Former BP Chief Links with AI Data Center

Introduction

The former BP chief links with AI data center infrastructure in a move that captures the fastest shift in corporate leadership this decade. Bernard Looney ran BP from February 2020 until his resignation in September 2023. He has now taken over as chief executive of Prometheus Hyperscale, leading a Wyoming buildout the company puts above thirty billion dollars. Rocky Mountain Power told reporters that the planned campus will consume more electricity than every residential household in Wyoming combined, a load confirmed by TechRepublic’s Wyoming AI power report. That single sentence reshapes the energy transition story that defined the first half of the decade. It frames a hyperscale AI campus not as a software project but as a utility-scale power deal with grid, water, and workforce consequences. The move also signals a repeatable pipeline: energy veterans with capital allocation experience are now the candidate pool for AI infrastructure leadership. This article unpacks the deal, the engineering, the governance, and the risks that come with gigawatt scale AI compute landing in rural Wyoming.

Quick Answers on the Former BP Chief and the AI Data Center Buildout

Who is leading the AI data center buildout at Prometheus Hyperscale?

Bernard Looney, the former BP chief executive, took over as CEO of Prometheus Hyperscale in April 2026. He succeeded founder Trenton Thornock, who remains on the board as an advisor.

How large is the Wyoming AI data center project?

The AI data center program plans two Wyoming campuses, one in Evanston and one in Casper, with each starting around 1.2 gigawatts. The combined pipeline could approach ten gigawatts and thirty billion dollars.

Where does the power come from for a gigawatt scale AI campus?

The AI data center starts behind the meter with natural gas, then adds solar, wind, and small modular nuclear reactors through an Oklo twenty year letter of intent. A grid connection is planned later.

Key Takeaways on Bernard Looney’s Move Into AI Infrastructure

• Bernard Looney went from BP chief executive to the top job at a Wyoming AI data center developer in roughly thirty months, a shorter arc than any comparable Big Oil to technology move on record.
• Prometheus Hyperscale is planning more than thirty billion dollars of investment across Evanston and Casper, with a combined pipeline that could approach ten gigawatts of installed capacity.
• The campus will consume more electricity than every residential household in Wyoming combined, according to Rocky Mountain Power, which puts the project at the center of the state’s grid planning for the rest of the decade.
• In-Q-Tel, the venture arm backed by the United States intelligence community, holds a stake and a board seat, giving the project a governance signature no earlier AI campus has carried.

Understanding Why a Former BP Chief Links with AI Data Center Projects

The former BP chief links with AI data center work because gigawatt scale AI compute is primarily an energy, land, and water problem that oil majors have solved before. Looney now leads Prometheus Hyperscale, pairing capital allocation experience with the technical demands of behind the meter AI campus construction.

From North Sea Oil Fields to Gigawatt Scale AI Campuses

Looney spent his entire corporate career at BP, rising from petroleum engineer in the North Sea to chief executive in February 2020. He left in September 2023 after failing to disclose past personal relationships with colleagues at BP. The board later classified the exit as serious misconduct in a formal review. BP denied Looney roughly thirty two million pounds, about forty million dollars, in pay and share awards, as Data Center Dynamics documented. That reputational hit shaped his public silence for most of 2024, which makes the Wyoming move look even sharper. The energy connection to AI infrastructure is not accidental, and it is why oil veterans are increasingly on hyperscale shortlists.

Building on that foundation, Prometheus Hyperscale needed a leader comfortable with multi billion dollar capital stacks, long construction timelines, and politically sensitive energy contracts. Looney joined as non executive chairman in late 2024, during the company’s relaunch from Wyoming Hyperscale Whitebox, and quickly took a visible role in investor and policy briefings. His BP tenure included the pivot toward lower carbon operations, which gives the Prometheus narrative a credible operator at the top. The elevation to chief executive in April 2026 formalized what the market had already priced in. Analysts covering the energy to AI pipeline have begun comparing his profile to Big Oil peers eyeing similar moves.

