GridAI to Buy Gas Producer Pronghorn to Power AI Data Centers With 300 MW by 2027
AI data centers are short on electricity, and grid connections can take years. That is why GridAI Technologies is buying Wyoming gas producer Pronghorn and wants to build power plants right next to data centers. The first 300 MW, enough to power several large data centers, is promised for 2027. GridAI estimates the gas supply can support 600 MW in total. The company has no customers yet, so the 2027 deadline is a target, not capacity that exists.
AI-processed from TNW; edited by Hamidun News
AI data centers today lack not only chips but also electricity. In some regions, a grid connection takes years, and training and running models cannot wait. US-based GridAI Technologies has decided to approach the problem from a different direction: buy its own fuel source and build gas-fired power plants directly for data centers. To do this, it has agreed to acquire Wyoming gas producer Pronghorn Resources. The goal is to deliver the first 300 MW of capacity in 2027. For scale: megawatts measure how fast a data center consumes energy, and 300 MW is enough to power several large data centers. But there is a caveat: GridAI has no customers yet.
What exactly GridAI is buying
Pronghorn produces almost 100 million cubic feet of gas per day from about 3,200 wells in the Powder River Basin. Beyond the wells, the company owns more than 450 miles of pipelines and rights to nearly a million acres of land and mineral interests.
GridAI's own estimate is that current production can support about 600 MW of electrical capacity. That is a calculation, not finished generation: the power plants still have to be designed, built, and paid for. The first 300 MW, which is half of that estimate, is slated for 2027.
The deal is entirely in stock, not cash. Pronghorn's owners will receive about 1.25 million GridAI shares plus preferred shares. Closing is expected by November 22. GridAI trades on Nasdaq, and in October its market value was about $31 million. The buyer is therefore very small, and the announcement does not show where the money to build the plants will come from.
Why the idea makes sense but carries risk
The logic is simple. If gas is produced next to the plant, there is no need to wait for utilities to run lines, and no need to buy fuel from middlemen. Consider a startup renting thousands of GPUs: every month that hardware sits idle costs money, so connection speed may be worth more than the cheapest electricity.
The main risk is equally clear: GridAI has not named a single customer. Building plants without signed consumption contracts is a bet that demand will appear on its own. The market already shows that plans and real demand do not always match: last week Crusoe abandoned its plans in favor of Google, reportedly after failing to find enough customers. The source material gives no details on that story, so drawing direct parallels would be premature.
GridAI also gains rights to use a small modular nuclear reactor project in Wyoming. It currently exists only as a planned development, and the company itself stresses that its main focus remains gas.
What this means for the industry
Wyoming is becoming a notable site for AI infrastructure at a time when new data centers face resistance elsewhere. According to the source, Google plans its own project there, which will run mainly on gas. Access to fuel and the stance of local authorities are beginning to shape the geography of AI as much as taxes or proximity to users.
Practical takeaways for those planning capacity:
- Energy is a separate project with its own timeline. If you need a large cluster, expect that a grid connection may take years.
- In press releases, distinguish three things: the stated target (300 MW in 2027), the resource estimate (600 MW), and power plants that are actually built. Right now only the first two exist.
- Alternative energy suppliers like GridAI may become an option for data centers, but without contracts and confirmed financing, their timelines should be treated with caution.
Bottom line: the race for AI capacity is shifting from graphics cards to wells and pipelines. Whether GridAI can turn its gas assets into working power plants depends on customers and the money to build them. For now, this is more a claim on a niche than a finished business.
- Nov 22, 2026Anthropic Shares Fell 8.5% in a Day, Nearly Back to the $83 IPO Price
- Nov 22, 2026Linger Runs an AI Agent on a 16 GB Mac With No Token Fees and Promises a 5x Speedup
- Nov 22, 2026Ahrefs: A Brand With 10x the Web Mentions Gets Only 3.7x More in AI Answers
- Nov 22, 2026xhigh Mode in Gemini 4 Hydrogen Costs 6x More Than high: 23.61 Cents Instead of 3.99 Per Response
Key Questions Answered
What exactly GridAI is buying?
Pronghorn produces almost 100 million cubic feet of gas per day from about 3,200 wells in the Powder River Basin. Beyond the wells, the company owns more than 450 miles of pipelines and rights to nearly a million acres of land and mineral interests. GridAI's own estimate is that current production can support about 600 MW of electrical capacity. That is a calculation, not finished generation: the power plants still have to be designed, built, and paid for.
Why the idea makes sense but carries risk?
The logic is simple. If gas is produced next to the plant, there is no need to wait for utilities to run lines, and no need to buy fuel from middlemen. Consider a startup renting thousands of GPUs: every month that hardware sits idle costs money, so connection speed may be worth more than the cheapest electricity. The main risk is equally clear: GridAI has not named a single customer. Building plants without signed consumption contracts is a bet that demand will appear on its own.
What this means for the industry?
Wyoming is becoming a notable site for AI infrastructure at a time when new data centers face resistance elsewhere. According to the source, Google plans its own project there, which will run mainly on gas. Access to fuel and the stance of local authorities are beginning to shape the geography of AI as much as taxes or proximity to users. Practical takeaways for those planning capacity: Energy is a separate project with its own timeline. If you need a large cluster, expect that a grid connection may take years.
Ship AI Faster with Founders, Engineers & Weekly Architecture Reviews
Inside Vibe Club (hamidun.com/club) — weekly live masterminds with Zhemal Khamidun, Tech Radar, full LMS access, and 370+ production Claude Code skills.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.