What is speed-to-power?

Speed-to-power (or time-to-power) is the elapsed time from controlling a site to energizing it with firm power at the scale a data center requires. It has become a headline metric because, in the AI buildout, capacity that arrives in 2031 is worth far less than capacity that arrives in 2026 — and the grid, on its own, increasingly delivers the former.

Three queues compound:

• Interconnection — four to seven years to connect new load or generation in the major markets.
• Turbines — large gas units are quoted five-plus years out and have risen sharply in price as orders more than doubled.
• Transformers and gear — substation transformers now exceed a hundred and sixty weeks.

When demand needs power in two to three years and the grid offers five to seven, the bottleneck is no longer capital or land. It is time. Speed-to-power is the term for what is in shortest supply.

The fastest path is usually to bypass the grid:

• Behind-the-meter generation — building dedicated on-site power (often natural gas) so the facility does not depend on an interconnection queue.
• Bridging power — mobile and aeroderivative turbines that can be deployed quickly to energize early load while permanent power is built.
• Hybrid supply — pairing on-site generation with grid power and storage to balance speed, cost, and reliability.

These are structuring and procurement problems as much as engineering ones: securing the units, the gas, the permits, and the capital, in the right order. Mobile units such as the GE TM2500 are a common bridging tool precisely because they shorten the gap to first power.

Behind-the-meter generation delivers speed, but it adds fuel, emissions, permitting, and operating complexity. The right answer is rarely "all on-site" or "all grid" — it is the supply mix and deal structure that hits the target energization date at an acceptable cost and risk. That trade-off is exactly where development and capital structuring earn their keep.