Between Nov 11 and Nov 13, four banks moved Nvidia to Hold inside the same week: HSBC, Truist, Bernstein and Baird. Three more kept their Buy ratings and cut what they would pay. Goldman Sachs went from 300 to 220, Citi from 285 to 225, Bernstein from 400 to 200, and JPMorgan stayed positive with a lower target.
The target changes are the loud part. The quiet number is the multiple. Nvidia now trades at about 13 times expected earnings, against 20 times before the cuts. Seven points of multiple is not a rounding error in a rating model; it is the difference between pricing a franchise and pricing a hardware cycle.
Read the two kinds of action separately. A Hold from Truist or Baird is a statement about risk-adjusted return at the current price. A Buy with a target cut from 300 to 220 is a statement about the price at which the analyst still wants the stock. Both land in the same place: less willingness to pay per dollar of earnings.
The cuts arrive four days before the Nov 17 third-quarter report, and they name what the market wants to see. Data center revenue, gross margin, guidance for Blackwell, and how the interim chief executive arrangement around Colette Kress is meant to work.
That last item carries more weight than usual. A company reporting a quarter in the middle of a leadership handover gets asked two questions at once: whether the numbers are good, and who owns the next set. Analysts cutting targets while holding ratings are saying the cash flows may survive and the certainty around them has fallen.
Which is what a multiple is: a price for certainty. Nvidia at 20 times earnings was priced as a company whose growth was underwritten by a founder who set supply commitments and pricing himself. At 13 times, the market prices the same cash flows with a wider range of outcomes attached. Nothing in the earnings model changed; the discount applied to the story did.
The bull case at 13 times is straightforward. If gross margin holds and Blackwell guidance lands near expectations, 13 times is a low price for a business with Nvidia’s installed base, and four Hold ratings become the setup rather than the warning. A de-rating that happens before an earnings beat tends to reverse faster than it arrived.
The bear case is that 13 times is not cheap if the earnings estimates are the next thing to move. A target cut from 400 to 200 by an analyst who also moved to Hold is a 50% reduction in expected value; a cut from 300 to 220 is 27%. Those are not small edits to a model. They suggest the out-year numbers in the consensus are still too high.
The comparable that matters is Nvidia against Nvidia a week ago. Nothing about the competitive position changed between Nov 11 and Nov 13. What changed is the number of analysts willing to put a lower number on paper, and the multiple the street is now using as an anchor.
There is a mechanical reason multiples compress before a print. Analysts who expect a guidance reset cut the target first and explain afterward, because a stale target damages a franchise more than a cautious one. When four banks move to Hold in three days, the desk reading is that the private expectation for the quarter sits below the published one.
The four items to check on Nov 17, ordered by how much they move the multiple: gross margin first, because it is the cleanest read on pricing power. Data center revenue second, because it is the growth line. Blackwell guidance third, because it sets the 2027 model. The Kress transition last, because it is the one item management can settle with a single sentence.
What the handover costs, if it costs anything, shows up in the multiple before it shows up in revenue. Buyers of a stock during a leadership transition want a discount for uncertainty, and the sell side has handed them one: seven points of multiple, four Hold ratings and a cluster of target cuts inside 72 hours.
For anyone holding the stock, the question is not whether 13 times is a good price for Nvidia. It is whether 13 times is the right price for Nvidia without a permanent chief executive, with a Blackwell cycle that has to be guided rather than described, and with customers who have already committed capital on the assumption that supply arrives on schedule.
The verifiable hypothesis: if gross margin holds near its recent level and Blackwell guidance does not come in below the current range, 13 times is the floor of this de-rating and the Hold ratings get reversed within a quarter. If margin slips by even a point or two, the multiple has further to fall, because the market will begin pricing Nvidia as a cyclical, and cyclicals do not trade at 13 times peak earnings.
The targets are already public. What is left is the number the company prints, and whether the four banks that moved to Hold this week were early or simply first.
Nov 17 settles the first half of that. The second half takes a quarter, because a guidance cut is visible in a day and a margin trend is visible only in the following print. Until then, 13 times is the price of the argument, and the spread between the highest target at 400 and the lowest at 200 measures how little the street agrees on what the next four quarters contain.


