Skylar CIO: AI & Power Story Will Play Out Over Years

A Bloomberg segment highlights a new ETF offering direct exposure to short-dated electricity futures in PJM and ERCOT, shifting risk from utility stocks to power-price moves. The pitch is that demand should rise as manufacturing reshoring, data centers, and the AI buildout expand load. Impact is likely more incremental for broader markets, but it may support fresh positioning in electricity price risk.

Analysis

This is more a volatility/pricing story than a broad "utilities" story. The cleanest winners are merchant generators with meaningful exposure to PJM/ERCOT prompt power, where revenue reprices faster than in regulated utility models; that favors names like CEG, VST, and NRG over XLU-style rate-base businesses. If the ETF attracts real assets, it can also become a marginal buyer of front-end power contracts, amplifying short-term spikes and widening the gap between prompt power and longer-dated expectations.

The second-order loser set is large electricity consumers that cannot fully hedge at the prompt curve: data centers, hyperscalers, and industrials with clustered load in Texas/Mid-Atlantic. For them, higher delivered power costs may compress EBITDA before they can pass through pricing, especially if growth assumptions were built on cheap and abundant power. Longer term, the more durable beneficiary may be grid equipment and interconnection-capex names rather than utilities themselves, because sustained load growth forces transmission and substation investment before it raises end-user rates.

The market is likely overestimating the speed of AI-driven load translation into realized cash flow. Most of the fundamental impact is 6-18 months out because queue constraints, transmission bottlenecks, and utility rate cases slow the pass-through; near term, weather and gas prices still dominate. The biggest falsifier is a benign summer with weak peak load and falling gas, which would deflate prompt power and expose the ETF’s roll/mean-reversion drag.

Key Decisions for Investors

• Pair trade: long CEG/VST/NRG basket vs short XLU for 1-3 months. Thesis is that prompt power and merchant margins reprice faster than regulated utility earnings; risk/reward is strongest if summer load or outage risk tightens PJM/ERCOT. Cover if front-month power rolls over after the peak season or merchant guidance stays flat.
• Add an alert, not a full position, on any new short-dated power ETF until AUM and daily volume are visible. If the vehicle remains subscale, the better trade is in the underlying generators; if it gathers assets quickly, it can create a tradable front-end squeeze in prompt contracts.
• Long ETN or HUBB over 6-12 months as a second-order beneficiary of load growth and grid reinforcement capex. This is a slower, higher-quality way to express the AI/data-center power thesis than buying the ETF itself; stop if utility capex commentary rolls over or interconnection spending gets delayed.
• Avoid shorting utilities indiscriminately on this theme. The risk/reward is better in merchant generators and grid suppliers than in rate-based names, which can eventually pass through costs and are less exposed to spot volatility.