On Nov. 5, 2026, Google said it would invest $17 billion to build its first data center in Wyoming and that it would pay for 100% of the power it uses. Two days earlier, voters in ten states had rejected data-center restrictions in 40 of 41 ballot measures. The sequence matters more than the size: a hyperscaler that takes the whole power bill converts a political objection into a financial one, and that changes who can build.
Resources contracted for one campus
2.7 GW
Includes reserve margins; Black Hills contract terms run through 2048
Data center ballot measures rejected
40 of 41
2026 midterms across 10 states; Ohio alone had 21
2028 U.S. capacity at risk
~30%
UBS estimate of construction pauses, Nov. 5, 2026
The contract
Google Did Not Buy a Data Center. It Bought a Grid.
This is not a campus announcement; it is a power-procurement strategy that happens to include buildings. Google has lined up gas generation owned by Black Hills, a 702 MW fuel-cell plant built by AEP, and carbon storage at Tallgrass Energy's Eastern Wyoming Sequestration Hub, with Bloom Energy supplying the cells, so that no ratepayer absorbs a dollar of the cost.
| Party | What was signed | Capacity and capital | Timing |
|---|---|---|---|
| Pays for 100% of the power it uses and covers the infrastructure needs driven by its growth; four data centers plus an office building, logistics center and network facilities | $17B | Mass grading later in 2026; full buildout in the first half of 2031 | |
| Black Hills | Large Power Contract Services Agreement and Generation Facilities Agreement, with Google bearing all costs of serving the campus and no cost shifting to other customers | Up to 590 MW of grid-connected service; $1.8B of new company-owned gas generation (564 MW nameplate) | Energy service from late 2027, ramping to peak load in 2030; terms through 2048 |
| AEP | 20-year power purchase agreement with Google through Cheyenne Fuel Cell, LLC, about 92% owned by AEP and 8% by Basalt Infrastructure Partners | About $4B for 702 MW of Bloom Energy fuel cells with carbon capture and storage | Construction starts in 2026; fully operational by end-2028 |
| Tallgrass Energy | Operates the gas generators serving the campus; its Eastern Wyoming Sequestration Hub stores captured carbon, and a front-end engineering study tests fitting capture to its combined-cycle turbines | Nearly 1 GW of contracted carbon-sequestration capacity | Not disclosed |
| Bloom Energy | Supplies the fuel cells for the 702 MW block, manufactured in the United States | Largest planned commercial-scale fuel cell development of its kind | Not disclosed |
| Related Digital | Developer of the campus and a co-sponsor of the Community Council | Not disclosed | Not disclosed |
- A six-party Community Council — Google, Tallgrass Energy, Black Hills, Related Digital, AEP and Bloom Energy — will shape how Cheyenne absorbs the buildout.
- Google expects hundreds of millions of dollars in local property taxes for Laramie County and School District 1, plus thousands of jobs.
- More than 200 residents attended a downtown Cheyenne community meeting, and no community benefits agreement is signed yet.
- About 2,500 workers at peak construction and 250 permanent operating roles once the campus runs.
The political trigger
Voters Rejected 40 of 41 Measures. Two Days Later, Someone Answered.
Voters did not reject the machines; they rejected the bill. Of 41 data-center ballot measures across ten states in the Nov. 3, 2026 midterms, voters backed restrictions or rejected pro-development measures in every contest with results except one, and Ohio alone carried 21 of them. The single exception was Harwood, North Dakota, where voters declined to add new review of industrial projects.
Two days after that vote, Google put a price on its answer. The company did not frame $17 billion as a data center; it framed it as investment in Wyoming's infrastructure, energy and workforce, with a six-party Community Council already in place as mass grading approaches.
The mechanism
The New Rule: Draw the Load, Pay for the Grid
The template already exists in policy. The White House's Ratepayer Protection Pledge of March 4, 2026 asks hyperscalers and AI companies to build or buy the new power they need, to pay for every power-delivery upgrade their data centers require, to pay those rates even if they never use the electricity, to hire locally, and to make backup generation available during scarcity. Google went a step further: instead of promising to pay, it signed contracts with the utilities that own the assets and put its own credit behind the promise.
| Model | Who carries the grid bill | Contract length | Who can use it |
|---|---|---|---|
| Self-funding — Google | The developer pays the full cost of power and of the upgrades it drives; the utility keeps cost pass-through, stranded-asset and early-termination protection | Through 2048 | Investment-grade balance sheets |
| Policy pledge — OpenAI | Commits to pay the full cost of electricity and serving infrastructure, posts financial assurances before major spending, uses no NDAs with public officials and publishes construction plans | No fixed term disclosed | Companies without utility-grade credit |
| Third-party lease — neoclouds | The landlord and utility carry the build; the tenant takes occupancy only when power arrives | Project by project | Credit-constrained tenants |
| Hybrid — Microsoft | Owns generation but also leases, and exited the Nscale UK anchor deal in October 2026 when power-connection delays pushed out the supply date | Mixed | Balance sheets able to do both |
Which produces a two-tier market. Capital-rich builders buy power certainty, and with it speed; everyone else sells the same politics as a promise. Bloom Energy shows both sides at once: it makes the fuel cells Google just ordered, and its shares fell 16.4% on Nov. 5, in a session when AI infrastructure stocks were sold on fears that data-center limits would now bite.
