A price target is a forecast dressed as a number. When a bank cuts one by a third, the forecast has been rewritten, and the note that Wells Fargo sent to clients on Friday rewrote two of them.
The bank lowered its twelve-month targets on four U.S. technology companies. SanDisk went from $1,550 to $1,300, a reduction of 16.1%. Intel went from $120 to $80, a cut of 33.3%. Alphabet moved from $417 to $371, down 11%. Amazon was trimmed from $329 to $326, a change of less than 1%.
The pattern inside those four numbers is the interesting part. The two companies that make things, memory chips and processors, took the deepest revisions. The two that sell cloud computing and advertising were adjusted at the margin, and one of them barely at all.
Both SanDisk and Intel reported quarterly results this week, which is the usual occasion for this kind of reset. Analysts build targets from projected earnings and a multiple applied to them. When a quarter arrives with numbers that differ from the model, the model is rebuilt, and the target moves with it. A cut of a third is not a comment on the share price. It is an admission that the earnings power assumed in the previous note no longer holds.
That is why the Intel cut carries more information than the others. A target of $80 against a previous $120 implies the analyst expects either materially lower profit or a materially lower multiple, or both. Intel’s problem is not demand for processors. It is the cost of building the capacity to make them, and the time it takes for that capacity to generate a return. A foundry business consumes capital before it earns any, and the gap between the two is measured in years. Analysts who model that gap have to decide how much of it to capitalize and how much to charge against near-term earnings, and small changes in that assumption move the target a long way.
SanDisk’s 16% reduction has a different origin. Memory is a cyclical business, and the analyst’s task is to decide where in the cycle a company sits. When prices for NAND flash are rising, margins expand faster than revenue and earnings estimates climb quickly. When the rate of increase slows, the same model produces a lower number even if the business is still growing. A target cut that leaves the new figure far above the old one’s starting point is a statement about the pace of improvement, not its direction.
Alphabet’s 11% trim follows a week in which the shares fell more than 11% in a single session after the company reported third-quarter results. Analysts rarely leave a target untouched after a move of that size, because the target is supposed to describe where the stock goes from here, not where it has been. Cutting a target to $371 keeps a premium to the market while acknowledging that the assumption of uninterrupted earnings growth has been qualified.
Amazon’s $3 reduction is close to noise. It reads as housekeeping, the kind of adjustment a bank makes when the model changes by a fraction and the analyst wants the published number to match the spreadsheet.
Taken together, the four revisions describe a market that has begun to separate companies by their capital intensity. The two names that build physical capacity took the largest cuts. The two that rent capacity took small ones, and the one with the strongest cash generation was left nearly alone.
That separation has been building for a year. Investors spent most of the past two years rewarding any company with exposure to artificial-intelligence demand, without much regard for how the revenue was produced. The distinction that matters now is between businesses that convert demand into cash and businesses that convert demand into factories. Wells Fargo’s note is one bank’s version of that distinction, and its rankings are unambiguous.
There is a limit to what a single note tells you. Sell-side targets cluster: when one bank revises, others tend to follow within weeks, partly because they share the same inputs and partly because standing apart is uncomfortable. A change of this size at a large bank often precedes a round of similar moves across the Street rather than standing alone.
The other caveat is timing. Targets are set with a twelve-month horizon and revised after the fact, which makes them a lagging indicator. By the time a bank publishes a 33% cut, the shares have usually already moved. Intel’s own week included a fall of that order, so the note is documenting a repricing that has happened rather than predicting one.
What the note does show is where the analytical disagreement sits. Nobody on the sell side is arguing about whether AI infrastructure spending is real. The argument is about who keeps the profit. Memory makers keep it when supply is short and lose it when supply catches up. Foundries keep it late, if they keep it at all. Hyperscalers keep it only if their own customers pay enough to cover depreciation that is now measured in tens of billions of dollars a quarter.
For investors holding all four names, the practical read is that the market is no longer paying the same multiple for the same theme. Two of these companies are being valued on the cycle and two on the platform, and the bank’s numbers reflect that split more clearly than its prose would.
Wells Fargo did not publish a thesis that the sector is finished. It published four numbers, and three of them moved down. In a week when several of the largest companies in the world reported results and several of them fell, that is the least dramatic way to say the same thing.
