Dek: Nvidia lost its founder, refused to take him back, and then shipped the fastest product ramp in its history inside seven days. The interesting question isn't whether the quarter was good. It's who decides what happens next.
The week Jensen Huang left Nvidia, in order
The sequence matters, because each step explains the next one.
Jensen Huang resigned as Nvidia's president and CEO, announced on 12 November. Board chair Harvey Jones said the resignation followed "disclosure of personal behavior inconsistent with the company's code of conduct, standards, and cultural norms".
On 17 November the board held an earnings call. Jones said the board didn't intend to reinstate Huang and wouldn't take questions about the circumstances or its investigation. Hours earlier, Huang had said publicly that he'd be willing to return.
Same day, same company: the first Nvidia earnings call and the first all-hands without the founder.
Colette Kress and the continuity speech
Colette Kress has been Nvidia's CFO since 2013, and she's now also interim CEO. She opened the call with the line every board wants said out loud: the company's ambition, its appetite for innovation, and its commitment to customers, partners and shareholders wouldn't change.
She said the past month had been "business as usual", and that customers, suppliers and partners were "100% behind us". She also mentioned that TSMC's leadership had been at Nvidia's offices that week, which is the kind of detail that does more work than an adjective.
Read the framing carefully. Kress didn't argue that nothing had happened. She argued that the operating rhythm hadn't changed, and then spent the rest of the call proving it with product dates.
Vera Rubin: the ramp that made the case
The strongest evidence for continuity wasn't a sentence. It was a shipping schedule.
Vera Rubin delivered the fastest product ramp in Nvidia's history, contributing close to $20 billion in its first production quarter. Hyperscalers brought it into production in about a month, roughly half the time Grace Blackwell needed.
That number is the one to hold onto. A platform transition that lands in half the usual window, in the month after the CEO leaves, says more about the organization than any statement from the board.
Metric | Q3 fiscal 2027 |
|---|---|
Revenue | $115.8 billion, up 103 percent year over year |
Data center revenue | $107.6 billion, up 110 percent year over year |
GAAP earnings per share | $2.70 |
The quarter closed on 25 October, before Huang's exit, so none of it tests the new arrangement. What it does test is whether the roadmap survives a leadership change, and the roadmap came with dates: Vera Rubin Ultra next year, Feynman in 2028, and supply agreements running to the end of the decade.
Nvidia also guided the current quarter to $130 billion, plus or minus 3 percent, excluding any data center compute revenue from China.
The Nvidia all-hands: raises, equity and a CEO search
Money answers the question employees actually ask first, and Kress answered it at the all-hands.
She announced plans to accelerate pay raises and equity grants. Jones walked through the CEO search, which will consider internal and external candidates. Ian Buck covered the roadmap, according to reporting from WIRED and Business Insider cited by SiliconANGLE.
Accelerated equity in the week your founder leaves is a retention move, and it's a candid one. The people who build the next platform have options elsewhere, and the board knows it.
Jones framed the hire in terms the company has been using for two years: advanced silicon and national security dynamics are changing the market, and the next CEO will need to navigate those waters carefully.
Nvidia's export control committee outlasts the search
One decision from that week will still matter after a CEO is named.
On the morning of earnings, the board created a standing national security and export control oversight committee of three directors, with outside experts retained. That's a permanent structure, not a search committee that disbands.
The same week, Kress said Nvidia is no longer shipping licensed Hopper data center products to China, and that the US Commerce Department issued new export controls that the company still has time to study.
Put the two together and you get a company that has stopped treating export policy as a legal department problem. A standing committee with outside experts changes how fast Nvidia can say yes to a customer in a restricted market, which is a commercial fact, not a governance ornament.
What Nvidia allocation looks like for buyers now
For anyone trying to buy GPUs at scale, the practical rules got clearer on this call.
Allocation now weighs land, power and shell readiness first, then capital, then proof of off-take for customers that aren't investment-grade. Kress said all buyers are getting the same pricing, which is a direct answer to the accusation that the largest cloud customers get a discount.
The guarantee structure for neocloud partners also came up. Nvidia's backstop takes effect only after the partner has raised its own debt and equity, built the facility and installed the compute, and it lapses if the project stalls. Kress was blunt about whether that's new: "Is it new? No. Did we change anything? No."
On capital, she said the top priority is suppliers, through prepayments, with customer financing reserved for "very strong customers". That ordering tells you where the bottleneck sits.
Not demand.
The demand picture she described is agentic. Kress cited OpenRouter data showing agentic workloads account for roughly 80 percent of tokens, and said agentic AI is driving CPU demand that's far outstripping supply. Vera CPU is in use at CoreWeave, Meta, Nscale and OpenAI.
The interim CEO question the week didn't answer
A week of good execution doesn't settle a succession.
The interim CEO is the CFO, which works because the CFO already owned the numbers Wall Street grades the company on. It works less well for the parts of Nvidia that aren't financial: systems design, software, and the supply chain relationships that took a decade to build.
That's the question the board has to answer with a hire. Nvidia's moat is systems, software and supply chain, not a single chip, and a successor who reads it as a chip company would inherit the wrong map.
Watch the CEO announcement for one signal above all: whether the next person is introduced as an operator of the platform or as a caretaker of the quarter. The first keeps the roadmap. The second fights the last war.






