AI Infrastructure Will Mint More Millionaires Over the Next Decade: 3 Stocks to Buy Right Now

The article highlights AI infrastructure demand—citing a projected 26.6% CAGR for the global AI infrastructure market (2026–2034)—and positions Marvell, Coherent, and Vertiv as key beneficiaries. It expects Marvell adjusted EBITDA to grow at a 44% CAGR (FY2026–FY2029), with Coherent also at a 44% CAGR (FY2025–FY2028) and Vertiv at a 38% CAGR (2025–2028). Valuations are described as roughly 43x (Marvell), 39x (Coherent), and 34x (Vertiv) current-year adjusted EBITDA, implying growth-led rather than “cheap” setups.

Analysis

This is less a pure AI demand trade than a bottleneck trade: the market is paying for whichever vendors sit closest to the physical constraints of higher rack density, not for “AI exposure” in the abstract. That argues for a hierarchy of winners. VRT has the cleanest claim on incremental spend because thermal and power constraints are hard requirements, while MRVL and COHR are more exposed to customer concentration, platform standardization, and price pressure as hyperscalers push suppliers to absorb more of the build-out economics.

Near term, the biggest catalyst is not AI enthusiasm but capex cadence. If hyperscaler spending holds into the next 1-2 earnings cycles, these names can grind higher; if capex pauses, they can re-rate down quickly because the market is already paying for multi-year growth. Over 6-18 months, the key risk is substitution: better chip efficiency can reduce some networking intensity, while in-house silicon and multi-vendor sourcing can cap MRVL’s margin power. COHR benefits from the copper-to-fiber migration, but that transition is lumpy and vulnerable to inventory digestion.

The consensus may be missing that “AI infrastructure” is not one trade. VRT is the most defensive way to own the theme, MRVL is the most exposed to design-win volatility, and COHR is the highest beta to rollout timing. This makes relative value more attractive than outright long beta at current valuations, especially if the market starts rewarding earnings quality over narrative.

Key Decisions for Investors

• Prefer VRT over MRVL on a 6-12 month basis; initiate a long VRT / short MRVL pair sized for a 15-20% spread target, with a stop if MRVL prints accelerating design-win commentary or VRT backlog growth slows.
• Do not chase COHR after a positive theme article; wait for a 10-15% pullback or confirmation from hyperscaler capex commentary before adding, because timing risk is the main driver of near-term downside.
• Use the next NVDA and hyperscaler capex updates as the catalyst monitor: if capex guidance or data-center power availability softens, trim AI-infrastructure longs by 25-50%.
• For investors needing directional exposure, favor VRT over a basket of MRVL/COHR for the next 1-3 quarters; the risk/reward is better because cooling spend is harder to defer than networking upgrades.
• Avoid buying MRVL as a core long here unless there is evidence of margin expansion in the next earnings print; the upside depends on sustained customer leverage, not just AI narrative momentum.