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

Hewlett Packard Enterprise, Micron, and Palantir all reported strong revenue growth and earnings beats tied to AI demand. HPE revenue rose 40% year over year with EPS of $0.79 versus $0.53 expected, Micron revenue nearly tripled to $24 billion with EPS of $12.20 versus $9 expected, and Palantir revenue grew 85% to $1.63 billion versus $1.54 billion expected. The article is broadly positive on AI-driven infrastructure and software demand, though it flags valuation risk for Micron and Palantir.

Analysis

The common thread is not “AI” as a theme, but a sharp reallocation of capex toward the infrastructure layers that monetize it fastest: networking, memory, and application software that sits closest to operating workflows. HPE and MU are the clearest second-order beneficiaries because their revenue inflects when hyperscalers and enterprise buyers move from pilot spending to deployment scale; that usually produces a burst of order growth before margins normalize. PLTR is different: it is less a capex beneficiary than a workflow capture story, where each additional deployment expands switching costs and makes customer growth understates expansion in spend per account.

The market is likely underappreciating how this cycle feeds back into suppliers and peers. If HPE’s cross-sell momentum persists, Juniper becomes less about standalone networking upside and more about forcing rivals to discount on integrated deals; that can pressure Arista, Cisco, and white-box OEMs over the next 2-4 quarters. In memory, the key marginal swing factor is capacity discipline: if Samsung or SK Hynix opens the taps, MU’s earnings power can compress quickly even if unit demand stays healthy, because the stock is now pricing a much longer-duration AI shortage than the industry usually deserves.

The main contrarian point is that consensus may be conflating strong demand with durable slope. MU at a richer multiple is the most vulnerable to a narrative reset because earnings are levered to price rather than just bits shipped; that makes it more of a trade than a compounder at current levels. PLTR, by contrast, can stay expensive longer than skeptics expect if it keeps converting large accounts, but the risk is valuation compression if revenue growth decelerates even modestly from current hyper-growth levels. For HPE, the setup is best over the next 1-2 quarters as guidance and backlog conversion can outrun sentiment, but less attractive once the market starts capitalizing 2028 targets before execution is visible.

Key Decisions for Investors

• Long HPE on a 1-3 month horizon into the next print; use pullbacks to add, with upside driven by order-to-revenue conversion and guidance revisions. Favor a call spread over outright equity if you want defined risk, since the multiple is still low enough that rerating can happen quickly.
• Short MU or buy put spreads 3-6 months out against strength if the stock continues to price in a permanent memory supercycle. The asymmetry is negative here: a modest supply response or AI capex pause could compress the multiple faster than earnings can grow.
• Long PLTR only on dips, preferably via call spreads or a staggered equity entry over 2-4 weeks, because the thesis is durable but valuation leaves little room for execution misses. Best reward/risk is if management can prove sustained net expansion in large accounts rather than just top-line beat-and-raise.
• Pair trade: long HPE / short a networking peer basket on a 2-quarter view if you believe integrated AI infrastructure deals are taking share. The edge is the acquisition-driven bundling effect, which can improve HPE deal size faster than competitors can match without margin sacrifice.
• For more aggressive exposure, express the AI hardware thesis via NVDA/TSM rather than MU when looking 12+ months out; they capture structural demand with less direct exposure to memory pricing volatility.