Beyond the biography, the move captures a broader shift in who the AI industry hires to run physical infrastructure. The AI boom is turning data center construction into a program management problem that resembles upstream oil and gas: siting, permitting, pipelines, water rights, workforce housing, and tax base negotiation. A former chief executive who ran a two hundred billion dollar energy major understands those variables in a way few technology executives do. The Prometheus hire is a signal flare, as aiplusinfo coverage of AI data centers driving the gas surge shows. Expect more such hires in the next twelve to eighteen months.

Inside Prometheus Hyperscale and the Wyoming Buildout

Shifting focus to the project itself, Prometheus Hyperscale is a Wyoming data center developer. Trenton Thornock founded it in 2020 as Wyoming Hyperscale Whitebox and relaunched it under its current name in September 2024. The company has publicly disclosed two Wyoming campuses: one in Evanston on part of the Thornock family ranch, and one outside Casper on a separately assembled site. The Evanston plan began at one hundred twenty megawatts under its original scope. The company revised the design to a gigawatt scale after securing the Looney commitment and the In-Q-Tel investment. The company has also signaled campuses in Arizona, Colorado, and outside Dallas, though those are earlier in the development stack. The disclosed pipeline is the clearest window the industry has into the Prometheus thesis.

Capacity numbers at Prometheus have moved fast and are worth separating by source. Cowboy State Daily reported each Wyoming campus starting at roughly 1.2 to 1.25 gigawatts. Evanston could reach four to five gigawatts and Casper two to two and a half gigawatts. Looney himself cited a broader pipeline approaching ten gigawatts in April 2026, a figure captured in the Cowboy State Daily reporting on the thirty billion dollar buildout. Data Center Dynamics gave a more conservative 2.5 gigawatt pipeline number earlier in the year, before the latest site revisions. The discrepancy reflects live planning across the pipeline, not conflicting facts about the sites. Analysts should track the next quarterly update for a reconciled schedule.

Site selection in Evanston was not accidental, and it tells you how Prometheus thinks about risk. The campus sits on a roughly one square mile parcel of the Thornock family ranch, which simplifies land assembly and gives the company a long dated option on expansion acreage. The site is close to natural gas lines, wind and solar resources, and transmission that could tie into regional grids later. Casper offers a different mix, with established workforce depth and proximity to a college willing to build a dedicated training pipeline. Taken together, the two sites hedge each other on energy, workforce, and regulatory exposure.

Project phasing is the final piece of the picture, and Prometheus is explicit about it. The first phase goes behind the meter with on site generation, trades grid dependency for schedule certainty, and begins commissioning compute in staged tranches rather than one giant handoff. Later phases add renewable PPAs, bring the Oklo small modular reactor contract online, and connect to the grid once interconnection studies clear. The company has not published a formal commercial operation date, but public statements point to initial compute capacity before 2028 and the first nuclear generation by 2029 or 2030. Readers tracking similar programs can compare notes with Blackstone’s data centers navigating AI power challenges, which show how private capital stacks are solving the same timing puzzle.

The Energy Argument Driving Oil Veterans Into AI Infrastructure

Turning to the energy argument, the Prometheus thesis rests on a simple observation: AI compute is bottlenecked by electricity, not by chips. The International Energy Agency projects data center electricity demand will roughly double by 2030, with AI focused centers tripling to around 465 terawatt hours. Those numbers put AI infrastructure on a trajectory that oil veterans recognize instantly. The IEA AI and data centre electricity surge analysis keeps resurfacing in strategy decks. Looney’s background in upstream capex and firm supply contracts maps onto behind the meter campus economics directly. The energy transition narrative that framed BP’s pivot now rhymes with the AI build.

Beyond the top line forecasts, the operational playbook is also familiar to oil industry alumni. Hedging gas supply, structuring long dated offtake agreements, and managing dispatch across multiple generation sources are standard oil and gas practice. Those are also the daily tasks of running a behind the meter AI campus with gas, wind, solar, and nuclear in the mix. Looney and his peers can treat compute like a petrochemical complex: a fixed capital base with variable fuel costs and strict reliability constraints. That mental model, visible across the AI chip wars coverage, makes the former BP chief and his cohort newly valuable to AI infrastructure.

