Christopher Helman

Forbes Staff

Source: Forbes

Nate Franklin, CEO of Pacifico Energy, has a grand plan to build the nation's largest electric power generating complex on 8,400 acres of flat West Texas scrub desert in desolate, sparsely populated Pecos County (14,000 people spread over 4,800 square miles). He’s got an option on the land and has already received air emission permits from the Texas Commission on Environmental Quality giving Pacifico the green light to erect dozens of natural gas turbines, generating 7.5 gigawatts of power. The complex he envisions will also include solar panels producing 750 megawatts of power, plus batteries with a storage capacity of 1.8 gigawatt hours and will put out sufficient juice to power five million Texas homes (or enough for all of New York City on most days).

What Franklin doesn’t have yet is the $12 billion he estimates it will take to build the power complex or a commitment from the hyperscalers (the big cloud service providers like Amazon, Microsoft, and Google) to build AI data centers to consume all that energy. Those commitments are needed to shake loose the capital to build Franklin’s dream. “The demand for this intelligence will justify all the power projects underway now,” says Franklin, who is confident he’ll get those commitments.

The project, which Franklin has christened GW Ranch, might not be as farfetched as it sounds considering that the hyperscalers are expected to spend $650 billion in 2026 alone and that Elon Musk and others are talking about putting AI data centers in space to run on solar power.

Moreover, at a time when some local communities are rejecting data centers in part because of their energy consumption, Franklin intends to keep costs (and opposition) low by staying entirely disconnected from the state power grid run by ERCOT, the Electric Reliability Council of Texas. He was drawn to this spot by what he refers to as “the cheapest most abundant gas in the world.” A few miles away is the Waha hub, an intersection of a dozen pipelines that collect natural gas produced from surrounding oilfields. The gas is a byproduct of more lucrative oil drilling. When oil prices are high and rigs are busy, the price of gas at Waha can go negative. Why not build the nation's biggest power plant if you can get your fuel cheap or even for free?

If it gets built, the project could power some $200 billion worth of AI supercomputers — millions of square feet filled with racks of GPUs from the likes of Nvidia, says Franklin. “Certain customers want the ability to scale larger and larger,” he observes. “Two years ago we were talking about tens of gigawatts, and now it’s hundreds.”

Needless to say, this is a high stakes gamble. Despite having no customers, Pacifico already has gas turbines on order and expects its first deliveries late this year, with an eye to ramping up an initial 1 GW of gas-fired power in 2028. So far Franklin has funded startup costs in-house. But the big dollars will have to come from others, perhaps (given Franklin’s unusual, international experience), