Prediction: These 2 Stocks Will Be Worth More Than Palantir 5 Years From Now

Celestica's Q2 revenue rose 62% year over year, and management raised its full-year outlook, citing new program wins and improving customer forecasts. Marvell's custom AI silicon has grown to roughly a quarter of data center revenue; its fiscal 2027-to-2031 EPS is projected to increase from $4.22 to $27.77. The article sees Marvell as having greater potential to outgrow Palantir over five years, while Celestica's potential for roughly tenfold growth would require both sustained earnings growth and P/E expansion.

Analysis

The investable distinction is revenue visibility versus earnings-duration risk. Celestica’s networking and systems role can benefit across accelerator architectures, giving it some insulation if buyers diversify away from a single chip platform. But broader deployment does not guarantee attractive economics: program wins matter only if they convert to sustained shipments and margin improvement. Track gross margin, cash conversion, and customer/program concentration—not headline revenue growth alone. Marvell has more upside if custom silicon and optical demand compound, but its thesis is more exposed to design-win timing, hyperscaler insourcing, and customers shifting workloads or vendors. The cited EPS ramp is a forecast, not evidence of booked revenue; any multiple that capitalizes it far ahead of realized programs is vulnerable to estimate cuts. NVIDIA’s partnership may validate Marvell’s ecosystem position, but the disclosed investment does not establish the scale or profitability of future product sales.

Over days, AI-infrastructure sentiment may lift both names; that is not a fundamental entry signal. Over 1–3 months, watch Celestica’s conversion of new programs into margins and Marvell’s reported AI revenue/bookings against its ambitious trajectory. Over 6–18 months, scale-up networking and custom accelerators could broaden the addressable market, while customer concentration and architecture changes remain reversal risks. The contrarian point: investors may be over-crediting chip demand and underweighting the possibility that networking integration and deployment execution capture durable value—but Celestica’s cross-platform exposure is not automatically high-margin exposure. Falsifiers include weakening customer forecasts, stalled program ramps, margin deterioration at Celestica, or Marvell bookings/revenue that fail to support successive estimates.

Key Decisions for Investors

• Prefer a measured long in Celestica over chasing Marvell’s forecast-driven upside as the steadier infrastructure expression; add on weakness rather than a sentiment spike. Reassess if gross margin or cash conversion deteriorates despite growth.
• Keep Marvell on an earnings-confirmation watchlist rather than underwriting the long-range EPS projection today. Require reported AI revenue or bookings to convert across custom silicon and interconnect; cut the thesis if estimates rise while conversion or customer breadth fails to follow.
• Do not treat the NVIDIA investment or announced partnerships as proof of material near-term earnings. Verify product timing, customer concentration, and revenue contribution in filings and earnings commentary before sizing either position aggressively.
• Avoid a forced pair trade: relative valuation and current positioning are not provided. A sustained gap between Celestica’s program growth and margin conversion, or Marvell’s AI bookings and realized revenue, would be a more useful trigger for relative exposure.