Nvidia Just Lost Its CEO, Its Moat, and 30% of Its Value in That Order

Let's get the scoreboard straight before anyone tries to sell you a story. NVIDIA closed Thursday at $166.19, down 4.34%, with a market cap near $4 trillion. It's been in freefall for a month: down roughly 25% on the month, 26% off the quarter's start, and about 30% off its October 6 peak near $243. This isn't a correction. It's a de-rating with a body count.

Here's the part the headlines keep under-rotating. The selloff didn't start with the earnings. It started with the CEO walking out the door.

The CEO Left Before the Numbers Did

Jensen Huang departed in November — reports of extramarital affairs tangled up with a China-export and national-security row, plus a board investigation into what was described as a "foreign intelligence effort" aimed at him. CFO Colette Kress stepped in as interim CEO. Here's what I actually care about: the November 17 earnings call will be NVIDIA's first one held without Huang.

You can't separate that from the tape. NVIDIA has spent a decade selling the market a single, irreplaceable mind. Take the mind out of the pitch, and suddenly the "$279 billion in supply commitments prove demand" argument gets a lot harder to make. The market doesn't price earnings misses first. It prices uncertainty first, and earnings second.

This Wasn't a Maco Sell-Off. It Was a Circuit Breaker on AI.

The scary number here isn't NVIDIA's drawdown. It's the $5 trillion in market value erased across roughly 47 AI and chip stocks since October 6 — while the rest of the S&P 500 gained about $1.3 trillion in the same window. That's not a broad panic. That's investors auditing the entire AI build-out at once and deciding the financing story had more loose ends than they were willing to hold.

The loose ends are real and they're not hypothetical anymore. Management publicly acknowledged the "circular financing" critique. Guarantee obligations sit at ~$108.5 billion. Supply commitments doubled to $279 billion in a single quarter. Days-sales-outstanding climbed from 45 to 60. You can read all of that as "demand so strong we're inventing new accounting" or as "we've tied ourselves to a demand curve we can't un-sign." Both are true at the same time.

The Bulls Have a Case. It's Just Not Cheap.

I'll defend the other side fairly. NVIDIA closed at roughly 21x trailing and low-teen forward earnings. The Q3 revenue guide of ~$108 billion is a monster number. Vera Rubin is guided to be about 20% of data-center revenue this quarter, and supply stays a bottleneck into FY2028. Data Center was ~92.5% of Q2 revenue — that's not a bug, it's the most profitable segment on Earth doing exactly what it was hired to do.

And yes, roughly 59 of 62 analysts still rate it a Buy.

But "cheap for a reason" is the oldest trap in the book. The bears aren't wrong about the stuff — margin compression (Q4 gross margins guided down to 71–72% from 75%), single-segment dependence, custom-silicon competition, TSMC constraints, the China exposure. They're wrong if you think the number itself was ever going to save the stock. NVIDIA has closed down the session after 7 of its last 13 beats. A beat is now the entry fee, not the catalyst.

My Read

Everyone's treating November 17 like a verdict. It's not. It's a measurement of a business that is still printing money while simultaneously losing its founder, its governance, and its narrative.

Here's what I actually care about: the market has already priced a disaster into this tape — 30% down, forward P/E hovering near 10.6. If Kress walks in, hands over $108 billion in revenue, and says the financing concerns are overblown, this pops. If she stammers on the circular-financing question or the China row gets louder, "cheap" becomes "cheap forever" in about four minutes.

The irony of this whole month? NVIDIA never had to prove it could build chips. It had to prove someone else could run the company. So far, the market hasn't gotten an answer — and it's charging for the privilege of waiting.