whole structure wobbles. Hyperscalers like Microsoft and Google will pull compute inward, prioritizing their own models and products. They become the safe haven. Everyone else scrambles.
Here's where it gets ugly for Nvidia's neocloud partners:
- They lose their patron just as creditors and equity holders get nervous
- Their GPU assets could flood the secondary market at fire-sale prices
- Supply chain ramps mean new chips keep arriving even as demand softens
- Hyperscalers, now seen as safe, capture the spending that remains
The turmoil Nvidia is in isn't just about leadership transition. It's about whether the company will continue underwriting an ecosystem that only made sense because one man was willing to bet the company on it. Kress is a seasoned operator, but CFOs optimize. Founders take existential risks.
The Implication
Watch Tuesday's earnings call for what Kress doesn't say. If there's no clear commitment to supporting neocloud partners, that's your signal. The smart money will move to hyperscaler compute deals and away from the smaller players. For companies building AI agents or applications, this means rethinking your infrastructure bets. Hyperscaler lock-in suddenly looks less like a risk and more like insurance.
The bigger question is whether the Web4 buildout—the agent economy that needs all this compute—can generate revenue fast enough to justify the hardware spend. If it can't, we're looking at a correction that makes the dot-com bust look orderly. Nvidia's stock price has priced in a future where AI agents are everywhere. That future still has to actually arrive.