Nvidia has shed some 1,800 billion dollars in market value since its October 6 record close after CEO Jensen Huang stepped down on Thursday amid a security scandal. The deeper risk, per Zonebourse’s analysis, is structural: Nvidia’s biggest customers are building their own AI chips, and Google — with TPUs, Gemini and its own chip-design AI — is best positioned to win what comes next.
Jensen Huang resigned as Nvidia CEO, sending shares down 4.3% to $166.19 and wiping out roughly $1.8 trillion in market capitalization since the October 6 record close.
Jensen Huang turned Nvidia into the world’s most valuable chipmaker, and his exit landed at the exact moment the economics of AI compute are shifting — from renting general-purpose GPUs to owning the entire stack, silicon included. That collision of a governance crisis and a business-model inflection is what makes this week extraordinary.
Why did Huang resign?
Nvidia announced Huang’s resignation last Thursday, after weekend reports of an extramarital affair with a woman suspected by US intelligence of being a Chinese spy. Colette Kress, Nvidia’s CFO since 2013, has stepped in as interim CEO while the board hunts for a permanent successor, Euronews reported.
The industry’s reaction split immediately. Silicon Valley has mobilized behind the man who founded Nvidia in 1993, with an online petition and an open letter drawing support from across the tech establishment, Al Jazeera reported.
“Jensen drove supply chain brilliantly at Nvidia, but many of those deals were a handshake with Jensen, not a written contract. Without Jensen, at least some of those critical deals will be lost.”
said Elon Musk, in a post on X on Saturday, November 14, 2026.
The “Stand With Jensen” letter — signed by Musk, Y Combinator CEO Garry Tan and YouTube cofounder Steve Chen, among others — has gathered more than 172,000 signatures. It states that Huang was “forced out without due process and without even a few days to make his case,” and that signers have “not been shown evidence substantiating the allegation that the woman at the centre of this controversy was a spy.”
Investors joined too. Brad Gerstner of Altimeter Capital and Gavin Baker of Atreides Management, both Nvidia shareholders, wrote to the board urging a “slow, measured and deliberate process” that leaves “all options on the table, including the potential return of Jensen Huang.”
How much value has Nvidia left behind?
The damage is measured in trillions. Nvidia’s stock closed at $166.19 last Thursday, down 4.3% on the resignation announcement and near its 52-week low, after wiping out roughly $1.8 trillion of market capitalization since its record close on October 6, when the stock finished at $239.24, according to Euronews.
| Metric | Figure |
|---|---|
| Record close (October 6) | $239.24 |
| Close after resignation (last Thursday) | $166.19, down 4.3% |
| Market value erased since record | ~$1.8 trillion |
| Forward P/E — Nvidia vs. AMD | 13.7x vs. 38.9x |
That valuation gap is the tell. Nvidia trades at 13.7 times expected earnings against 38.9 for AMD, per Koyfin data cited by The Information — a discount that prices in real doubt about the future, not just a bad news cycle.

Not everyone sees lasting damage. Kunjan Sobhani, senior semiconductor analyst at Bloomberg Intelligence, told Euronews the departure should not change Nvidia’s earnings forecasts for fiscal 2027 and 2028, while flagging longer-term risks around succession, deals and tightening China scrutiny as investors reprice the so-called “Jensen premium.”
Dan Coatsworth, markets head at AJ Bell, framed it the same way: the day-to-day business keeps running, but it is the long-term vision that now comes into question.
Over the past week, I’ve been overwhelmed by the love and support of friends, industry colleagues, and people I've never met. Thank you to everyone who has shown kindness to my family and me. It means the world to me. I came to the United States 53 years ago, and becoming an…
Who is Nvidia’s real threat?
Here is where the story stops being about one man. Zonebourse’s analysis argues the biggest danger to Nvidia is not AMD or China’s chipmakers — it is Nvidia’s own customers, who are internalizing ever-larger slices of the AI value chain for reasons both technical and financial.
The incentive is obvious. Nvidia’s margins run close to 75%, and every hyperscaler and AI lab would rather keep that money than rent it back. OpenAI plans to deploy 10 gigawatts of custom accelerators developed with Broadcom, while Anthropic was already running more than a million of Amazon’s Trainium2 chips as of April, according to Zonebourse. TrendForce projects that shipments of AI-specific ASICs could overtake GPUs as early as 2027.
