Guggenheim's Schwartz Says US Power Crunch Hurts AI Push

Alan Schwartz warned that the US could fall behind in artificial intelligence development if the electricity grid is not upgraded fast enough. The comment highlights infrastructure constraints as a potential bottleneck for AI expansion, but it contains no company-specific numbers or immediate policy action. Market impact is limited and mostly sentiment-driven for AI and power infrastructure names.

Analysis

The bottleneck here is less about AI demand and more about the capex plumbing needed to monetize it. If grid expansion and interconnection timelines stay stretched, the marginal beneficiary is not just utilities but the entire “picks and shovels” stack: transformers, switchgear, gas turbines, cable, and power-management software. That shifts value away from pure software winners toward the industrial names that can turn backlog into revenue over the next 12-36 months.

The second-order risk is that AI growth becomes self-throttling in certain geographies, especially dense data-center markets where power availability is already the binding constraint. That creates a relative advantage for operators with captive generation, on-site power agreements, or faster permitting exposure, while slowing down hyperscaler rollouts in constrained regions. It also raises the odds of utility rate-base expansion and tighter regulation around who gets priority access to incremental load.

Consensus likely underestimates how positive this is for natural gas and distributed generation over the next 1-3 years. If grid upgrades lag, developers will increasingly bridge with gas peakers, microgrids, and behind-the-meter solutions; that supports equipment suppliers and midstream gas infrastructure even if headline AI sentiment cools. The contrarian angle is that the market may be too focused on GPU scarcity and not enough on power scarcity—the latter is more durable, less cyclical, and harder to solve with capital alone.

Tail risk is a policy response: accelerated permitting, DOE support, or emergency grid upgrades could compress this thesis faster than expected, particularly if AI becomes a national competitiveness issue. But those remedies are slow to execute, so near-term downside is more about project delays and capex repricing than outright demand destruction. Over the next two quarters, watch for utility interconnection queues and hyperscaler guidance on power-constrained deployment as the key catalyst set.

Key Decisions for Investors

• Long VRT / ETN basket on pullbacks, 3-12 month horizon: play the infrastructure bottleneck through electrical equipment and thermal management beneficiaries; risk/reward is attractive because order books are already extending before revenue catches up.
• Long NEE or other regulated utility with data-center load exposure, 12-24 months: rate-base expansion and contracted power demand should support mid-teens total-return potential, with lower volatility than pure AI names.
• Pair trade: long CEG / short AI software basket for 6-12 months: if power becomes the constraint, operators with firm generation and merchant pricing capture the economics while software multiples remain vulnerable to slower deployment cadence.
• Long KMI or other gas midstream exposure, 12-36 months: use as a delayed beneficiary of behind-the-meter gas buildout and incremental gas-fired generation; upside is steadier cash flow re-rating rather than growth multiple expansion.
• Buy call spreads on CMI or GTLS into any grid-upgrade policy headlines, 3-6 months: these names can rerate quickly on backlog visibility, but call spreads cap premium outlay in case policy remains slow-moving.