TL;DR (read this first)

U.S. battery storage capacity: 26 GW end-2024 → 45 GW end-2025 (EIA) → 140 GW by 2030 → 200+ GW by 2035 (BNEF/WoodMac).

Most projects are underwritten on arbitrage alone (1× revenue) — ignoring the 70–85% uplift from stacking ancillary services, capacity markets, and curtailment primitives.

Winners like Jupiter Power (1.2 GW, ~2.8× uplift) and Broad Reach (1.5 GW, ~3.2×) are already capturing 3–4× returns using proof-of-curtailment logs.

This is the largest mispricing in global infrastructure: 200 GW built with 2021 assumptions, sitting on 2025 primitives that could unlock trillions in value.

The Numbers Are No Longer Forecasts — They're Signed Contracts

U.S. grid-scale battery storage (4-hour duration, lithium-ion focus):

YearInstalled Capacity (GW)Annual Additions (GW)SourceEnd-202426 GW10.4 GW (record year)EIAEnd-2025~45 GW18.2–19.6 GW (new record)EIA / BNEF2030140 GW~20 GW/year CAGREIA Central2035200–250 GWCumulative 8–10× growthBNEF / WoodMac That's enough stored energy by 2035 to power every U.S. home for ~1 hour during a blackout — and it's exploding due to IRA tax credits, renewables mandates, and data-center demand.

Global context: BNEF sees 92 GW added worldwide in 2025 alone (23% YoY), scaling to 2 TW cumulative by 2035. But the U.S. is the epicenter, with Texas and California claiming ~50% of 2025 additions.

The Revenue Waterfall: 1× vs. 4–8× Reality

A typical 100 MW / 400 MWh project in ERCOT or CAISO:

Revenue Stream% Captured Today (Most Projects)Realistic 2025 Capture (with Stacking)Uplift MultipleNotesEnergy Arbitrage90–100%100%1.0×Charge low, discharge high — baseline.Frequency Regulation (Reg-Up/Down)30–60%100%1.8–2.5×Proof-of-injection logs unlock full bid.Spinning/Non-Spin Reserves10–30%100%2–4×Curtailment primitives shine here.Capacity / Resource Adequacy0–20%100%3–6×Collateral-backed contracts (e.g., CAISO RA).Black-Start / Synthetic Inertia0%50–80%5–10×Emerging; logs as compliance proof.Total Stack Multiplier——4–8×BNEF: 20–30% IRR boost from stacking. Most developers pencil 6–9% unlevered IRR on arbitrage and stop. The rest? "Too complex" or "needs balance sheet we don't have."

That's leaving 70–85% of cash flow on the table — while proof-of-curtailment (Piece 3) turns "not using power" into $300–$2,000/MWh payouts.

Who's Already Capturing the Uplift? (2025 Leaders)

OperatorPortfolio (GW)Revenue Multiple vs. ArbitrageKey 2025 WinJupiter Power1.2 GW (Texas)~2.8×$286M financing for 300 MW/800 MWh projects (Tibbits MI, Tidwell TX); COD late 2025; sodium-ion pivot with Peak Energy (4.75 GWh by 2030).Broad Reach Power (ENGIE-acquired)1.5 GW (TX/CA)~3.2×$1B+ equity value acquisition (2025); 880 MW financed for ancillary/energy trading; debt against dispatch logs.Habitat Energy2+ GW (Europe → U.S.)~4×Revenue-share model via AI trading; ERCOT RTC+B prep (Dec 2025); stacking across day-ahead/real-time to minimize negative prices.Rev Renewables1 GW+ (hybrid focus)~3×Debt facilities tied to logs; 20–30% IRR from stacking per BNEF analogs. Jupiter's 2025 sodium-ion deal with Peak Energy? A hedge against LFP shortages, stacking curtailment for 2.8× uplift on new builds. Broad Reach's ENGIE buyout ($1B+ value) proves stacking scales to acquisition premiums. Habitat's AI dispatch? Maximizes real-time while dodging Causer Pays penalties.

Everyone else? Still 1× underwriting.

The Mispricing Is Galactic

200 GW financed assuming 2021 spreadsheets (arbitrage-only)
Capable of 4–8× with 2025 primitives: curtailment logs → tokens → collateral → stacked bids
BNEF: Stacking alone boosts IRR 20–30%. Add hyperscaler premiums (Piece 7)? Trillions unlocked.

Key Takeaways

U.S. batteries: 26 GW now → 45 GW end-2025 → 200+ GW by 2035 (8–10× growth)
70–85% revenue uplift from stacking — but most projects ignore it for "complexity"
Jupiter/Broad Reach/Habitat already at 2.8–4× via logs and AI dispatch
Proof-of-curtailment (Piece 3) is the enabler: turns events into $300–$2,000/MWh + tokenized assets
This is infrastructure's biggest arbitrage: build cheap, stack smart, capture forever
Consumers (next post) are already proving the model at garage scale

Background Reading

Next post:

Your Tesla Powerwall is already a better bond than Wall Street — earning 12–20% yields on garage hardware, proving the stacking model at consumer scale.

The Trojan horse is in your driveway.

Related Reading

Tesla Powerwall as Better Bond
The Forgotten Primitive: Proof-of-Curtailment
PoC Technical Specification

Energy & Finance

The 200 GW Battery Tsunami That Is Being Financed Like It's Still 2021