Datacenters driving US clean energy growth while still threatening climate

Datacenter-driven electricity demand is accelerating growth across clean energy, with the iShares Global Clean Energy ETF up about 52% over the last year after falling roughly 80% from late 2021 to early 2025. Demand is benefiting utility-scale solar, batteries, and fuel cells, including Nextpower’s 20% YoY growth, Bloom Energy’s 1,338% stock rally, and large projects from Google, Microsoft, Oracle, and DTE Energy. The article also flags material caveats: grid delays of up to 12 years, continued fossil-fuel buildout, and the risk of an AI demand bubble.

Analysis

The important second-order effect is that datacenter load is turning clean energy from a policy-driven market into a time-to-power market. That flips the competitive edge toward developers and equipment vendors that can deliver behind-the-meter or near-term capacity, even if the marginal electron is not pure renewables. In practice, the bottleneck is not ideology but interconnection and execution, which means the winners are increasingly those with permitting, manufacturing, or storage flywheels rather than pure-play project developers dependent on grid queues.

This creates a bifurcation inside the clean stack. Utility-scale solar, batteries, and fuel cells tied to enterprise customers should see demand elasticity improve over the next 12-24 months, while residential rooftop solar remains relatively disconnected from the AI capex cycle and may continue to lag. The same dynamic also raises input-chain pressure on transformers, switchgear, gas turbines, and pipeline capacity, where pricing power can persist even if overall clean-energy sentiment cools.

The market is likely underestimating how much of the current move is a capex displacement trade rather than a secular decarbonization trade. If AI spending slows, the visible growth rates in some clean names could compress quickly because the customer base is concentrated and project timing is lumpy; that argues for favoring businesses with manufacturing backlog and service content over pure order-book optionality. The counterpoint is that the grid congestion problem itself is a multi-year secular constraint, so the demand for local power solutions may persist even through an AI equity reset.

Key Decisions for Investors

• Long BE into the next 6-12 months on continued datacenter fuel-cell adoption; risk/reward is attractive while deployment speed remains the gating factor, but trim if the stock rerates to fully price 2026 capacity expansion.
• Long GOOGL and MSFT on a 6-18 month horizon as the most likely self-suppliers of grid-constrained load; use pullbacks to add, since vertical integration should reduce dependency on utility delays and improve power optionality.
• Long DTE as a structural beneficiary of customer-funded storage buildouts; the thesis is 12-24 months of regulated asset growth, but size modestly because utility approvals can slow the multiple expansion.
• Pair trade long BE / short a basket of residential solar names if available, because datacenter-driven demand is concentrated in utility-scale and onsite power, not rooftop; expect relative outperformance over the next 2-4 quarters.
• Avoid chasing broad clean-energy ETFs after the recent rerating; instead use any AI-related drawdown to buy quality equipment and storage names, as the market is likely overpaying for headline growth but underpricing backlog visibility.