Cheap AI Stocks Still Worth Buying After Selloff

Artificial intelligence infrastructure stocks are trading at a discount, but the price drop isn’t because earnings are bad. The market is selling off because sentiment turned sour, not because the business models are broken.

Samsung Electronics reported preliminary operating profit that jumped 19 times year over year to nearly $60 billion, driven almost entirely by AI memory demand. This data contradicts the bear case that the AI cycle has peaked. International Business Machines pre-announced a second-quarter revenue miss after CEO Arvind Krishna revealed clients were pulling capital out of software and consulting deals to buy supply-constrained servers, storage, and memory ahead of expected price hikes. This cannibalization of budgets in one sector to fund another shows the underlying demand is real, even if the balance sheets of some companies look strained.

One must consider the gap between what insiders build conviction around and what the public feels comfortable owning. Billionaires and brand partners see the future in products like Meta Platforms’ video-recording AI glasses, but the public views them as “creepy” and invasive surveillance devices. This disconnect often precedes a market correction, as retail investors abandon a sector they no longer understand or trust. The current dip in AI infrastructure stocks feels similar to the skepticism faced by face computers, yet the financials for major players remain solid.

Five Stocks Trading Below Their Peaks

SpaceX Technologies Inc. is the most argued-about stock in the market. Oppenheimer, Goldman Sachs, and Morgan Stanley all carry “buy” ratings with targets ranging from $205 to $300. Revenue estimates jump from $18.6 billion to $38.7 billion this year, then to $74.2 billion in 2027 and $135 billion in 2028. Twenty-one Wall Street firms have already penciled in 2030 estimates, clustering around $330 billion in revenue. At $150 a share, the math works.

TeraWulf Inc. pivoted from Bitcoin mining to leasing power for AI campuses. The company signed a 20-year, $19 billion deal with Anthropic for a 401-megawatt AI facility in Kentucky. Revenue growth estimates run 89% this year, 210% in 2027, then 72% and 56% after that, taking the company from $168 million in trailing revenue toward $3.3 billion within five years. Gross margins expand from 50% to 70% over that stretch. The stock trades at 33.6 times EBITDA, which is remarkably cheap for triple-digit growth.

Amazon.com, Inc. tapped the debt market for $25 billion to fund AI infrastructure and launched 29 more low Earth orbit satellites, bringing its total to 396. This confirms the satellite broadband race has moved from concept to commercial deployment. Amazon trades at 22.6 times forward earnings and 11 times forward EBITDA, both essentially five-year lows, while revenue growth holds steady in the low double digits and margins expand from the mid-20s toward the mid-30s because of Amazon Web Services.

Palantir Technologies Inc. got caught in a broader software selloff, with shares sitting roughly 26% to 27% below highs. The stock remains trapped below a declining 200-day moving average, but the growth profile underneath is extraordinary. Revenue growth is expected at 73% this year, followed by 46%, 44%, 52%, and 49% in the subsequent years. The stock trades at 75.5 times forward earnings and 56.5 times forward EBITDA for what could become a 70% to 80% compounder.

Micron Technology Inc. sits 22% below its highs. The bears say memory chips have peaked, but Samsung’s blowout quarter says otherwise. The real fear is demand 6 to 12 months out, once new memory supply comes online. This question gets answered in about three weeks when hyperscalers report earnings. Micron’s past pullbacks during this cycle have all bottomed in the 20% to 30% drawdown range, and the stock sits at a 22.6% drawdown right now.