Why power is the binding constraint on the AI buildout
The AI buildout has made power — not chips, not capital — the gating factor on every hyperscale project. Generation, grid interconnection, and on-site backup are all being procured at the same time, by the same operators, on overlapping timelines. The scale of the demand is the story: analysts project data-center-driven electricity load could reach 90 GW by 2030, and data centers account for only about a third of total expected U.S. demand growth, with manufacturing reshoring and electrification making up the rest.[3] One longer-range estimate puts total U.S. data-center power demand on a path from 47 GW in 2025 to more than 176 GW by 2035.[13]
A record 86 GW of new U.S. utility-scale capacity lands in 2026 — the largest single-year addition since 2002. Solar and storage make up 79% of it. The grid, the gas-turbine supply chain, and the skilled-trades workforce — not financing — are now the binding constraints. The roles in shortest supply are experienced project managers, superintendents, field engineers, and the electrical trades that connect it all.
The full picture sits in data center power & energy news 2026, the grid-side mechanics in AI data center power, substation & grid coordination, and the federal-policy angle in the White House and tech giants on AI data center power costs. For the upstream picture, see the Data Center Construction guide and live announcements in the Data Center News hub.
The 2026 capacity wave
2026 is the largest U.S. energy infrastructure cycle in a generation, and the numbers make the scale concrete. Developers plan to bring 86 GW of new utility-scale generating capacity online in 2026 — the largest single-year addition in more than two decades, nearly double the 53 GW added in 2025.[1] For an industry that spent years adding capacity at a steady, predictable clip, this is a step-change driven directly by load growth.
The mix tells you where the work is. Solar leads at roughly 43 GW, battery storage follows at a record 24 GW, and wind contributes nearly 12 GW — meaning solar and storage together account for about 79% of all new capacity.[1] The 2026 buildout is, in effect, a portfolio answer to a single demand problem: gas for fast-track firming, nuclear and SMRs for long-term baseload, and an overwhelming volume of solar-plus-storage for the bulk of new megawatts. Each technology carries its own construction profile, its own permitting path, and its own workforce — which is why staffing an energy build now means understanding the whole stack, not one slice of it.
Where the capacity is landing
The buildout is not evenly distributed. Texas (ERCOT) leads on both solar and storage, helped by a fast interconnection process and abundant land; California continues to add storage to firm its solar base; and the PJM territory — the mid-Atlantic and Ohio Valley where much of the data center load concentrates — is where the grid strain is most acute. For recruiters and contractors, this geographic concentration means the same scarce leadership and craft talent is being pursued by clustered projects in a handful of markets at once.
Eighty-six gigawatts of new capacity is not an abstraction — it is thousands of concurrent projects competing for the same project managers, superintendents and field engineers. The constraint shifts from "can we finance it" to "can we staff and energize it," and the second question is harder.
Generation: gas, nuclear, renewables
The 2026 generation buildout is a portfolio answer to a single demand problem. Hyperscalers are signing power purchase agreements across the entire generation stack — gas peakers for fast-track capacity, nuclear and SMR for firm baseload, and renewables plus storage for the bulk of new megawatts. Each path is being pursued at once, because no single source can fill the gap on the timeline required.
Gas is back as the fast-track option
On-site and behind-the-meter natural gas has returned as the speed-to-power play, as in the Nevada data center natural gas approval. Turbines can in principle be sited faster than a grid connection clears, which is why gas keeps appearing in projects that need power before the interconnection queue will deliver it — though, as the next section covers, turbine lead times have themselves become a bottleneck.
Renewables carry the volume
Solar and wind are the bulk of new capacity, and the segment is drawing fresh construction entrants — like Eastern International's entry into wind power construction. The volume is enormous, but so is the competition for crews who can deliver utility-scale projects on schedule.
Nuclear anchors the firm-power story
Nuclear — covered in depth in the Nuclear & SMR Construction Workforce guide — is at the center of the firm-power conversation, offering carbon-free baseload that solar and storage cannot match for duration. It is the slowest path to build but the one hyperscalers increasingly want in the mix.
