3 AI Tech Stocks That Just Crushed Earnings: Are They Still Buys for the Long Term?

Hewlett Packard Enterprise, Micron, and Palantir all reported earnings and revenue above Wall Street expectations, with HPE revenue up 40% YoY and Palantir revenue up 85% to $1.63B. Micron’s February quarter revenue nearly tripled to $24B and EPS reached $12.20, though the article notes valuation and cyclicality risks. Overall, the piece is constructive on AI-driven demand, but more cautious on Micron at current valuation levels.

Analysis

The setup is less about “AI winners” broadly and more about which layer of the stack is seeing pricing power versus temporary order pull-forward. HPE’s surprise is not just top-line acceleration; it suggests enterprise AI capex is moving from pilot budgets into network refresh and integration spend, which tends to be stickier than server-only demand. The Juniper combination matters because it gives HPE a chance to bundle hardware, networking, and automation into procurement cycles, creating a modest but real moat expansion versus point-solution rivals.

Micron is the most interesting secular-vs-cyclical debate. The market is implicitly underwriting that AI memory demand will keep capacity tight long enough for margins to remain elevated, but memory suppliers usually destroy their own pricing power within 12–18 months once capex normalizes. The second-order risk is that hyperscalers, after securing near-term HBM supply, may push design changes that reduce per-unit memory intensity per chip, which would cap the upside even if AI compute keeps growing.

Palantir’s move is the cleanest example of software monetizing AI through workflow entrenchment rather than model ownership. The key takeaway is that customer growth is slowing relative to revenue, which implies expansion within the installed base is doing the heavy lifting; that is bullish for durability but also means the valuation depends on sustained net revenue retention staying very high. The market is likely underappreciating how much government demand can offset enterprise budget hesitation, but it is probably overpaying for the assumption that 50%+ earnings growth can persist without any compression in multiple.

Net-net, the closest thing to a mispricing is the market treating HPE as a one-quarter beneficiary rather than a multi-quarter infrastructure beneficiary, while treating MU as if the cycle has been permanently broken. The more probable outcome is that HPE and PLTR can compound longer than consensus expects, but MU remains the most vulnerable to any capex digestion or memory oversupply catalyst over the next 6–12 months.

Key Decisions for Investors

• Long HPE on a 3–6 month horizon: favor a starter position on any post-earnings consolidation, targeting a rerating toward low- to mid-teens forward P/E if order momentum persists; downside is limited if AI infrastructure spend merely decelerates rather than rolls over.
• Reduce or hedge MU after the recent run: use call spreads or trim outright exposure into strength; risk/reward is unfavorable because the stock is already discounting a very tight supply environment, while any HBM capacity normalization could compress multiples quickly over the next 2–4 quarters.
• Long PLTR but only via defined-risk structure: buy shares or call spreads on pullbacks, not momentum breakouts; the trade works if revenue growth stays above ~50% and large contract wins continue, but the valuation leaves little room for a guide-down or multiple compression.
• Pair trade: long HPE / short MU for 6–12 months to express ‘AI infra picks-and-shovels with durable backlog’ versus ‘AI memory cycle with supply-side reversion’; this isolates secular demand from cyclical pricing risk.
• For risk-managed exposure to AI software, pair long PLTR with a short of a lower-quality high-multiple enterprise software name if available in the book; the goal is to own the best revenue conversion while neutralizing sector beta and valuation compression.