Grid interconnection vs Behind-the-meter generation
With large-load interconnection queues at 3–7+ years in the top hubs and each energized gigawatt worth roughly $10–12B of AI revenue per year, speed-to-power now dominates the energy decision — which is why ~90 GW of behind-the-meter generation has been announced and gigawatt-class BTM gas plants are actually being built.
| Axis | Grid interconnection | Behind-the-meter generation |
|---|---|---|
| Speed to power | 3–7+ yr interconnection in top hubs (to ~10 yr worst) | gas turbines + gensets can energize in 1–3 yr; the whole point |
| Cost certainty | known tariffs / PPA structures; basis and shape risk manageable | fuel price exposure for the asset's life; hedging is on you |
| Regulatory risk | utility process — slow but well-trodden | air permits (6–36+ mo), co-location fights at FERC, local opposition |
| Reliability posture | grid + backup generation, the classic stack | islanded operation is a serious engineering program (Ch 4.8), not a procurement line |
| Carbon story | can contract 24/7 CFE, nuclear PPAs (>10 GW contracted) | gas BTM is a carbon liability; SMRs are the later-decade answer |
| Endgame | the destination — cheap, firm, someone else's spinning reserve | usually a bridge: island today, interconnect later, redeploy the turbines |
How the decision falls
The winning 2026 structure is rarely either/or: it is grid-plus-bridge or co-located hybrid — BTM generation to energize years early, an interconnection application running in parallel, and contracts that let the bridge redeploy. Underwrite the portfolio, not a single supply.
Full derivations, worked examples, and the numbers behind this matrix: Energy supply strategy: the four structures (Ch 3.4) · On-site and BYOP generation (Ch 3.5) · The power-bound era (Ch 16.1)