Behind the Meter Implementation: Natural Gas, Nuclear, Solar and the New Power Stack

Looking at the power stack in detail, Prometheus starts behind the meter with on site natural gas generation at commissioning. The company plans a measured transition to renewables and small modular nuclear reactors. The company calls this an island solution, meaning the campus runs independently of the grid during early operations. Natural gas is the only fuel that can be sited and commissioned at gigawatt scale fast enough to match the AI capex curve this decade. Prometheus signed a twenty year power purchase letter of intent with Oklo, a commitment detailed by Sustainability Magazine on the Prometheus program. That contract alone changes the risk profile of the project. It gives Prometheus a long dated clean baseload hedge that most hyperscalers do not yet have.

Renewables fit into the stack as additive capacity rather than base load. Wyoming has strong wind resources and growing solar penetration, and both can be dispatched into on site microgrids with modern inverters and battery storage. Prometheus has indicated it will layer wind and solar PPAs on top of the gas and nuclear base over the first five years of operation. The campus power management system handles ramping and dispatch between sources, with gas available as a firming resource whenever variable supply dips. This design pattern mirrors the hybrid stacks you see in the broader coverage of Amazon and Nvidia AI energy sources.

Grid connection is deferred by design, not deferred by accident. Behind the meter construction lets Prometheus avoid the multi year interconnection queues that have strangled other AI developers. Once the campus is operating stably, Prometheus can file for interconnection studies and sell firm capacity back into the regional grid. That turns the data center into a flexible load and occasional supply resource. That optionality is one of the more interesting features of the Wyoming design. Readers can trace the broader pattern through separate aiplusinfo reporting that explains AI datacenter energy use quadrupling by 2030.

Liquid Cooling, Closed Loop Water and the Thermal Challenge

Stepping back from fuel, the thermal problem is where modern AI campuses either win or lose. GPU racks dissipate heat densities that air cooling cannot match without burning twenty to forty percent of total energy on fans and chillers. Prometheus plans liquid cooling across its Wyoming halls, cutting power use by up to fifty percent versus air cooling, as the Sustainability Magazine cooling profile describes. The approach reduces mechanical overhead, lowers plant complexity, and makes it easier to hit tight power usage effectiveness targets at gigawatt scale. It also unlocks waste heat capture, which Prometheus plans to use for industrial and agricultural partnerships adjacent to the sites.

Water is the other half of the thermal story and the one that keeps community lawyers busy. Prometheus has committed to a closed loop system that circulates non potable water drawn from deep aquifer wells that are not connected to community supplies. The company contrasts that approach with Phoenix and Virginia data centers that rely on municipal utility water. Closed loop cooling also reduces evaporative loss, cutting daily make up water compared with cooling tower designs. The design does carry questions about long term aquifer draw and brine disposal, which regulators and local hydrologists will scrutinize. That is why environmental review belongs squarely in the hyperscale conversation.

Capital, Governance and the In-Q-Tel Connection

Among the capital partners, the one that stands out is In-Q-Tel, the venture arm backed by the United States intelligence community. In-Q-Tel took a stake in Prometheus Hyperscale and holds a seat on the board, giving the project a governance signature that no other announced AI campus carries at this scale. Fortune framed the move as an intelligence community bet on domestic AI infrastructure security, as its March 2026 Looney appointment report captures. The investment amount has not been disclosed, which is standard practice for the fund. The board role, by contrast, is public and material to how Prometheus will make procurement, siting, and customer decisions.

Beyond In-Q-Tel, the capital stack blends private equity, strategic energy investors, and allied family office capital. The company has not published a complete investor list, but filings and local reporting suggest a syndicate structured around long dated infrastructure holders. That matters because a thirty billion dollar project needs patient equity willing to see ten to fifteen year payback curves. Short duration growth capital would reprice the campus economics every quarter and risk forcing early concessions. Prometheus’s choice of partners is therefore itself an engineering decision. It also sets up the governance frame for how the project interacts with Wyoming regulators and with potential anchor customers.

