THE RATEPAYER IS THE COUNTERPARTY
Everyone charts the capacity. A few chart the tokens. Nobody charts who pays for it.
THE SECOND OPINION
There is a large and growing pile of municipal paper whose coverage ratios are underwritten on forecast load. Not contracted load. Forecast. And the forecast was built in a political environment that stopped existing sometime around March.
Evaluated pricing on those names is not wrong. It is answering a different question than the one you should be asking. ICE, BVAL, and LSEG will mark the sector off comparable trades and curve moves. None of those inputs know that 530 local ordinances now seek to ban or restrict the thing generating the load, that more than 50 projects have been canceled this year — better than double all of 2025 — or that eight died in July alone.
The mark is not lying yet. It is early. That is a different problem, and a worse one, because early is the only time the work is worth doing.
THE ARGUMENT
Three legs to this. Two get charted constantly. The third is the one that prices credit.
Leg one: demand is real. Google went from 9.7 trillion tokens a month in May 2024 to roughly 480 trillion by I/O 2025 to north of 3.2 quadrillion in May 2026. That is 330x in twenty-four months and 7x year over year. OpenRouter is at 25 trillion a week across eight million users. Fireworks is running 40 trillion a day. Anyone selling you an AI unwind story built on empty data centers is arguing with a public meter that reads vertical.
Leg two: it is all spoken for. Northern Virginia vacancy is 0.3%. Atlanta 1.0%. Dallas–Fort Worth 1.8%. Core-four inventory grew 33% year over year and absorption still ran 34% higher at 2,236 MW. Eighty percent of what is under construction is preleased. This is not a market with a supply overhang. It is a market where developers are in catch-up mode signing leases on buildings that deliver in 2029.
Leg three: nobody plots this one. In September 2025, Americans split evenly on a data center near them — 43% for, 42% against. By May 2026, more than seven in ten were opposed. A forty-nine point swing in nine months. Over the same window, the share blaming data centers for their electric bill went from 28% to 53%. A majority now supports a national moratorium; forty percent support it strongly. Roughly six in ten back a moratorium at some level of government, and majorities of both self-identified Democrats and MAGA Republicans back it at all three.
Legs one and two are denominated in tokens and megawatts, and they accrue to shareholders. Leg three is denominated in a monthly utility bill, and it lands on a ratepayer who has a vote and a county commission. Both sets of facts are true simultaneously. The error is treating them as a debate.
WHY THIS IS A CREDIT STORY
The transmission mechanism is short and it is not subtle. Poll moves, local ordinance passes, interconnection or entitlement stalls, load arrives later than the feasibility study said, and debt service coverage that was built on ramped load gets tested on unramped load.
New York already did it: a first-in-the-nation one-year pause on new data center permit approvals. Pallone has called for a national moratorium. Warren, Van Hollen, and Blumenthal have an open probe into the ratepayer pass-through. This is no longer a siting story confined to a county zoning board.
The names to work through are the ones where the take-or-pay is thin and the forecast is doing the load-bearing. Public power systems that upsized generation and transmission against a single anchor tenant. Municipal utilities whose rate stabilization funds assume a step-function in industrial sales. Special districts and IDBs where the data center is the credit. Where there is a real contract with a real investment-grade counterparty and a termination fee that actually stings, you are fine. Where there is a letter of intent and a growth assumption, you are underwriting a poll.
The cost is already visible in the physical inputs. Copper is +18.6% year to date and +51.9% over twelve months. Aluminum +9.8% YTD. Coal +19.3%. Those are the materials that go into the transmission, the switchgear, and the generation that the load requires, and they are not being bid by data center operators — they are being bid by every utility in the queue trying to serve them. That input inflation arrives in a rate case eighteen months from now with a ratepayer's name on it.
The offset holds. Washington will keep tolerating energy inflation as the price of AI deflation — that trade has not changed. What has changed is that localities were never party to it. That gap is where the spread is.
THE WIRE
530+ local laws now seek to ban or restrict data centers. — More than 200 pending projects contested; 50+ cancellations YTD, double all of 2025. A supply constraint now, not a headline.
New York's one-year permitting moratorium stands as the only statewide pause. — Watch Illinois. A second state converts this from anomaly to trend.
Pallone calls for a national AI data center moratorium ahead of E&C markup. — Ratepayer Protection Act is the vehicle. A federal cost-allocation standard reprices the sector.
CBRE: global inventory 16 GW in Q1, +25% y/y; vacancy 6.7% from 8.3%. — Supply growing fast and still losing the race.
Google monthly tokens exceed 3.2 quadrillion, +7x y/y. — The demand denominator nobody in the moratorium debate is using.
Distillate squeeze: 3-2-1 crack at $59.34/bbl, heating oil +82.7% YTD. — Crude is the quiet leg here. Products are doing the work. Gulf tensions and a fading Hormuz deal keep the risk premium in the barrel, not the curve.
Copper +18.6% YTD, aluminum +9.8%, coal +19.3%. — The grid buildout priced in metal. Utilities are the marginal bidder and the ratepayer is the residual claimant.
30Y at 5.210 trades through the 20Y at 5.216. — Six-tenths of a basis point of inversion at the very long end. Not a signal on its own; worth watching if it widens into next week's supply.
Data: Koyfin. Rates, FX and commodities live 8/7/26 07:29 ET; corporate credit indices and fund prices as of 8/6/26 close. Chart sources as marked.
The Bond Bro Dispatch is published by Positive Carry LLC. All content is general market commentary provided for informational and educational purposes only and does not constitute investment advice, a recommendation, an offer, or a solicitation to buy or sell any security. Nothing herein is tailored to the circumstances of any recipient. Data are drawn from sources believed reliable; accuracy and completeness are not guaranteed. [email protected] · © 2026 Positive Carry LLC. All rights reserved.