A ~$100B, decade-long program in which the companies building the desert’s future use their own power build-out to refill the river they depend on, and are seen leading the fix instead of fighting it.
The full stack The thesis The first stepFill the stranded transmission at the Colorado River dams with solar and storage → that power backfills the dams, runs the data centers, and manufactures water → the water and the goodwill refill the basin.
The Colorado River is short about 1.5 million acre-feet a year and its rulebook expires in 2026. The hyperscalers are pouring $100B+ of water-hungry infrastructure into Arizona and Nevada at exactly the wrong moment, and the backlash has started. But they hold something no one else does: the balance sheet, the 2030 water pledges, and a reason to act, and the desert throws off more sun than they can use. Energy and water are substitutes at the margin. Cheap, abundant, carbon-free power is the bridge between what these companies need and what the basin needs.
Full derivations, the interactive model, and an adversarial technical review are at the evidence site. The short version: the crisis is an accounting problem as much as a hydrology one, and every current proposal prices evaporation while ignoring groundwater and the ecological cliff.
Not one mega-project, a stacked program where the energy layer pays for itself and the companies’ own build-out powers the rest. Illustrative capital over ten years:
Fill ~3.5 GW of idle interconnection at Hoover, Glen Canyon, and the retired Navajo Generating Station’s 500-kV lines to Phoenix and Vegas. Scale toward ~12–15 GW. ~17→50 TWh/yr of clean power via existing wires.
Wastewater reuse-to-recharge (the Orange County model), brackish and ag-drainage desalination, and aquifer storage, 1.5–3 MAF/yr of drought-proof water, powered by layer 1.
A verified consumptive-use conservation market ($400/AF), a public measurement ledger, and the Glen Canyon temperature retrofit, the honest core that survived adversarial review.
FERC surplus-interconnection to reuse the wires, batteries to firm the variable solar, and selective new transmission where the headroom runs out.
Navajo and Hopi jobs, revenue, and water infrastructure on the reclaimed coal land the buildout reuses, the goodwill core, not an afterthought.
~$100B over a decade. Layer 1 is investable and largely self-funding. Layers 3–5 are public-good and grant/coalition-funded. The split is the point (see Financing).
As the dams derate, the transmission built for them strands ~3.5 GW of capacity pointed at Phoenix and Las Vegas. Surplus interconnection puts solar+storage on it in months, not the 5+ year queue.
Frontier did it for carbon, LEAF for forests. The coalition pre-buys verified, additional, shepherded consumptive-use reduction at a floor price, the first AMC for water, the demand signal that makes the market real.
After backfilling the dams and powering the data centers, the surplus electrons desalinate, treat, and recharge, reuse-led and drought-proof. Energy stops being the constraint.
Batteries smooth the day, not the season. Recharge and pumping bank summer’s solar surplus underground as water, covering winter’s deficit no battery can.
Utility-scale solar on federal land usually takes 3 to 7 years, most of it environmental review. This program has a shortcut most developers don’t: build on the already-disturbed Navajo Generating Station and Kayenta mine land, under tribal jurisdiction, using surplus interconnection. That combination sidesteps the three things that kill timelines.
| Ownership model | Best for | The hard part |
|---|---|---|
| Private IPP + hyperscaler PPA | speed, capital, standard structure | land control + interconnection risk sits with the developer |
| Tribal (NTUA or joint venture) | sovereignty, the Navajo Nation’s existing 500 MW of NGS transmission rights, IRA/BIL tribal funds, goodwill | a “regulatory jumble” (up to 49 steps), tax-credit access, fractionated land |
| Federal (BLM right-of-way) | vast public land, WAPA transmission access | NEPA 18–36+ months, fees, multi-agency |
| State trust land (AZ) | 9M+ acres, state streamlining orders | public-auction leasing, revenue duty to schools |
| Utility (APS / SRP / NV Energy) | rate base, they already own the wires | slower, regulated, and they’re the incumbents being displaced |
The critical path is NEPA plus grid interconnection. The single biggest bottleneck is an Environmental Impact Statement (24–48 months, sometimes far longer) stacked on top of Section 106 tribal-consultation (no statutory deadline) and a 5-year interconnection queue. Three moves collapse it:
Building on the former mine and plant footprint drops the review from a full EIS toward a ~9–15 month Environmental Assessment (or a categorical exclusion), with less opposition and existing roads and grid, plus the IRA “energy community” 10% bonus.
FERC Order 2023 lets the project use the derated dam’s or NGS’s existing interconnection headroom on an expedited path, skipping the 5-year queue when no new network upgrades are needed.
On Navajo trust land, the Nation’s own process and its 500 MW of transmission rights give a willing owner and a faster path than greenfield BLM, the Kayenta solar facility is the proof.
Where public land is used, the 2024 Western Solar Plan’s pre-screened zones get the streamlined EA, not the full EIS.
Realistic outcome: the fast-path combination can bring first projects online in ~2–3 years instead of 5–7. The ESA / desert-tortoise consultation (~6 months) and honest tribal engagement are the parts you cannot rush, and shouldn’t.
| Dimension | Result | Note |
|---|---|---|
| Verified water freed (conservation core) | ~0.6–1 MAF/yr | additional, shepherded, at ~$400/AF |
| Water made/saved (reuse + desal + efficiency) | 1.5–3 MAF/yr | reuse-led, drought-proof. Bounded by capture & brine, not energy |
| Clean energy through existing lines | 17→50 TWh/yr | backfills the dams, powers the data centers, runs the water |
| Dam operations | decoupled | releases freed for temperature & ecology, no power penalty |
| Jobs | ~28,000+ | construction job-years + permanent O&M, Native-hire priority |
| Carbon avoided | 3–7 MtCO2/yr | displacing market gas. More as it scales |
| Goodwill | the narrative flip | from “draining the desert” to “refilled the river,” on a public ledger |
A program this size makes losers. Naming them is how it survives contact with politics.