The In-Q-Tel seat raises a legitimate policy question about dual use AI infrastructure. Compute capacity large enough to train frontier models can also be used for sensitive national security workloads, and the line between commercial customers and intelligence partners is becoming more permeable. Prometheus has not published an acceptable use policy, nor a public statement on the share of capacity reserved for government or classified workloads. Civil society groups and local reporters have begun asking those questions in advance of permit hearings. The answers will shape how the public understands the project. Separate reporting on data centers driving up electricity costs on aiplusinfo adds useful context on the broader public interest stakes.

Workforce, Training and the Casper College Pipeline

Shifting focus to the workforce question, each Prometheus campus is projected to employ roughly six hundred skilled workers once fully commissioned. The company has partnered with Casper College to launch a data center technology training pipeline. Cowboy State Daily reporting on the Wyoming buildout profiled the workforce commitment. The curriculum covers mechanical, electrical, instrumentation, and network specializations that match the on site engineering needs. Prometheus wants a locally sourced workforce because it reduces housing pressure and ties the campus to Wyoming’s political economy. The training pipeline also addresses one of the biggest grievances communities raise against hyperscale builds, which is that jobs go to out of state contractors.

Workforce planning for AI hyperscale is more complex than it looks, and it begins before the first shovel. Wages for data center operations in rural markets routinely clear seventy thousand dollars for mid level technicians and six figures for shift managers. Each project where a former BP chief links with AI data center work needs reliable labor. Prometheus is funding apprenticeship cohorts, housing partnerships, and relocation stipends. Those investments double as local political insurance, because they create visible benefits for state legislators and county commissioners. Readers tracking the training economics can compare notes with energy efficient AI training techniques on aiplusinfo for how the operations side of AI compute is professionalizing fast.

Ratepayer, Water Table and Community Impact in Wyoming

Turning to community impact, the single most important number in the entire deal is the ratepayer effect. Rocky Mountain Power has publicly stated that the Prometheus campus will consume more electricity than every residential household in Wyoming combined. Industry commentary tracking this comparison, including coverage at babl.ai on the Cheyenne-adjacent AI power footprint, has shaped the state’s conversation about rate design. Prometheus argues the behind the meter design means households will not subsidize the campus, which is a defensible position for phase one. Longer term grid connection could change that calculus, which is why rate design is now on every regulator’s whiteboard. Expect a Wyoming Public Service Commission docket within the next year.

Water table concerns are the second community issue, and they are quieter but structurally important. Wyoming is a semi arid state, and even non potable aquifer withdrawals can affect neighboring agricultural wells over multi year draw cycles. Prometheus has committed to monitoring wells and public reporting, but the state’s water rights framework is complex and litigious. Local ranchers have begun asking for independent hydrology studies before major pumping begins. Those requests will be the first real test of how Prometheus handles community pushback.

Tax base and infrastructure effects round out the community story. A thirty billion dollar capex program generates sales tax, use tax, and property tax at a scale Wyoming has rarely seen. State and county officials will see transformational revenue once the first phase hits the tax rolls. The political alignment to protect the project is already visible in Casper and Evanston. The tradeoff is heavy road wear, construction traffic, and short term housing shortages that create friction with existing residents. Programs like optimizing AI data centers for sustainability suggest how a disciplined operator can mitigate those externalities over time.

Regulatory and Environmental Review of AI Hyperscale Projects

Beyond the community lens, regulatory review for Prometheus will combine federal, state, and local processes with timelines measured in years. Air permits for the gas fleet, water rights filings, interconnection studies, and nuclear licensing for the Oklo units each run on their own clock. The Recharge News profile on oil chiefs backing gas and nuclear describes the political framing Prometheus is leaning into. Environmental impact statements may be required for the Oklo reactor and for larger surface disturbances at the Evanston site. Stakeholders should expect public comment periods and formal hearings on each permit. Interveners will come from both environmental groups and competing developers across the sector.

