Developers are rapidly shifting to hybrid solar-plus-storage configurations as gas turbine supply constraints make new gas capacity a multi-year wait at rising cost
Decision Focus
Two converging supply-chain realities are forcing a structural change in how U.S. data center energy portfolios get built. Gas turbine lead times averaged roughly five years in 2025, and prices are forecast to reach $600 per kilowatt by 2027 from 2019 levels. At the same moment, solar plus battery storage is delivering utility-scale power within 18 to 20 months, at levelized costs between $54 and $82 per megawatt-hour in high-resource U.S. regions. For energy heads managing growth targets against a nearly 360% projected rise in U.S. data center power demand by 2030, the procurement calculus is shifting away from gas-first in ways already visible in hyperscaler contracting and developer consolidation.
90-Second Brief
In recent days, u.S. Developers are rapidly shifting to hybrid solar-plus-storage configurations as gas turbine supply constraints make new gas capacity a multi-year wait at rising cost. Wood Mackenzie projects U.S. Data center energy demand will reach 110 GW by 2030.
What Is Really Happening?
The gas turbine shortage is not a temporary disruption. It reflects a collision between surging AI infrastructure demand, re-industrialization pressure, and a manufacturing base that cannot quickly scale combustion turbine capacity. Five-year average lead times compress deployment planning beyond most data center construction cycles, and the expected price trajectory only worsens the option value of waiting.
Clean technology economics have also shifted materially. Battery installation costs have fallen 90% since 2010, solar 87%, and wind 55%, according to IRENA. Solar-plus-storage is therefore no longer just the fast option in high-resource regions; it is increasingly the cheap one. Primergy Solar, which supplies Amazon and Microsoft via VPPAs, notes that 500 MW of photovoltaics paired with 2 GWh of storage can be constructed in 18 to 20 months, with equipment accessible at construction-start timing in a way that gas turbines are not.
Developers are also beginning to layer in longer storage durations to close the firm-power gap that has historically constrained solar-storage as a baseload answer. Noon Energy’s agreement with Meta to deploy up to 1 GW and 100 GWh of ultra-long-duration storage — using reversible solid oxide fuel cell technology — targets system costs below $20 per kilowatt-hour, with a clean firm power goal of $60 per megawatt-hour. That cost target remains a developer projection; if confirmed at commercial scale, it would place dispatchable clean power at rough parity with natural gas across most U.S. markets.
The competitive dynamics are also reshaping developer ownership structures. Google’s $4.75 billion acquisition of Intersect Energy in December 2025 signals that the largest buyers are moving toward owning developer capability rather than contracting for output alone. NextEra Energy’s announced merger with Dominion Energy, combined with a 110 GW battery storage pipeline, positions it as a dominant integrated counterparty across both generation development and utility delivery.
Why It Matters for Global Heads of Data Center Energy
The 18-to-20-month solar-storage deployment window is now materially shorter than the 5-to-7-year grid interconnection queue in most U.S. markets. That gap has created a genuine behind-the-meter pathway, particularly in Texas, where favorable permitting and the ability to bypass the large-load queue allow co-located generation projects to align with data center construction schedules rather than waiting on them. BTM co-location is now a credible near-term alternative to queued grid capacity in certain markets, not a niche workaround.
PPA structuring is also evolving. Primergy’s recent refinancing of its Gemini project in Nevada — a 690 MW solar and 380 MW battery asset with a 25-year PPA with NV Energy — secured financing spreads described as among the lowest for a U.S. clean energy project in roughly a decade. For portfolios that have historically treated hybrid PPAs as higher-risk than gas-backed agreements, that financing trajectory warrants a fresh assessment.
For portfolios carrying 24/7 carbon-free energy commitments, ultra-long-duration storage matters beyond cost. Noon’s SOFC model and comparable technologies represent a potential path to firm clean power without waiting for new transmission — an accommodation the CFE commitment timeline often cannot make under current grid interconnection realities.
Forward View
Three fronts warrant active monitoring. First, whether ultra-long-duration storage economics arrive at commercial scale on the timeline developers are projecting. The 2028 demonstration date for the Meta SOFC project is a near-term data point, but commercial-scale cost confirmation will require operating history beyond that milestone. Second, whether Texas BTM permitting advantages persist as load growth accelerates and grid operators revisit the large-load queue structure. Policy conditions favorable in one market are inherently vulnerable to regulatory revision as that market becomes crowded. Third, whether the NextEra-Dominion merger creates counterparty concentration risk for portfolios that rely on NextEra as developer, financier, and utility across multiple regions simultaneously.
What Is Still Uncertain
Several variables remain unresolved. The $60 per megawatt-hour target for firm clean power from Noon Energy is a developer projection, not a confirmed delivered cost at scale. The gas turbine pricing forecast to $600 per kilowatt by 2027 reflects one analytical scenario; supply-chain relief or demand-side moderation could alter that trajectory. The Wood Mackenzie 110 GW demand figure carries scenario uncertainty at the upper range of AI infrastructure build forecasts. The IRENA LCOE range of $54 to $82 per megawatt-hour applies specifically to high-resource regions; U.S. markets with higher financing costs, grid connection charges, or permitting complexity sit above that band, as IRENA itself acknowledged.
One Question for Your Team
Given that behind-the-meter solar-storage in Texas now offers an 18-month deployment path against a 5-to-7-year grid queue in most primary growth markets, which sites in your current development pipeline have the load profile and land footprint to make BTM co-location a viable near-term alternative to waiting for queued interconnection capacity?
Sources
- Reuters — US solar-storage build spurred by gas plant waits (Link)