Gas turbines & the speed-to-power crunch
Gas re-emerged as the fast-track answer to AI load — and then the turbines themselves became the bottleneck. Only three manufacturers build large-scale gas turbines at volume: GE Vernova, Siemens Energy, and Mitsubishi Heavy Industries. All three are booked solid. Industry reporting puts wait times for new gas-fired turbine equipment at five to seven years depending on model and location, with delivery slots sold deep into the next decade.[9]
The order books make the scale vivid. GE Vernova's combined gas-turbine backlog and slot-reservation agreements grew from 83 GW to 100 GW in a single quarter in early 2026, and the company expects to reach at least 110 GW by year-end — with its CEO projecting turbine reservations sold out through 2030.[8] Siemens Energy's order book reached a record level on the same data-center-driven demand. This is why a project that needs firm power on a two-year horizon increasingly cannot simply order a turbine — the equipment lead time can exceed the entire rest of the build.
The same squeeze hits grid equipment. GE Vernova's electrification segment — which makes substations, switchgear, transformers and HVDC systems — saw orders roughly double year-over-year to $7.1 billion, with $2.4 billion tied to data center customers in a single quarter, more than all of the prior year combined.[10] The read-through for hiring is direct: the manufacturers are racing to add capacity (GE Vernova alone is adding roughly 1,800 U.S. production workers across 2025–2026), and the projects that depend on this gear are competing for both the equipment and the people who install and commission it.
Grid & transmission
The U.S. transmission system was not designed for hyperscale point loads, and the grid bottleneck is widely understood as the single biggest schedule risk facing operators. The equipment tells part of the story: high-voltage transformer lead times have stretched from about 140 weeks in 2023 to more than 160 weeks in 2026, making the gear itself a gating item.[5] New substations, interconnection upgrades, and cross-border power flows are all in motion to relieve the strain.
See the engineering recruiting angle in recruiting engineers for grid-scale projects, the international scale in GCC-Oman cross-border grid construction, and a major OEM bet on domestic capacity in Siemens' $1B U.S. energy infrastructure expansion. The read-through for hiring: substation and transmission expertise is now as scarce and as critical as the generation side.
The interconnection queue, explained
If one number captures why power is the constraint, it is the interconnection queue. Roughly 2,300 GW of generation and storage is stuck waiting to connect to the U.S. grid — more than the entire installed capacity of the country — with projects now averaging around five years in queue, and considerably longer in the busiest markets.[4] A project can be financed, permitted and shovel-ready and still wait years for permission to energize.
The queue exists because connecting new generation requires studies of how it affects grid stability, and the volume of applications has overwhelmed the system operators who run those studies. The practical effects ripple through every hiring decision on an energy build:
- It reorders the schedule. The grid connection, not the construction, is usually the critical path — so the people who manage the interconnection process and utility relationship are as valuable as those who build.
- It drives the behind-the-meter surge. On-site gas, solar-plus-storage, and increasingly nuclear are all ways to bypass the queue entirely, each spinning up its own workforce demand.
- It makes storage strategic. As the next section covers, batteries are increasingly used as a bridge to interconnection — letting a project energize years earlier than a traditional utility upgrade would allow.
Battery storage (BESS): the fastest-growing segment
Battery energy storage is the segment with the most explosive growth in 2026 — small, fast to deploy, and increasingly bundled with solar. Developers plan a record 24.3 GW of new utility-scale storage this year, up from 15 GW in 2025, pushing total U.S. capacity from about 44.6 GW toward 67 GW by early 2027; roughly 48% of installed storage is co-located with solar.[2] Longer term, BloombergNEF projects U.S. storage reaching 204 GW by 2035, with hyperscalers alone representing a ~20 GW BESS opportunity over that horizon.[7]
What is making storage strategic, beyond its growth rate, is its use as a bridge to interconnection. Aligned Data Centers struck a deal for a 31 MW / 62 MWh battery alongside a Pacific Northwest data center, sized specifically to let the facility interconnect years earlier than a traditional utility upgrade would allow.[6] Google went further, committing to a 300 MW / 30 GWh multi-day iron-air system paired with 1.6 GW of new renewables for a data center campus.[6] Storage has moved from backup power to a core grid and siting asset.
The deal flow is heavy and the projects are large. Start with battery storage construction recruitment, then the active projects: a 300 MW Arizona solar-plus-storage build, Georgia Power's BESS facility, Aypa Power's $1.5B financing, Spearmint's Texas project, Lydian's $689M solar-plus-storage round, and Arevon's $600M California project.