Prometheus has staffed up government relations and environmental teams well in advance of these filings. The company’s hiring pattern favors former state regulators and utility planners, which signals an intent to run a disciplined review process. Smart developers treat permitting as an engineering discipline, not a political afterthought. Prometheus clearly understands that framing from its first environmental filings. The project’s visibility also means every filing will be scrutinized by national press and advocacy groups. That attention is a tax on execution speed, yet it is the price of being the biggest private build in state history.

Risks and Open Questions for Prometheus Hyperscale

Looking at the risk register, demand risk is the first thing a disciplined investor flags. If the AI capex cycle softens, a ten gigawatt pipeline can look oversized in a hurry. Prometheus’s thesis depends on continued hyperscaler and sovereign demand for GPU hours. Data Centre Magazine on the AI boom doubling demand treats that as locked in. Even so, a single pause by a major anchor customer could reshape the economics. Analysts should model multiple demand scenarios before accepting the stated capacity schedule.

Technology risk is the second category, and it maps directly onto the Oklo contract and the liquid cooling platform. Oklo’s small modular reactor technology still needs commercial scale validation and ongoing Nuclear Regulatory Commission milestones. Liquid cooling at gigawatt scale is proven in enterprise and HPC settings, but gigawatt deployments stress the service, maintenance, and plumbing supply chains in new ways. Both risks are manageable with staged deployment, and Prometheus’s design accommodates that staging. Mitigation does require real contingency capital, which is why the project needs patient investors.

Governance and reputational risk are the third category, and they tie back to Looney’s BP exit and the In-Q-Tel seat. A high profile executive with a documented serious misconduct finding becomes a story the first time a public hearing goes sideways. The In-Q-Tel seat invites civil liberties and transparency questions that cannot be answered by press release. Prometheus will need a steady communications discipline and clear workload disclosure practices to manage those pressures. The absence of either will become a story in itself.

Finally, grid and interconnection risk sits in the fourth bucket. Behind the meter is a smart first phase, but grid interconnection is where many AI campuses have stalled for two or three years. Regional transmission organizations are drowning in queue backlogs, and interconnection studies have become a bottleneck across the entire sector. Prometheus’s plan to connect later gives it time, but the slot assignments it eventually receives will shape expansion economics. Readers tracking similar supply dynamics can review AI reveals hidden oil wells for an adjacent case of AI meeting legacy energy.

Ethics of Former Oil Executives Leading the AI Energy Transition

Shifting to the ethics lens, Looney’s move raises questions the AI industry has avoided for most of the current cycle. Hiring executives from Big Oil to build AI compute bets that operational discipline matters more than past climate exposure. The tension in generative AI’s rising energy costs impact explores that trade in depth. It is defensible on the merits, because AI compute desperately needs people who can run multi fuel energy portfolios at scale. It is also uncomfortable, because climate commitments made by hyperscalers now depend on the choices of leaders whose careers were built on fossil extraction. Both of these observations can be true at once in the current cycle.

The deeper ethical issue beneath the move is corporate accountability for climate outcomes. Oil executives have faced lawsuits, shareholder revolts, and reputational cycles that constrain their decision making. AI data center operators have not yet faced equivalent scrutiny. Giving oil veterans the keys to the next decade of AI infrastructure risks importing the opacity practices that made Big Oil a political target. Civil society groups will press for transparent power purchase agreements, published acceptable use policies, and third party environmental audits as a baseline. Those demands are reasonable, and operators who meet them will have an easier time with permits and politics, as aiplusinfo’s coverage of AI and climate change underlines.

A separate dimension is personal accountability tied to the executive at the center. When a former BP chief links with AI data center work, Looney’s BP exit remains part of the public record. The BP board classified it as serious misconduct, and the finding is unlikely to recede soon. Prometheus has chosen to look past that history to secure the operational depth he brings. That is a legitimate corporate choice, but it is one that invites continued scrutiny from reporters, policy analysts, and institutional investors. Transparent governance, independent board committees, and clear executive performance frameworks are the way Prometheus manages this. The company’s willingness to publish those structures will shape perceptions over the first eighteen months.

