A 716-acre campus south of Cheyenne — codenamed first ‘Jade,’ then ‘Tembo’ — has been unmasked as Google’s most ambitious infrastructure bet yet: 2.7 gigawatts of compute, its own gas generation, and a developer left behind.
For months, one of the largest infrastructure projects in American history had no publicly named owner. Planning documents filed in Laramie County, Wyoming referred only to ‘Project Jade’ — later rebranded ‘Project Tembo’ — a 716-acre campus in the Switchgrass Industrial Park, eight miles south of Cheyenne, on land that until recently served as seasonal cattle grazing. On July 8, 2026, Cowboy State Daily unmasked the owner: it is Google.
The scale is almost hard to parse. At 2.7 gigawatts, the campus would draw roughly three times the electricity the entire state of Wyoming currently consumes. Planning filings suggest the site could ultimately scale toward 10 gigawatts — a ceiling number, not a built figure, and still moving through the permitting process. Google has said little publicly; ownership is established via planning documents and the reporting of Cowboy State Daily. But the footprint is unambiguous.
Two details elevate this beyond a routine ‘big data center’ story. First, power: the campus is paired with approximately 2.7 gigawatts of new natural-gas-fired generation developed with Tallgrass — Google isn’t simply plugging into Wyoming’s grid, it is building its own power plant alongside the servers. Second, developer displacement: Bloomberg, via Blockspace, reports that the original project developer Crusoe was pushed out once Google moved in. Laramie County approvals have pegged the combined data-center-plus-power scope at roughly $50 billion — a figure that is still evolving as permitting advances.
Crusoe assembles the site under ‘Project Jade’ in Laramie County; land secured on 716 acres of former grazing land in Switchgrass Industrial Park
Late 2025
Project rebranded ‘Tembo’; Laramie County filings peg combined campus-plus-power scope at ~$50B; Tallgrass engaged for ~2.7 GW of new natural-gas generation
Cowboy State Daily unmasks Google as owner via planning documents; project becomes public record
The key insight:Google didn’t just build a data center in Wyoming — it ran a land-and-power acquisition campaign behind a shell codename, assembled its own electricity generation because the grid couldn’t supply what it needed, and absorbed a developer who did the hard early work. That sequence is not incidental. It is the playbook.
The Structural Read
Project Tembo is the physical AI buildout in its rawest, most revealing form. Strip away the Wyoming geography and the cattle-grazing backstory, and three structural dynamics emerge — each one a signal about where the AI infrastructure race is actually heading.
The codename is the strategy. Hyperscalers don’t run projects under shell LLCs and changing codenames out of bureaucratic habit. They do it because the moment a Google-sized buyer is publicly associated with a land parcel or a power corridor, the market reprices everything in the vicinity. ‘Jade’ became ‘Tembo’ and stayed nameless through the critical assembly window precisely so that 716 acres of grazing land could be acquired at grazing-land prices. The secrecy is not incidental to the infrastructure play — it is how the land-and-power grab stays economically viable.
Power is now the product. The decision to build ~2.7 gigawatts of new natural-gas generation alongside the servers — via Tallgrass — is the most structurally significant detail in this story. It means Google concluded that no grid operator in the continental United States could reliably deliver that load on the timeline the AI buildout demands. The binding constraint of the AI era has moved upstream from chips to electrons. The winners are vertically integrating backward into energy generation, treating electricity infrastructure as part of the compute stack itself. This is the same electron-scarcity logic driving Meta’s $50 billion Louisiana Hyperion campus and the same force behind Goldman’s ~$1 trillion hyperscaler capex projection.
The developer gets squeezed out. Crusoe assembled the project, navigated the early permitting risk, and built the local relationships. Then Google arrived with a balance sheet that renders the original developer structurally unnecessary. This is the physical-layer version of a pattern visible across AI: smaller players de-risk the hard early work, and the hyperscaler with infinite capital absorbs the upside. At a capex scope that now registers as a measurable share of US GDP growth and is turning hyperscaler free cash flow negative, only a handful of entities on earth can play at this table.
Map of AI — Infrastructure Layer
“When a hyperscaler builds its own power plant to feed its own servers on land it assembled anonymously, the infrastructure layer of the AI stack has effectively seceded from the public grid. That is not a data center story. That is a vertical integration story — and it redraws the Map of AI from the bottom up.”
Three Implications
IMPLICATION 1 — Energy Is Now a Competitive Moat
A hyperscaler that controls its own generation at gigawatt scale has a structural advantage no grid-dependent competitor can easily replicate. Power access — not chip access, not modelquality — is becoming the durable differentiator in the AI infrastructure race. Google’s Tallgrass partnership is not a utility arrangement; it is a moat under construction. Expect Microsoft, Amazon, and Meta to close the same gap through similar off-grid or co-generation plays within 18 months.
IMPLICATION 2 — The Codename Playbook Will Spread (and Eventually Break)
Shell-LLC land assembly is now a documented hyperscaler strategy, which means local governments, utility commissions, and competing buyers will start pricing it in. Wyoming won’t be the last state where a ‘mystery project’ turns out to be a $50 billion Google campus — but the information asymmetry that made the strategy work will narrow as the pattern becomes visible. The next Tembo will face harder questions earlier in the process.
IMPLICATION 3 — Neoclouds and Developers Face a Structural Ceiling
Crusoe’s exit from Project Tembo is not an isolated outcome — it is a preview of the consolidation dynamics that will play out across the neocloud and AI infrastructure developer space. Smaller players are indispensable for absorbing early permitting risk and navigating local politics, but once a project crosses a certain scale threshold, hyperscaler balance sheets make independent developers structurally redundant. The viable path for neoclouds is either to stay below the threshold where hyperscalers want to take over, or to build a capability — like specialized cooling, co-location density, or edge deployment — that the hyperscaler genuinely cannot replicate in-house.
Business Engineer Framework
The Map of AI — Infrastructure Layer
Project Tembo sits at the deepest layer of the Map of AI: physical infrastructure. Understanding where Google, Meta, Microsoft, and Amazon are placing their infrastructure bets — and why the energy layer is now the most contested — is the analytical lens that separates structural insight from headline noise. The Map of AI tracks all 9 layers, 200+ companies, and the competitive dynamics reshaping each one.
Project Tembo is not a data center story — it is a vertical integration story told in gigawatts and shell LLCs. Google assembled 716 acres anonymously, paired the campus with its own gas generation because the grid physically couldn’t keep up, and absorbed the developer who did
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
