Behind-the-Meter Power for Data Centers: Why Gas Turbines Are Back in the Stack
Grid delays are pushing data center developers to evaluate on-site generation, but it is not a shortcut around power risk.
Behind-the-meter power for data centers means electricity generation or storage located on the customer side of the utility meter. For developers, the term now usually points to gas turbines, reciprocating engines, fuel cells, batteries, microgrids or hybrid systems used to support large loads when grid capacity is delayed, constrained or too expensive to secure on the required timeline.
The reason this is back in the development conversation is simple. AI demand is moving faster than utility infrastructure. Berkeley Lab's 2024 United States Data Center Energy Usage Report estimated U.S. data center electricity use could rise from 176 TWh in 2023 to 325 TWh to 580 TWh by 2028. JLL's 2026 Global Data Center Market Outlook estimates nearly 100 GW of new capacity globally between 2026 and 2030. Grid delivery cannot absorb that curve evenly.
Reuters reported in April 2026 that GE Vernova raised its outlook as data center demand lifted orders for gas turbines and grid equipment. Power Engineering reported that GE Vernova expects to reach 20 GW of annualized gas turbine output in 2026, with further expansion targeted by 2028. That is not a niche signal. It shows developers, utilities and hyperscalers are treating firm power equipment as a scarce input.
Why behind-the-meter power is getting serious again
Data center developers are evaluating on-site generation for three reasons.
First, grid interconnection timelines are too slow in constrained markets. A site can have land control, zoning support and customer demand but still wait years for transmission upgrades, substation equipment or utility approvals.
Second, AI workloads need firm capacity. Solar and wind can support procurement strategy, carbon goals and blended energy cost, but they do not by themselves solve 24/7 uptime for high-density compute. Batteries help with short-duration balancing. They do not replace long-duration firm generation at campus scale.
Third, power certainty now affects customer commitments. A tenant or hyperscaler may not care who owns the generation asset, but they will care whether the capacity, redundancy, emissions profile and operating cost are credible.
Behind-the-meter power can make a site more competitive if it bridges a delay, supports resilience or creates a phased path to energization. It can also make a site unfinanceable if the permitting, fuel, noise, emissions and operating assumptions are weak.
Gas turbines are a schedule hedge, not a magic answer
Gas turbines can provide large-scale firm power. That is why they are returning to the center of data center planning. But a turbine strategy creates its own diligence stack.
Developers need to answer:
Is there high-pressure gas access with enough capacity for full load?
Are firm transportation rights available or only interruptible service?
What air permits are required for normal operations and backup modes?
What are the local emissions limits for nitrogen oxides, carbon monoxide and greenhouse gases?
How will noise, visual impact and community concerns be handled?
Who owns, operates and maintains the plant?
What is the plan when grid capacity eventually arrives?
The last question is often missed. On-site generation may be temporary, permanent, backup-only or part of a long-term microgrid. Each case has different capex, contracting, accounting and customer implications.
The underwriting problem is now multi-variable
Traditional data center underwriting focused on land, power price, customer revenue, construction cost and exit value. Behind-the-meter power adds another layer. The developer now has to model delivered energy cost, heat rate, fuel escalation, outage risk, emissions compliance, maintenance downtime, reserve margins and grid interaction.
A power plant also changes the entitlement story. Residents who might tolerate a data center may resist a project that looks like industrial power generation. Local officials may ask whether the plant affects air quality, utility rates or climate targets. In some markets, the political risk will matter as much as the engineering.
That does not mean developers should avoid the strategy. It means they should underwrite it honestly. A gas-backed campus with a credible fuel path, permits, community package and transition plan may be stronger than a grid-dependent campus waiting for speculative upgrades. A poorly explained turbine plan can kill an otherwise strong site.
Where AI helps the power strategy
AI is useful because behind-the-meter diligence pulls from many disconnected sources. A development team needs utility filings, gas pipeline maps, air permitting rules, emissions thresholds, equipment lead times, public meeting records, tariff schedules, outage data and customer requirements.
An AI workflow can:
Track grid interconnection milestones and utility capacity signals.
Compare on-site generation options by scale, lead time, fuel requirement and permit path.
Monitor turbine, transformer, switchgear and engine procurement signals.
Extract air permit requirements from state and local rules.
Build a risk register that links power assumptions to schedule and capex.
Draft scenario models for grid-only, bridge generation and permanent microgrid cases.
The human layer remains essential. Power engineers validate the electrical design. Environmental counsel tests the permit path. Development leaders decide whether the political and commercial tradeoff is worth it.
What developers should evaluate before choosing on-site power
A behind-the-meter strategy should clear five gates before it becomes part of the investment thesis.
First, the power source must be technically capable of supporting the required load profile. Second, the fuel supply must be firm enough to match customer uptime obligations. Third, the permit path must be realistic, not assumed. Fourth, the economics must work after fuel, maintenance, emissions compliance and eventual grid integration. Fifth, the community story must be credible.
The strongest sites will not be the ones with the boldest power slide. They will be the ones where the power plan survives diligence from the utility, customer, lender, local government and operator.
Behind-the-meter power is back because grid scarcity is real. Treat it as infrastructure, not a workaround.