What Hyperscale Operators Can Learn From the Prometheus Playbook

Beyond the headline deal, there are at least three operational lessons from the Prometheus playbook that other hyperscale operators can borrow. Behind the meter construction, long dated nuclear hedges with vendors like Oklo, and closed loop water with aquifer monitoring form a reproducible template. The pattern echoes AI solutions to cut energy use and emissions on aiplusinfo. The template reduces time to first megawatt and lowers early execution risk for operators. It locks in a clean baseload option over the long term through the Oklo nuclear hedge. The template also defuses the biggest community objection to large campuses by using closed loop aquifer water. Operators who ignore any of the three risk a messier siting cycle and worse financing terms.

The second lesson is about executive leadership of hyperscale programs over long construction schedules. Hyperscale projects need program managers with multi billion dollar capital experience, and the technology industry has been chronically short of them. Partnering with or hiring experienced energy executives is one direct way to close that gap quickly. The third lesson concerns disclosure practice at the project level across energy and water. Projects that publish their energy, water, and workforce metrics early tend to earn political capital that smooths later phases. Prometheus has adopted that posture at least in part, and the industry will watch how far it carries the practice through construction. Operators who publish nothing create a vacuum that activists fill with their own numbers.

The Future of AI Data Center Siting in the American West

Looking ahead, the Prometheus thesis points to a broader shift in where AI campuses will be built this decade. Wyoming, New Mexico, North Dakota, and parts of Idaho offer the exact combination of land, gas, wind, and workforce depth that gigawatt scale AI compute needs. Analysts tracking the sector increasingly describe this as a new Western power corridor, a frame reinforced by the Yahoo Finance coverage of Looney taking over the Wyoming developer. The corridor’s defining feature is cheap behind the meter power paired with early stage nuclear, which is unavailable at scale in traditional East Coast and Pacific Northwest hubs. Expect at least four comparable projects to be announced before the end of 2027.

Policy implications follow the geography of the new Western AI power corridor directly. States that offer streamlined air permits, workable water rights, and expedited interconnection will capture a disproportionate share of the capex. Those that do not will cede both tax base and workforce opportunity to neighbors. Wyoming is the furthest along in using the Prometheus deal as a template, and the state has begun positioning itself as a destination for similar projects. New Mexico and North Dakota are following close behind Wyoming on these same policy levers. Readers can trace adjacent policy stakes through aiplusinfo’s reporting on Meta’s derailed nuclear bee crisis.

The pattern of a former BP chief links with AI data center style move is only just forming across the industry. The next twelve to eighteen months should bring two or three high profile moves, mostly from upstream and midstream firms whose core skills translate directly. The reputational halo around Looney’s appointment will fade quickly if Prometheus hits its milestones, which will accelerate the trend. If Prometheus stumbles, the opposite happens, and the technology industry will be more cautious about these imports. Either way, the former BP chief has already shifted the conversation about who runs AI infrastructure. The lessons for the industry are permanent regardless of how Prometheus’s first commercial operation turns out.

Key Insights From the Prometheus Hyperscale Case

• The International Energy Agency projection puts global data center demand at 950 terawatt hours by 2030. That doubling trajectory is what makes a thirty billion dollar Wyoming AI data center bet rational for the current cycle.
• Per the IEA briefing on surging AI compute use, AI focused data centers alone are projected to triple to 465 terawatt hours by 2030. Hyperscale developers are racing to lock in gigawatt scale sites quickly across the United States.
• Data centers already account for 4.4 percent of United States electricity today and could reach 12 percent by 2028. The TechRepublic Wyoming AI power coverage documents the shift in detail for regulators to track.
• Each Wyoming hyperscale facility may use two million liters of water per day, as the TechRepublic AI water analysis quantifies. That matches about six thousand five hundred households based on utility estimates for the region.
• According to Data Centre Magazine, five large United States technology companies invested four hundred billion dollars of capital in 2025. They plan a further 75 percent capex increase during 2026 across the United States technology sector.
• The TechRepublic Wyoming power report traces United States AI infrastructure requirements to two trillion dollars by 2030. The forecasts are drawn from utility and industry sources that regulators across the sector cite.
• Looney was denied roughly thirty two million pounds, equivalent to about forty million dollars, in pay and share awards after his BP exit at the oil major. Data Center Dynamics confirmed this detail in its coverage of the Prometheus Hyperscale CEO appointment story.