Hyperscalers (license to operate + goodwill), the Navajo & Hopi Nations (a real just transition), farmers paid to conserve, cities that get drought-proof local water, the utilities (APS, SRP, NV Energy, TEP) who own the wires and can build the solar, and the Grand Canyon’s fish.
Merchant gas peakers and gas suppliers (the real opposition). Imported-water wholesalers (MWD, CAP) whose sales shrink. Senior ag rights (IID) whose water loses scarcity value. Queue-stuck solar developers. Seat them as builders and offtakers, or they fund the campaign against it.
Full winners-and-losers analysis, with magnitudes, on the evidence site.
| Layer | How it’s funded | Cost of capital |
|---|---|---|
| Energy (PV+BESS) | project finance + infrastructure equity, anchored by data-center PPAs. Investable | market IRR |
| Water production | revenue bonds + WIFIA + corporate offtake (the AMC) | ~5% |
| Conservation + ledger + dam | coalition grants (the no-revenue public-good core) + a small ~$18/AF beneficiary assessment | grant |
| Just transition | coalition + IRA energy-community 10% bonus + tribal partnership | grant + credit |
The hyperscalers’ role is a coalition Advance Market Commitment plus grants for the public-good core, on the order of $10B of commitments across five or six trillion-dollar companies over a decade, a rounding error against the $100B+ of basin capital it protects. The energy layer largely funds itself. Blue bonds and WIFIA carry the revenue tier. The coalition grants what has no cash flow. No new federal appropriations required.
OpenET (with Google in it) is the measurement layer. VWBA is the accounting standard. Orange County has run reuse-to-recharge for 23 years, cheaper than imports. The ICS mechanism already banks conserved water in Lake Mead. We connect them into the first water AMC.
We handed the plan to simulated water-tech VCs, Xylem engineers, and the basin’s hydrologists to destroy it, then cut everything that didn’t survive, the inflated yields, the fake 54% IRR, the paper water. What’s left is smaller and true.
Two honest limits we lead with: shepherding conserved water to Lake Powell is legally unsolved, so we start in the Lower Basin where it already works. And the water numbers are bounded by brine disposal, capture infrastructure, and capital, not by energy or optimism.
An advance market commitment is a demand signal, not a capital stack. Frontier’s $1B and LEAF’s $1.5B don’t build anything, they promise to buy output so someone else finances it. Here the AMC underwrites one tier: ~$3–4B of conservation offtake. The other ~$95B is project finance (the energy layer self-funds off data-center PPAs), muni and WIFIA debt (water), and coalition grants (public goods). Anyone who knows finance and hears “finance $100B with an AMC” will dismiss it, so we don’t say that.
Five things have to be true for the coalition to form:
Our own honest read: a funded pilot is genuinely likely. A full basin-scale AMC over a decade is possible but gated on the five above. “The whole $100B via AMCs” is not how it works and shouldn’t be pitched. The energy economics carry their own weight. Water is the goodwill overlay.
We ran the proposal past twenty opponents and experts. Four objections landed hard. Here they are, answered straight.
1. The transmission is no shortcut, and public-land solar faces a hostile federal posture. True on both counts. The retired Navajo Generating Station 500-kV lines are co-owned by APS, SRP, TEP, NV Energy, and Reclamation, and a 12 to 15 GW build needs real interconnection studies and upgrades, not a plug-and-play claim. And with the current administration cool to solar on federal land, greenfield BLM permits are a poor bet. That is exactly why the plan leads with tribal and reclaimed coal land, which is sovereign or private rather than BLM, and reuses existing transmission under multi-year owner agreements. Phase one is sized to what can actually be studied and built.
2. The water cost stack was understated. Agreed. We drop the single blended dollar-per-acre-foot and unbundle the delivered cost by type: conservation near $400/AF, reuse $1,500 to $2,200, brackish desal near $900 to $1,100, seawater exchange $2,800 to $4,000, plus conveyance, brine, and debt service. Priced like infrastructure, not a slogan.
3. Brine disposal is the real inland-desal constraint, and it is solvable where the geology allows. One million acre-feet of desal product makes roughly 170,000 acre-feet of brine. The proven template is El Paso’s Kay Bailey Hutchison plant: deep-well injection 22 miles out into isolated 4,000-foot formations, 30,800 acre-feet a year of fresh water at about $489/AF. So brine is a core line item, and inland desal is capped to sites with a credible injection sink, not claimed everywhere.
4. Manufactured water is not yet a basin credit, but the window to fix that is open. Reuse and desal water is a creature of state law. Today it reduces a state’s own demand and cannot float as an interstate credit without a new accounting rule. The honest move is to say so, and to make that rule one of the coalition’s policy asks. The post-2026 renegotiation of the operating guidelines is the once-in-a-generation opening to establish crediting for verified new supply, and it is far more winnable now, mid-crisis, than it has ever been.
Fund the measurement-ledger pilot and a first verified conservation tranche in one Lower-Basin sub-basin, the branded, public proof layer.
Co-develop one PV+BESS project on a derated-dam or NGS interconnection via surplus interconnection, prove the wires, the water-load pairing, and the tribal partnership.
Set the integrity standard together, additionality, shepherding, uncertainty holdback, so the coalition defines a real water acre-foot, not just buys one.
Scale on proof. If it verifies clean, the anchors bring the full coalition and the stack builds out over the decade. If not, you’ve spent a pilot budget and learned something real.
The coalition terms The AWS-specific brief The full evidence