10 Oversold AI Stocks: Where Wall Street Analysts Are Spotting 50%+ Upside

The AI selloff has opened a wide gap between share prices and Wall Street’s expectations. Some of the stocks below have lost nearly half their value in 30 days, while their average analyst targets still imply gains well above 50%. It sounds like an opportunity but can also mean the market has moved faster than the research notes.

The selloff has punished AI capacity builders and hardware suppliers particularly hard. Between September 23 and October 23, Cerebras lost 46.0%, CoreWeave fell 37.6% and Nebius dropped 30.0%. Seagate and Coherent each lost more than 30%. These businesses face different pressures: neoclouds must finance construction, optical suppliers depend on customer deployments, and storage manufacturers face the possibility that new competing capacity will weaken pricing. Rising demand alone cannot protect their shares.

On October 14, Gartner warned that more than $3.5 trillion of AI infrastructure spending could face impairment by 2030 if providers maintain their investment pace. A data center can fill up and still earn too little to justify its construction cost. Interest accrues before rent arrives, and delays extend that gap. CoreWeave has floating-rate debt that exposes it to higher interest expense; IREN faces financing risk on future expansion even where existing borrowing is fixed or hedged. See how those differences affect the cash left for shareholders here.

Financing can also change what shareholders own. Raising equity may keep a project on schedule while spreading its eventual profits across more shares. Borrowing preserves ownership initially but gives lenders a prior claim on cash. A company can therefore deliver its promised revenue growth without delivering the return implied by an old price target. Customer prepayments offer another source of funding, but cash collected today still has to pay for future services. Oracle’s cash-flow breakdown shows how much that distinction matters after the construction bill.

Semiconductor investors face another problem: AI spending can keep growing while customers shift orders between competing chip designs. That makes an industry growth forecast a poor substitute for a company-specific earnings forecast. A chip designer must defend its share and pricing; a foundry may retain the manufacturing work when customers switch designs, but still suffers if overall orders slow. See how Nvidia and TSMC divide those risks here.

Investors should not assume analysts have identified the bottom. A target set 20% above a stock’s former price can suddenly imply 100% upside after the shares fall 40%, even without any improvement in the business. Analysts can update late, and further cuts over the coming weeks could erase much of the apparent opportunity. Some cuts have already arrived: UBS lowered its CoreWeave target from $120 to $105 on October 23.

That is why we pair targets with short interest and hedge fund ownership. Heavy short interest can expose disagreement, although convertible-debt and transaction hedges can inflate it. Quarterly holdings cannot identify who bought this month’s dip. Coatue’s Intel and Cerebras positions show why institutional backing also needs context. Neither measure guarantees a recovery. Here, oversold means sharply sold down, not necessarily below a technical RSI threshold.

Methodology

We revisited a 50-stock selloff shortlist drawn from a review of 124 U.S.-listed companies across the AI ecosystem. The shortlist used split-adjusted closing-price changes from September 23 through October 23, 2026, and required sufficient price history and at least three identifiable analyst targets in the consensus dataset. Its stocks had declined at least 12.41% over that window.

For this large-company list, we required a market capitalization of at least $20 billion and direct commercial exposure to AI computing infrastructure: chips, cloud capacity, networking, data-center equipment, memory or storage. These restrictions left 27 companies from the original selloff shortlist. The group includes focused infrastructure suppliers and diversified companies with substantial infrastructure businesses; it is not a pure-play-only screen.

We ranked those 27 companies by upside to their average 12-month S&P Global analyst price target displayed by Stock Analysis. Upside equals the target divided by the October 23 closing price, minus one. Missing, insufficient or unreconciled target coverage was excluded. The ten highest-upside stocks appear below, counting down to number one; alphabetical ticker order breaks ties. Market capitalizations and targets use the October 23 snapshot.

Short interest uses the September 30 settlement-date snapshot and is expressed as a percentage of estimated float, except for Cerebras, where conflicting float estimates make the number of shares sold short more reliable. Hedge fund counts come from Insider Monkey’s Q2 2026 database and are compared with Q1 where meaningful. These positioning measures provide context and do not determine the ranking. Float estimates vary by provider and can lag share issuance. All prices and targets are in dollars.

Why are we interested in the stocks that hedge funds pile into? The reason is simple: our research has shown that we can outperform the market by imitating the top stock picks of the best hedge funds. Insider Monkey’s quarterly newsletter strategy selects 14 small-cap and large-cap stocks every quarter and returned 599.2% from May 2014 through June 2, 2026, beating its benchmark by 372 percentage points (see more details here).

#10. Micron Technology, Inc. (NASDAQ:MU)

Analyst Upside: 77.88%

30-Day Decline: 18.13%

Short Interest: 2.31% of Float

Number of Hedge Fund Holders: 184

Micron’s selloff has exposed the main risk in owning a memory producer in that earnings can peak before demand does. The shares fell 18.1% during a semiconductor retreat fueled by rate worries and competition. Its September 30 outlook also offered little further margin expansion as next-quarter adjusted gross margin was projected at 86.25%, down from 87% in the quarter just reported.

Bulls have more than an AI label to work with because Micron’s Cloud Memory and Core Data Center businesses generated a combined $34.29 billion of fiscal fourth-quarter revenue, 63% of the total. High-bandwidth memory feeds AI accelerators, while servers also need conventional memory. Management’s $61.5 billion next-quarter revenue forecast implies another 13.4% increase from $54.23 billion. The question is how much of that growth survives in earnings if memory supply catches up and pricing softens.

Q2 hedge fund holders rose to 184 from 154 in Q1. Coatue increased its Micron stake from 165,000+ shares to 3.14 million, worth $3.63 billion at June-end and 7.46% of its reported 13F portfolio. That was a substantial commitment before the selloff. September 30 short interest was just 2.31% of float, so bearish positioning was relatively limited. Even so, the $1,561 analyst target depends on future earnings assumptions. If memory pricing weakens, the apparent bargain changes with it; our Micron earnings stress test shows how a lower profit base changes the stock’s valuation.