Those seven data points fit together as a sharper thesis than any Prometheus press release has acknowledged. Energy demand from AI compute is on a doubling trajectory that favors operators who can land land, gas, and water at gigawatt scale. The United States grid cannot absorb that growth without careful behind the meter construction, which is exactly the model Prometheus is pioneering in Wyoming. Capital is pouring into AI infrastructure faster than the construction and workforce base can absorb it, which pushes an advantage toward disciplined operators. The Looney hire is a bet that oil industry program management can bend that curve in the project’s favor. The market will know by 2028 whether the bet pays off.

How Prometheus Compares With Other AI Hyperscale Programs

The former BP chief links with AI data center work in a landscape already crowded with hyperscale programs. The comparison matters for anyone sizing up the Wyoming bet. The table below sets Prometheus beside the major United States AI campuses on capacity and generation. It also compares cooling, capital stack, and community pressure points. The differences read at a glance for anyone who tracks the sector today.

Program dimension	Prometheus Hyperscale (Wyoming)	xAI Memphis Colossus	Meta Louisiana (Richland Parish)	Microsoft Mount Pleasant (Wisconsin)	Oracle Abilene Stargate
Announced initial capacity	1.2 to 1.25 gigawatts per site	~200 megawatts expanding to ~1 gigawatt	2 gigawatts at full buildout	~450 megawatts at phase one	Up to 1.2 gigawatts at phase one
Headline pipeline	~10 gigawatts across US sites	Memphis plus adjacent sites	5 million square feet Louisiana campus	$3.3 billion Wisconsin investment	Reports of 1.2 to 2 gigawatt total
Primary generation	Natural gas, behind the meter, nuclear and renewables layered	Grid plus on site gas turbines	Grid with Entergy, new gas plants and solar PPAs	Grid utility, utility scale solar PPAs	Grid plus natural gas
Nuclear strategy	Oklo twenty year PPA letter of intent	None announced	None in initial build; separate Meta nuclear track reported	Microsoft Three Mile Island restart PPA	None announced
Cooling design	Direct liquid cooling, closed loop aquifer water	Mix of air and liquid	Direct liquid, open loop regional water	Mostly liquid, utility water with reclaim	Advanced liquid cooling at scale
Capital and governance signal	Private equity plus In-Q-Tel board seat	Elon Musk vehicle, private	Meta corporate balance sheet	Microsoft corporate balance sheet	Oracle plus SoftBank plus OpenAI consortium
Community pressure point	Electricity consumption vs residential load	Methane turbines, air quality permits	Water use and tax abatement	Water usage disclosure	Local power price impact

Real World Examples of Oil Executives Entering Technology

Bernard Looney at Prometheus Hyperscale

Looney stepped into Prometheus Hyperscale as non executive chairman in late 2024, then took over as chief executive in April 2026. He deployed capital into a thirty billion dollar Wyoming buildout tied to the Oklo twenty year letter of intent. The Evanston site was revised from 120 megawatts to a 1,200 megawatt design, roughly a 900 percent capacity increase. The measurable outcome is a pipeline that could approach ten gigawatts across United States sites, as detailed in the Cowboy State Daily thirty billion dollar buildout profile. The clear limitation is that the project still relies on natural gas at commissioning. Climate critics continue to challenge the sustainability framing of the first phase. The example captures how fast an energy veteran can reset a hyperscale plan.