Who captures the load
The Utility Is the New Infrastructure Company
Black Hills is the company that changes shape. It owns 1,481.5 MW of generation across Colorado, Montana, South Dakota and Wyoming and serves about 218,000 electric customers; the Google contract brings 2.7 GW of contracted resources, roughly 1.8 times its entire owned fleet, and 564 MW of new gas generation it will build and own.
| Black Hills' Google contract | Disclosed figure |
|---|---|
| Capital investment in new generation, 2027–2029 | $1.8B |
| New company-owned natural gas generation | 564 MW nameplate |
| Grid-connected energy service | Up to 590 MW |
| Third-party resources managed via a private microgrid | About 2.1 GW |
| Total contracted resource mix, including reserve margins | 2.7 GW |
| Estimated net income in 2030 | About $150M |
| Unlevered free cash flow through 2048, net of capital spending | About $2.4B |
| Contract term | Through 2048 |
| 2026 adjusted EPS guidance, reaffirmed | $4.25–$4.45 |
| Q3 2026 GAAP diluted EPS and adjusted EPS | $0.54 and $0.57 |
| Net debt to EBITDA, trailing twelve months | 6.2x |
- The Wyoming commission approved $4.0 million of new annual revenue at a 9.60% return on equity on a 53% equity layer, and new rates take effect Dec. 1, 2026.
- $686 million of transmission certificates are pending: $80 million for South Cheyenne and $606 million for Aeolus, roughly 125 miles of line upgradable to 345 kV and completing mid-2029, with decisions expected mid-2027.
- The Large Customer Transmission Cost Adjustment Mechanism, filed June 19, 2026, decides who ultimately pays for that expansion; a Wyoming decision is expected in the first quarter of 2027.
- Management puts $2.5 billion of additional Wyoming generation and transmission investment on the table as significant upside.
- The all-stock merger with NorthWestern Energy needs only Montana's approval to close by year-end 2026, with each NorthWestern share converting into 0.98 Black Hills shares.
The market reaction
The Tape Drew the Line: Contracts Paid, Prompts Sold
Nov. 5, 2026: the utility rose, the buildout fell
Single-session percentage moves on Nov. 5, 2026. Peer moves as reported by Reuters; closing prices for Alphabet, Black Hills and Bloom Energy.
Unit: %
Q3 revenue up 138%
-16.4%
Neocloud power exposure
-15%
Five-year credit spread at a record
-10%
Chip demand read-through
-9%
Power equipment
-7.4%
Rented power
-3.6%
Merchant generation
-2.5%
$17B self-funded
-2.2%
22-year contract signed
0.5%
Read the tape as a verdict on contract structure, not on technology. Black Hills rose 0.5% to $84.63, within 2% of its 52-week high, on a day the Nasdaq fell 5.5%. Alphabet fell 2.2% to $282.57, 31% below its 52-week high of $408.61, and roughly 4% above its 52-week low of $270.70 — a market that had stopped paying for AI capacity on promise.
Bloom Energy is the sharpest illustration. Third-quarter revenue rose 138% to $1.24 billion, gross profit reached $417 million, and operating income was $225 million against $8 million a year earlier. The stock still fell 16.4% to $167.52, 52% below its 52-week high of $351.28, on about 121 times trailing earnings. The market is not paying for fuel cell order books; it is paying for rate-based cash flows with a credit standing behind them.
Horizons
Two Clocks: Permits Now, Cash Flow in 2030
- First quarter of 2027: the Wyoming commission rules on the Large Customer Transmission Cost Adjustment Mechanism — the first real decision on who funds grid expansion for large loads.
- By year-end 2026: Montana's approval closes the Black Hills and NorthWestern Energy merger, removing the last condition on the combined balance sheet.
- December 2026: new rates take effect in Wyoming at $4.0 million of annual revenue and in South Dakota at $27.4 million, with a four-year base-rate freeze until August 2030.
- Now through 2027: Black Hills guidance assumes 50 million to 70 million shares of equity issuance to fund the build — dilution is the price of underwriting.
Short term, this is a regulatory calendar, not an earnings story: earnings at Black Hills begin in mid-2027, and energy service starts only in late 2027.
- 2027 to 2028: AEP's 702 MW fuel-cell block targets full operation by end-2028, the first commercial-scale fuel cell with carbon capture at this size.
- 2029 to 2030: the campus ramps to peak load in 2030, with about $150 million of net income at Black Hills and about $2.4 billion of unlevered free cash flow through 2048.
- 2028 is the swing year for the whole trade: UBS's roughly 30% capacity-at-risk estimate lands exactly when self-funded campuses are scheduled to come online.
- One to three years: if the model holds, power availability becomes a financing product rather than a scarce input, and the assets that capture it are the utilities with contracted, credit-backed load.
