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

Comfort Systems reported Q1 2026 revenue of ~$2.9B (+57% YoY) and net income of ~$370M (+119%), alongside a record $12.4B backlog (+81% YoY). Vertiv posted Q1 2026 revenue of ~$2.65B (+30%) and net income of ~$390M (+137%), citing AI-driven demand for power and cooling capacity. Both stocks have strong analyst support ("strong buy" consensus) with price targets implying ~45%-50% upside, while Comfort Systems screens cheaper on ~49x P/E vs Vertiv’s ~73x as AI infrastructure spending translates into rapidly growing earnings.

Analysis

Near term, the market is likely to keep rewarding the names that look most levered to the AI capex chain, but the second-order issue is capacity: the real constraint is not demand, it is execution. FIX is more exposed to project timing and labor bottlenecks, so its upside depends on converting backlog without margin leakage; VRT has a cleaner product pull-through and should monetize the install cycle faster, but that also leaves it more exposed to any pause in hyperscaler orders.

The less obvious winner set is the electrical/power ecosystem around these builds: ETN, HUBB, NVT, and specialty contractors with data-center-qualified field teams. The loser is generic commercial MEP work, where talent and equipment get bid away; that creates a margin headwind for non-AI exposed peers even if top-line growth looks fine. If AI build-outs stay hot for 1-3 quarters, this becomes a labor and lead-time story as much as a demand story.

Contrarianly, the bullish consensus may be underestimating how quickly sentiment can swing on a single hyperscaler budget reset. Both names are priced for continued beats, so the risk/reward is asymmetric if backlog growth decelerates or margins normalize; a 10-15% multiple compression would not be surprising on merely in-line execution. Over 6-18 months, the structural winner is the company with the stickiest service/aftermarket mix, not necessarily the cheapest multiple.

The key falsifier is any sign that data-center capex growth moderates below expected levels or that book-to-bill rolls over for two consecutive quarters. If that happens, the market will stop treating backlog as visibility and start treating it as a lagging indicator.

Key Decisions for Investors

• Relative value: long FIX / short VRT for 1-3 months if you want to own the faster-growth, cheaper multiple name; target 15-20% spread compression, but cut if VRT reaccelerates on better AI order intake or FIX margin guidance softens.
• Wait for a pullback to add FIX rather than chase it after the run; the entry matters because the stock already discounts several quarters of backlog conversion. Falsify the thesis if organic margin expansion stalls despite backlog growth.
• Use VRT as the higher-quality AI infrastructure hedge only on weakness, not after strength; its valuation leaves less room for disappointment. Risk/reward improves if the stock de-rates into the low-60s forward P/E range.
• Basket trade: long a data-center infrastructure basket (FIX, VRT, ETN, NVT) vs short a broad industrials ETF if hyperscaler capex remains firm over the next 1-2 quarters; this isolates AI capex share gain from broader cyclical exposure.