A way to close AWS’s water-positive gap in the Colorado River basin with the highest-integrity replenishment available, verified, additional, legally shepherded to Lake Mead. Already stress-tested against the objections your own engineers would raise.
The offer What your team will ask The first stepAWS is 75% of the way to water-positive by 2030, up from 53% in 2024. The remaining quarter is the hard part, and you are building major capacity in Mesa, Laveen, and Phoenix, in the middle of the worst Colorado River hydrology on record.
The credibility bar on replenishment is rising at the same time. “Paper water” and weak-additionality credits are getting torn apart in public, and Arizona is starting to say no, Tucson rejected Project Blue until it went zero-water, and municipalities are imposing caps. The last thing you need is replenishment volume your critics can pick apart, or a basin moratorium that stalls a campus.
What follows is not another offset catalog. It is a way to buy real, defensible water at scale, in the basin where it matters most to AWS, and to be seen leading the fix rather than negotiating against it.
Anchor a verified consumptive-use conservation market: pay farmers to retire real, additional water use, shepherd it into Lake Mead under legal protection, and settle only on volumes measured with an uncertainty holdback and ground-truthing.
What AWS gets that a scattered offset portfolio can’t give you: acre-feet that survive an auditor and a journalist, at a lower cost than any new supply, in the state where your buildout faces the most water risk, plus your name on the public ledger that proves the whole basin got measurably better, not just your own gallons.
The difference between this and a low-quality credit is four disciplines, baked into every acre-foot before it is sold:
1. Consumptive use, not diversion. We pay only for water actually taken out of the system, not efficiency “savings” that were return flow someone downstream already used. That distinction is where most water credits quietly fail.
2. Additionality. Only reductions that would not have happened anyway. Fallowing that was coming regardless doesn’t count.
3. Shepherding. The retired water is legally protected and delivered to Lake Mead or Powell, not left for the next diverter to pick up.
4. Honest measurement. Satellite ET (OpenET) flags anomalies but does not settle payments, its field-scale error is too high. Settlement is on metered flows and water-rights accounting, with a 25% uncertainty holdback on estimated volumes and funded flux-tower ground-truthing. The ledger is verification, not a black box.
Yes, in pieces. That’s the point, not the problem.
Your team already funds most of the ingredients. The measurement layer is OpenET (NASA, DRI, EDF, Google). The accounting standard is VWBA 2.0, which AWS-peer stewardship programs already use. Paying to conserve water into Lake Mead already works through the Lower Basin’s ICS mechanism. Corporate basin money already flows through Bonneville’s “Change the Course.” None of that is speculative.
What has never been built is the one instrument that ties them together: an Advance Market Commitment for water. Frontier did it for carbon, LEAF for forests, a forward demand signal that pulls high-integrity supply into existence. No one has done it for water, anywhere. That is the new thing AWS would be anchoring, and it is why a coordinated commitment beats another year of scattered per-site offsets.
Two honest caveats we lead with, not bury: shepherding saved water to Lake Powell is legally unsolved, so we start in the Lower Basin where conservation already stays in Lake Mead. And some major funders have cooled on water markets, which is exactly why integrity and a demand anchor, not enthusiasm, carry this.
We ran this proposal through a simulated adversarial review, written in the voices of water-tech investors, membrane-process engineers, and basin hydrologists, told to find every fatal flaw. Then we rebuilt it around what they found. Here is the short version of that stress test, so your team’s first questions already have answers.
| The objection | Our answer |
|---|---|
| “Saved water isn’t deliverable, the river’s not a pipe.” | Correct. AWS doesn’t receive wet water. It funds a verified reduction shepherded to system storage and takes the replenishment credit. No claim of private delivery. |
| “Efficiency savings are double-counted return flow.” | We pay for retired consumptive use only, not efficiency upgrades. That’s the whole design. |
| “OpenET has 10–20%+ error, you can’t settle money on it.” | Agreed. ET flags anomalies. Settlement is on metered flows and rights accounting, with a 25% holdback and ground-truthing. |
| “Tamarisk removal and cloud seeding don’t yield water.” | Removed from the water math entirely. USGS and the Wyoming trials are clear. They’re habitat/research line items, not supply. |
| “54% IRR is a fantasy for water infrastructure.” | Gone. Conservation returns a real benefit-cost above 2 at ~$400/AF. The capital-heavy public goods are funded as grants, not sold as equity. |
| “This reads as Big Tech buying the river.” | Which is why it leads with visible, local, community and tribal benefit, on a public ledger, conservation-with-community-benefit, in the open. |
The full seven-objection review and the before/after changes are documented and available to your team on request.
We are deliberately not showing you a headline that collapses under scrutiny. Here is the real shape of it:
| Metric | Value | Note |
|---|---|---|
| Conservation anchor cost | ~$400 / verified AF | Cheapest high-integrity replenishment available |
| Conservation benefit-cost | 2.15 | Water at $1,500/AF avoided supply, 5% discount, 30 yr |
| Full package capital | ~$15.7B | Ledger + conservation + ag + Glen Canyon + recharge |
| Verified water, full package | ~1.0 MAF/yr | ~0.6 MAF/yr shepherded to storage. Balance is regional recharge |
| Public-good pieces | grant-funded | Dam retrofit, habitat, no fake private return |
The conservation market is where AWS money does the most good per dollar and carries the least risk. The rest of the package is what the basin needs, financed by public and philanthropic capital, AWS doesn’t have to underwrite it to lead it.
Nobody signs a basin-scale check on a slide. So the ask is a pilot that proves the verified acre-foot before any large commitment.
Co-fund the measurement-ledger pilot in one well-instrumented Arizona / Lower-Basin sub-basin, the transparent proof layer, branded and public.
Buy a first conservation tranche of verified, shepherded consumptive-use reduction at ~$400/AF, and watch it land in Lake Mead on the ledger.
Set the integrity standard with your team, additionality, shepherding, holdback, so AWS defines what a real replenishment acre-foot is, not just buys one.
Scale on proof. If the pilot verifies clean, AWS anchors the founding coalition and brings the other hyperscalers to the table. If it doesn’t, you’ve spent a pilot budget and learned something real.
A non-binding letter of intent to:
No basin-scale commitment. No purchase of water AWS can’t legally receive. Just the smallest real step that proves whether this is as good as it looks.