← Back to blog
AiAbout 7 min read

The Jensen Discount: What Nvidia Loses When Its Founder Walks Out

Published Nov 17, 2026
The Jensen Discount: What Nvidia Loses When Its Founder Walks Out

On November 12, Nvidia's stock fell 14% in the first twenty minutes of trading. It closed down 4.34%, but market value still slipped below $4 trillion, the first time since mid-2025. Three days earlier it had already taken a hit: on November 9 it closed at $178.67, a seven-month low, with about $193 billion gone in a session.

The news itself is not complicated. Founder Jensen Huang resigned as president, CEO, and director, effective immediately. CFO Colette Kress takes over as interim CEO, and the board brought in Harvey C. Jones as the company's first board chair, a position that did not exist before.

Most commentary stops at the scandal. Line up the numbers from those days, though, and the market looks less like it sold a scandal than like it finally priced what the valuation model had never carried on its own: how much of this company's worth is attached to one man.

The discount is calculated, not emotional

The bluntest number on Wall Street comes from Dan Ives at Yorkville. With Huang, Nvidia is worth $300 a share. Without him, $225. That is a 25% gap, about $600 billion in market value. Crude, obviously. It still marks out a range the market will accept.

The multiple underneath is more interesting. Against FY2028 consensus earnings of $15.79 a share, the November 11 close works out to a P/E of about 11. A company growing revenue 90% year over year at a 74% gross margin, trading at 11 times. That is not a business getting worse. Analyst EPS revisions over 90 days were still +5.2%, so forecasts were not cut. The amount anyone will pay for those forecasts was.

Within four trading days, at least ten firms moved the stock to Hold. Bernstein halved its target from $400 to $200. Morgan Stanley dropped it from its top pick list, Bank of America from its US 1 list, Baird to Neutral. Ten downgrades in ten days, the tightest cluster for any single stock this year. Watch the order, though: all of them moved after the 12.75% decline.

Chart of Nvidia's share price drop and the valuation discount applied after Jensen Huang's resignation

The numbers in the report barely moved

After the close on November 17, Nvidia reported fiscal 2027 third-quarter results. The first report since Huang left, and the first time Kress sat alone at the head of the call.

The figures run roughly like this. Guidance of $108 billion, plus or minus 2%. Consensus of $108.7 billion to $109.3 billion, up about 90% year over year. Adjusted earnings of $2.47 a share against $1.30 a year earlier. Data center revenue around $101.8 billion. Gross margin guidance of 74%, a point below last quarter's 75%.

Nothing there looks like a collapse. Last quarter revenue was $96.2 billion and data center $89 billion, of which hyperscale cloud contributed $48.7 billion, up 13% sequentially, and AI cloud and industrial enterprise customers $40.3 billion, up 25%. Orders are not the problem. Capacity is.

What matters are three things the call may not settle. First, whether fourth-quarter guidance reaches the $124.3 billion to $126.4 billion consensus. With management just changed, a conservative guide is close to certain, and it will smother the good news from any beat. Second, the Vera Rubin cadence. Volume shipments began in August, and management had expected Rubin to be 20% of third-quarter data center revenue, but KeyBanc cut its 2026 Rubin GPU forecast from 1.7-1.8 million units to 1.5 million on tight HBM4 supply, with GB300 filling the gap. Third, and most delicate: does the company still stand behind its fiscal 2028 framework of about 70% growth? Morgan Stanley's model says 52%, some $70 billion below consensus, and the bank says it is not sure the company will give guidance next year at all.

The real risk is not on the income statement

Nvidia carries $145 billion in supply commitments and prepayments, a figure Kress gave on the August call, well above the prior quarter. The goods are ordered and the cash is out. If customers pick up on schedule it becomes revenue. If a customer delays energizing a data hall, it becomes inventory.

There is a second kind of commitment. By public disclosure, Nvidia holds more than $99 billion in equity investments, mostly in its own customers and ecosystem partners, including strategic stakes in labs such as OpenAI and Anthropic and support for the new cloud builders. The shovel seller is funding the prospectors. In an upcycle that is an accelerant. In a downcycle, an amplifier.

The third-quarter 13F shows a balance sheet being reshaped. Disclosed holdings fell from $63.44 billion to $48.33 billion. It cut about 53.69 million Intel shares, a 25% reduction, and added about 7.71 million SpaceX shares, up 6.3%, making SpaceX the largest disclosed position at $19.68 billion. None of that is wrong in isolation, but it landed in the week of the handover, which makes it look like clearing room for uncertainty.

Nobody was ever ready for this succession

A headhunter interviewed by the New York Times was blunt. Finding someone inside Nvidia to run the whole company on short notice is hard: most senior executives own one product line or one technical direction. They have never managed thousands of people.

Kress is one of the few exceptions. She is 59, joined in September 2013, and spent nearly 25 years in finance at Cisco, Microsoft, and Texas Instruments before that. She has co-hosted the quarterly call with Huang for years, so investors know her voice. Her strength is steadiness. Her weakness is as plain: Nvidia's moat is judgment about technical direction, and for 33 years one person made that call.

The internal name with the loudest support is Ian Buck, vice president for hyperscale and high-performance computing and an early contributor to CUDA. Polymarket gives him about 32% and Kress 22%. The outside names floated are Arm CEO Rene Haas and Fisher, an executive who has spent 32 years at Nvidia.

One detail matters. Nvidia is flat enough that more than 50 people report directly to the CEO. Under a founder that is speed. After a handover it is a single point of failure. The board inventing a chairman's role where none existed is, in essence, adding back a layer the company never needed.

For China, this is a window, not an opportunity

The hottest line in domestic discussion is that the window for domestic compute substitution has opened. That is half right.

The right half: when a customer evaluates a second supplier, the reason they need most was never that you are stronger. It is that they cannot have only one. For three years Nvidia pinned that argument down with CUDA. Now procurement departments have a reason they can put in the minutes with no technical justification at all: key-person risk.

The wrong half is timing. The weak spot in domestic GPUs is not peak performance. It is the software stack, framework support, and the two years it takes to move a generation of cards from working to working well. What the window opens is willingness to try, not capacity or ecosystem. On November 16 a new Commerce Department rule added another lock: data center customers in Indonesia, Malaysia, Singapore, Thailand, and Brazil need licenses to buy advanced AI chips, with a 90-day transition for facilities already serving Chinese customers. The route of renting cards through overseas data centers has been formally narrowed.

So the more accurate statement: the urgency of domestic substitution rose, and so did the difficulty. Both at once.

One last thing

Much of what Huang did across 33 years was not chips. It was turning a components company into a story: the black leather jacket, the two-hour roadmap on the GTC stage, the yearly sermon that compute is still not enough. That story carried Nvidia's premium, and it means the day he walked out the company had to answer who are you all over again.

From November 8 to November 17, nine days, a $4 trillion company did three things. The CEO left. The board was rebuilt. The first earnings report went out on schedule. Operations did not stop, orders did not reverse, the roadmap did not change. The organization is still standing.

But the 25% discount says something else. The organization is still standing, and the story has to be rewritten. A story is the one thing an earnings report can never fill in.

Related articles