Fact: Google has committed to pay for 100% of the power behind a $17 billion campus, and two utilities have put about $5.8 billion of generation behind it under contracts running to 2048. Inference: because those contracts move grid cost off ratepayers, builders with investment-grade credit can now buy speed that everyone else must negotiate for. Speculation: if that holds, the 2028 buildout gets smaller, more concentrated and more expensive to finance — and the utilities that signed the contracts capture the spread.
Risks
What Would Break the Thesis
- Scale is unproven: 702 MW of fuel cells with carbon capture is the largest planned commercial-scale deployment of its kind, and the supplier's shares fell 16.4% on the day of the order.
- Concentration: Black Hills commits $1.8 billion against a single customer, and the stranded-asset and early-termination protections are contractual, not a guarantee.
- Leverage and dilution: net debt of 6.2 times EBITDA, negative trailing free cash flow, and 50 million to 70 million shares of planned issuance in 2026.
- Politics is not only about power bills: water, noise, land use and truck traffic are separate ballot issues, and 40 of 41 measures went the wrong way for the industry.
- Demand: Alphabet is spending $159.8 billion of trailing capital expenditure against $19.7 billion of trailing free cash flow, and no one can price away the 30% of 2028 capacity UBS puts at risk.
- Carbon: capture on the combined-cycle turbines is a front-end engineering study, not an operating asset, and nearly 1 GW of contracted sequestration capacity still depends on it.
Who this actually touches
- Signs 2.7 GW of contracted resources through 2048, taking the grid bill off ratepayers and turning political risk into a balance-sheet line item.
- Q3 2026 revenue rose 24.7% to $127.6 billion with $40.6 billion of operating income; trailing capital spending of $159.8 billion makes $17 billion affordable.
- Shares closed at $282.57 on Nov. 5, 31% below the 52-week high, so the market has not yet paid for power self-sufficiency.
- Over 2027–2030, power certainty buys queue position in a market where UBS sees 30% of 2028 capacity at risk.
- Converts a 22-year contract with a hyperscaler into about $150 million of 2030 net income and $2.4 billion of unlevered free cash flow through 2048.
- Q3 2026 adjusted EPS of $0.57 against $0.45 a year earlier, with 2026 guidance of $4.25–$4.45 reaffirmed.
- Shares closed at $84.63 on Nov. 5, within 2% of the 52-week high, while the Nasdaq fell 5.5%.
- Over 2027–2029, $1.8 billion of gas generation and $686 million of transmission certificates turn a signed deal into rate base.
- Commits roughly $4 billion and 92% of Cheyenne Fuel Cell to a 702 MW fuel-cell plant with a 20-year Google contract.
- Operating by end-2028 puts the rate-base and earnings contribution after 2027, when the first commercial-scale fuel cells run.
- The offset: net debt of 6.0 times EBITDA and no operating precedent for a 702 MW fuel-cell fleet at utility scale.
- Near term, construction must start in 2026; slippage past end-2028 removes the contracted cash flow.
- Supplies the fuel cells Google just ordered, yet the shares fell 16.4% to $167.52 on Nov. 5 and sit 52% below the 52-week high.
- Q3 2026 revenue rose 138% to $1.24 billion, with operating income of $225 million against $8 million a year earlier.
- At 121 times trailing earnings on $3.83 billion of trailing revenue, the valuation assumes flawless delivery of a first-of-its-kind build.
- Over 2027–2028, execution on the 702 megawatt block is the test; further large orders are the upside.
- Exited the Nscale UK anchor deal in October 2026 when power-connection delays pushed out the supply date, showing that rented power is not power.
- Trailing capital spending of $135.9 billion through Sept. 30, 2026 makes the scale enormous; owning generation would price that risk directly.
- Shares fell 3.6% on Nov. 5 as the market repriced data-center buildout risk.
- Over 2027–2028, self-funding power or renting it becomes the central margin question for AI capacity.
- Fell 2.5% on Nov. 5 as hyperscaler power politics hit the whole merchant generation complex, not one name.
- Trailing revenue of $31.3 billion and operating income of $4.6 billion show a contracted generation base chasing the same hyperscaler demand.
- Net debt of 2.96 times EBITDA leaves room to underwrite large loads directly, which is now the differentiator.
- Over 2027–2030, self-funded supply favors whoever can commit capital at signing rather than whoever signs first.
- Lost 15% on Nov. 5 as Microsoft tightened oversight of projects sourced from it and from Nscale, exposing the tenant's power dependency.
- The landlord and utility carry the build while the tenant takes occupancy only when power arrives, the structure the new rule targets.
- Capital spending runs far above revenue, so each delay pushes more money into a build the company does not control.
- Over 2028, UBS's roughly 30% capacity-at-risk estimate lands hardest on tenants without contracted power.
- Awaits only Montana approval to close its all-stock merger with Black Hills at 0.98 Black Hills shares per NorthWestern share.
- Q3 2026 revenue of $423.8 million and EPS of $0.50 come with net debt at 6.2 times EBITDA.
- Combined, the two would span Montana, South Dakota and Wyoming, the states where large-load interconnection is contested.
- Over 2027–2030, integration determines whether Wyoming's stated $2.5 billion of large-load upside gets funded.
