The electricity bill for artificial intelligence is coming due, and no single fund fully captures it. U.S. data center consumption has jumped from 1.9% of national electricity in 2018 to 4.4% in 2023, and Lawrence Berkeley National Laboratory projects that share to reach between 6.7% and 12% by 2028. Owning that buildout requires three separate exposures: the hardware that moves electrons, the operators that generate and sell power, and the fuel that lights the marginal turbine.
Three ETFs help fully capture this theme. The First Trust NASDAQ Clean Edge Smart Grid Infrastructure Index Fund (NASDAQ:GRID) owns the equipment makers. The Utilities Select Sector SPDR Fund (NYSEARCA:XLU) owns the generators. The First Trust Natural Gas ETF (NYSEARCA:FCG) owns the fuel. Together they form a coherent thesis on the physical infrastructure driving AI compute.
Why the Grid Comes First #
Before a data center can even draw a gigawatt, someone has to build the substations, transformers, cables, and switchgear to deliver it. An individual hyperscale facility can consume more than a gigawatt of power (roughly equivalent to 750,000 homes), and PJM’s independent market monitor has already flagged data center load as the primary driver of tight capacity and rising prices in the largest U.S. grid. Currently, the physical bottleneck is a real constraint, favoring companies that sell the “picks and shovels” to utilities scrambling to add capacity (we have profiled seven of these suppliers, from power to cooling, in a free report you can grab here).
GRID: The Electrification Toolkit #
GRID reads like a directory of electrical equipment suppliers. Its largest positions are Eaton at 8.52% of net assets, Schneider Electric at 8.29%, ABB at 8.11%, Quanta Services at 8.10%, and Johnson Controls at 7.99%. That top-five concentration is the point of holding the fund. These vendors supply transformers, medium-voltage switchgear, transmission line construction, and building electrical systems for new data center campuses.
The fund extends into transmission operators such as National Grid at 4.11% and Italy’s Terna at 1.74%, plus power-generation equipment through GE Vernova at 1.06% and Siemens at 0.90%. That mix captures both the sale of a new HVDC line and the turbines feeding it.
Assets have expanded rapidly. Net assets stood at $7.65 billion at the end of March 2026 and reached $12.08 billion by June 30, reflecting inflows chasing the ongoing theme. And performance supports the narrative: GRID is up almost 19% year to date and roughly 29% over the trailing year. The tradeoff is heavy foreign exposure and cyclical industrial risk. A cooling in data center capex would hit these equipment vendors before it touched regulated utility revenues.
XLU: The Generators Selling the Kilowatt-Hour #
If GRID owns the toolbox, XLU owns the store. The Utilities Select Sector SPDR Fund holds the regulated and merchant utilities that collect checks from hyperscalers signing power purchase agreements. NextEra Energy dominates at 13.59% of net assets, followed by Southern at 7.47%, Duke Energy at 7.15%, and Constellation Energy at 6.11%. Constellation, the largest owner of U.S. nuclear generation, is arguably the cleanest pure play on hyperscaler contracts for around-the-clock carbon-free power. Vistra rounds out the top ten holdings at 3.36%, adding merchant power exposure that captures wholesale price spikes when data centers strain regional markets.
The fund’s pricing is hard to beat. XLU carries an expense ratio of 0.08%, roughly the cheapest way to own a diversified utility book. However, performance has been the quietest of the three: up about 2% year to date and roughly 3% over the past year. That understates the strategic role. Utilities offer dividend income, regulated returns, and slower moves in both directions, suiting investors who want AI power exposure without accepting industrial cyclicality. The catch is rate-case risk: political pressure on residential electricity bills, already elevated per the EIA, could crimp returns utilities earn on data center capex.
FCG: The Fuel Nobody Wants to Talk About #
FCG is the contrarian pick and the most direct expression of what gets burned to meet incremental AI load. The EIA’s Annual Energy Outlook 2026 projects that natural gas, solar, and wind together will supply roughly 80% of U.S. generation by 2050, with natural gas alone accounting for about 40% in the Counterfactual Baseline case. Gas is the only source that can keep a training cluster running around the clock at scale.
The portfolio is a concentrated bet on producers and midstream operators. Top positions include Western Midstream Partners at 4.91%, Hess Midstream at 4.83%, EOG Resources at 4.50%, ConocoPhillips at 4.33%, and Diamondback Energy at 4.03%. Appalachian gas pure-plays such as EQT at 4.02%, Expand Energy at 3.96%, Antero Resources at 3.50%, and Range Resources at 3.29% give direct leverage to Marcellus and Haynesville production feeding pipelines that supply new gas-fired plants.
FCG is far smaller than the other two, with net assets of roughly $590 million. Performance has been the strongest of the group: up nearly 36% year to date and about 41% over one year. But commodity risk is real. Henry Hub spiked to $7.72 per million BTU in January 2026 before settling to $2.89 by July, and producers remain exposed to that volatility. Investors get amplified upside on a demand story, with drawdowns to match.
How the Three Fit Together #
Each fund fails the AI thesis on its own. GRID captures the capex cycle but misses recurring electricity sales. XLU captures the meter while missing the machinery being installed to serve it. FCG captures the marginal molecule but leaves investors exposed to gas price swings and none of the electricity value chain downstream.
Owned together, however, they form a value chain. GRID benefits when utilities order equipment, which is happening now. XLU benefits when that equipment goes into rate base and earns a regulated return over the following decade. FCG benefits when new gas-fired capacity burns fuel to serve the load, the terminal state of the buildout. An investor bullish on hyperscaler electricity demand but unsure of timing across those layers gets diversification across the same thesis rather than a leveraged bet on one leg.
Weightings can tilt the risk. A conservative investor might overweight XLU for its 0.08% expense ratio and dividend base. A growth-oriented allocation would lean into GRID for equipment cycle exposure. FCG belongs as the smallest sleeve for anyone who accepts commodity risk in exchange for the highest torque to gas demand from data centers.
Contact [email protected] for any questions or corrections.