2 Hidden AI Infrustructure Stocks Up Over 100% in the Past Year

Comfort Systems (FIX) reported Q1 2026 revenue of ~$2.9B (+57% YoY) and net income of ~$370M (+119% YoY), alongside a record $12.4B backlog (+81% YoY). Vertiv (VRT) posted Q1 2026 revenue of ~$2.65B (+30% YoY) and net income of ~$390M (+137% YoY) as management cited AI-driven demand. Despite both trading at high earnings multiples (FIX ~49x vs VRT ~73x), Street consensus is “strong buy” for both with implied ~45%-50% upside, and the article favors FIX on faster growth and a lower valuation.

Analysis

The market is still underappreciating the split between "component intensity" and "installation intensity" in AI infra. Vertiv is a cleaner toll booth on ongoing capex, but Comfort Systems has a bigger operating-leverage wedge if hyperscaler projects keep converting from backlog into billable work; that makes FIX the higher-beta earnings surprise candidate over the next 2-4 quarters. The second-order winners are adjacent electrical and power contractors/suppliers such as ETN, PWR, and EME, because the binding constraint is increasingly grid interconnect and power distribution, not just cooling hardware.

The key risk is timing: both names are trading as if data-center demand is a multi-year straight line, but the first reversal usually comes from a pause in cloud capex plans, not a collapse in AI enthusiasm. If order conversion slows, FIX is more vulnerable than VRT because its backlog can look durable while gross margins get squeezed by labor, subcontractor, and schedule slippage. A 1-3 month catalyst path likely comes from next earnings and capex commentary from hyperscalers; a 6-18 month risk is customer self-supply or design standardization compressing margin premium.

My contrarian read is that the "cheaper" multiple on FIX may actually be the higher-quality setup if growth stays elevated, because the market is paying up for the more obvious winner and ignoring backlog durability. But if AI spend merely normalizes rather than accelerates, both stocks can de-rate fast because the valuation already embeds near-perfect execution. The thesis is falsified if data-center backlog stops growing, if FIX margin expansion stalls despite revenue growth, or if VRT guides to slower AI-related bookings growth on the next print.

Key Decisions for Investors

• Long FIX / short VRT as a relative-value pair for the next 1-3 quarters: FIX offers lower earnings multiple and stronger backlog momentum; exit if FIX underperforms VRT by >10% or if either company signals slower AI order conversion.
• Add to FIX on pullbacks rather than chasing strength; target a 2-4 month hold into the next backlog/earnings update. Risk/reward is best if the stock de-rates to the low-40s P/E without a corresponding backlog miss.
• Use VRT as the higher-quality momentum proxy only if hyperscaler capex commentary re-accelerates; otherwise avoid outright longs at ~73x P/E because multiple compression could offset fundamental growth.
• Watch ETN, PWR, and EME as second-order beneficiaries of AI power bottlenecks; they may offer better risk-adjusted exposure if the market shifts from cooling enthusiasm to grid and electrical spend.
• Set a stop/alert on any evidence of backlog conversion deterioration: if FIX growth stays high but backlog flattens or margins compress next quarter, the bull case weakens materially and the stock is vulnerable to a sharp re-rate.