Investing Lately? It's Tricky!

Main Street Alpha's Leo Nelissen argues that AI data-center investment remains a dominant growth driver and that fears of a slowdown are likely overblown. He expects hyperscalers to continue deploying capital and favors infrastructure, energy, and engineering companies including Comfort Systems (FIX), Vertiv (VRT), Quanta Services (PWR), and LandBridge (LB).

Analysis

The investable question is not whether AI remains a growth theme, but whether announced hyperscaler budgets convert into equipment orders and energized capacity. If spending persists, the nearer-term revenue transmission is strongest for cooling/power-management suppliers such as Vertiv (VRT), HVAC contractors such as Comfort Systems, and grid contractors such as Quanta Services (PWR). Grid connections, permitting, and power availability can become the binding constraint: that may extend project timelines while shifting value toward electrical infrastructure and thermal management rather than benefiting every AI-linked asset equally.

LandBridge (LB) is a less direct expression of this thesis. Any data-center or power-related upside should be treated as optionality unless supported by identifiable customer commitments and material revenue contribution; do not price it as equivalent to contracted equipment demand. The source is an analyst view, not independent evidence of order growth.

Near term, sentiment can move on hyperscaler capex commentary; over 1–3 months, order intake, backlog conversion, and guidance are the useful checks. Over 6–18 months, grid buildout supports structural demand, but project delays or a pause in compute spending could expose the cyclicality of supplier expectations. The contrarian risk is that the market may be right about timing even if wrong about the long-run AI opportunity: capex can remain high while deployments slip, utilization disappoints, and suppliers’ delivery schedules stretch.

Key Decisions for Investors

• Prefer a measured long bias in VRT and PWR over a broad AI-infrastructure basket, scaling in around earnings or hyperscaler capex updates rather than chasing theme-driven rallies. Reassess if order intake/backlog weakens or management lowers forward guidance.
• Treat LB as a watchlist/optionality position, not a direct data-center proxy. Before taking exposure, verify the revenue mix, any contracted power or data-center-related arrangements, and whether such activity is material to consolidated results.
• Track hyperscaler capex guidance, supplier order growth and backlog conversion, and utility interconnection or permitting delays over the next 1–3 months. A clear capex reduction, cancellations, or sustained deterioration in orders would falsify the near-term long thesis.
• Avoid assuming all infrastructure beneficiaries move together: delayed energization could defer contractor revenue even as demand for grid capacity rises. Favor companies with demonstrable order conversion over those whose case rests primarily on broad AI demand narratives.