John Browne and the Return of Energy Capital to Deep Tech

Lord John Browne is Looney’s distant predecessor and the former BP chief executive from 1995 to 2007. He has deployed energy capital into deep technology ventures through his investment vehicle BeyondNetZero since 2021. The portfolio crossed a one billion dollar deployed threshold during 2024, a 40 percent increase versus 2022, as the Recharge News profile of oil chiefs documents. Returns depend on the pace of grid interconnection and permitting as the measurable outcome. Both have slowed by roughly 15 percent since 2023, which is the clear limitation. Browne has publicly argued that AI campuses need gas and nuclear alongside renewables, a framing contested by pure play climate investors. The example shows the former oil leader to AI pipeline is not new.

Darren Woods and ExxonMobil’s AI Data Center Power Pilots

ExxonMobil, under chief executive Darren Woods, has not jumped seats like Looney but has rolled out an AI power supply strategy. The company deployed a plan to build a 1.5 gigawatt behind the meter power plant with carbon capture. ExxonMobil told investors the plant will deliver firm, lower carbon power to a hyperscale customer by decade end. The measurable outcome is a projected 25 percent reduction in lifecycle carbon versus unmitigated gas, per EnkiAI’s analysis of oil and gas AI infrastructure moves. The clear limitation is that gigawatt scale carbon capture has not been proven commercially. Critics argue the plant may become a stranded asset if customer demand softens. The example shows corporate strategy running parallel to executive mobility across the sector.

Case Studies of Hyperscale AI Infrastructure Projects

Case Study: Microsoft's Three Mile Island Nuclear Power Purchase Agreement

Microsoft faced the same core problem Prometheus is solving: sourcing firm, low carbon electricity at AI scale without waiting a decade. The company developed a solution in 2024 by signing a twenty year power purchase agreement with Constellation Energy. The deal restarts Unit 1 of the Three Mile Island nuclear plant in Pennsylvania to deliver roughly 835 megawatts of continuous clean power. The measurable impact is hundreds of millions of dollars per year of firm clean capacity once commissioned in 2028. The Data Centre Magazine analysis of AI driven electricity demand details that outlook. The limitation is that Three Mile Island carries one of the most famous nuclear accident legacies in United States history. Pennsylvania residents and policy groups have raised safety and precedent concerns, and licensing is still contested. The case illustrates the appeal and the controversy of long duration nuclear hedges for AI campuses.

Case Study: xAI's Memphis Colossus and the Gas Turbine Backlash

xAI's Memphis Colossus faced the problem of bringing an AI campus online without waiting for utility capacity to be built. The company built a rapid solution in 2024 by installing roughly 35 on site natural gas turbines generating 200 megawatts of behind the meter power. That rolled out training operations within a year of site selection. The measurable impact was dramatic, with Colossus going from idea to 100,000 GPUs online in roughly 122 days. The limitation and controversy is that the turbines were installed before full environmental permits were approved. The Southern Environmental Law Center and local groups sued over methane emissions and air quality, as the AI data centers drive gas surge coverage describes. Memphis residents raised concerns about health impacts and transparency throughout 2025. The case shows how speed can override process at hyperscale.

Case Study: Meta's Louisiana Richland Parish Campus and Water Trade Offs

Meta's Louisiana campus faced the problem of a rural site with limited existing infrastructure and major community sensitivities. The company launched a two gigawatt, five million square foot Richland Parish campus in late 2024 as its solution. Meta rolled out a utility partnership with Entergy that developed new gas plants and solar power purchase agreements sized to the load. The measurable impact includes committed capital over ten billion dollars and roughly 500 permanent jobs plus thousands during construction. The Blackstone data centers navigating AI power challenges coverage frames the private capital context. The limitation and controversy centers on local tax abatements, water usage disclosure, and whether residential ratepayers will subsidize new gas plants. Louisiana consumer advocates argued the Public Service Commission approval moved too fast. The case illustrates why Prometheus has stressed closed loop water as a response.

Frequently Asked Questions on the Former BP Chief and AI Data Centers