Private bankers see the same crack forming. Cité Gestion told Capital that training the largest models still favors Nvidia’s raw compute, but inference — the daily use of those models — rewards more targeted, cheaper and more energy-efficient architectures, opening the door to custom silicon from Alphabet, Amazon, Microsoft and Meta.
Its verdict: less the end of a monopoly than an industrial normalization, in which Nvidia keeps selling plenty of chips while its growth and margins lag the very market it created.
What is recursive self-improvement?
The technical force behind this shift has a name: recursive self-improvement, or RSI — AI systems contributing to the design of their own next generations. The effects are already visible.
Nvidia's board unanimously accepted former CEO Jensen Huang's resignation, and does not intend to reinstate him as chief executive, the board chair said at the start of the company's earnings call. https://t.co/PZCTYBT3Av pic.twitter.com/6FgOAfPr6V
A model can shrink its own algorithms, yet it remains boxed in by the physics of the silicon underneath — memory capacity, bandwidth, and how fast chips talk to each other. That’s why labs now want to design their own semiconductors: a feedback loop where smarter models design better chips, and better chips train smarter models.
This is where Nvidia’s business model shows its seams. Nvidia must build general-purpose processors that serve thousands of customers. A lab designing its own ASIC can optimize the die around the shape of its future models — before those models even exist. That is an advantage competitors simply cannot buy.
Why is Google the vertical-integration favorite?
Google has been running this playbook for over a decade with its TPU accelerators, and it is the only player that can tune chips to the future needs of Gemini while tuning Gemini to the characteristics of its next chips. Because every die has a finite area and power budget, Google can spend its silicon where its models actually bottleneck.
The numbers back the strategy. Google says its TPU 8i, built specifically for inference, carries three times the integrated memory of the previous generation and delivers up to 80% better price-performance.
The software-to-silicon loop is no longer theoretical either. AlphaEvolve, a DeepMind agent, contributed to the design of a circuit for a future TPU generation and optimized a compute operation by 23% — shaving 1% off total Gemini training time. Google has used AlphaChip to optimize the physical layout of its TPUs since 2020, per Zonebourse.
Then there’s distribution, the part rivals can’t replicate. Nvidia must still win over top AI labs; OpenAI and Anthropic must industrialize chip design. Google already owns the chain — TPUs, Gemini, cloud infrastructure, and the channels to users.

Can CUDA’s moat survive the models?
Nvidia’s classic lock-in has always been CUDA, the software platform that made GPUs the backbone of modern AI and kept developers from straying. AI itself may dissolve that barrier. A scientific study published in July found that specialized models could automatically convert CUDA code to AMD’s software environment with an 88.2% success rate, reaching performance comparable to native code in 85% of the cases studied.
Nvidia is fighting back on the same logic. Its Nemotron models and a proposed $12.9 billion acquisition of Hugging Face make sense only in one scenario: if open-source models stay good enough that enterprises keep buying or renting Nvidia GPUs, rather than consuming AI as a service that runs on the labs’ own accelerators.
NVIDIA Announces CEO Transition November 12, 2026 Jensen Huang Resigns Colette Kress, Executive Vice President and Chief Financial Officer, Appointed Interim CEO Harvey C. Jones Appointed First-Ever Chair of the Board SANTA CLARA, Calif., Nov. 12, 2026 NVIDIA (NASDAQ: NVDA)… https://t.co/csjWZuI3Tw
The scale of that bet is staggering. By the end of October, Nvidia’s multi-year commitments tied to cloud services and data-center rentals had climbed to $102 billion, up from $74 billion three months earlier — infrastructure partly earmarked for developing its own models.
Zonebourse poses the uncomfortable question outright: will a successor without Huang be willing to sacrifice part of Nvidia’s exceptional profitability to defend an ecosystem the company’s chip demand ultimately depends on?
Will Huang come back?
That is the open question the market is now trading on. Kress holds the interim seat while the board weighs its options, Wall Street is split between bargain hunters and those advising patience until the dust settles, and the “Stand With Jensen” campaign keeps pressure on the board to reverse course. The letter’s demand is blunt: “Jensen deserves due process, not mob justice.”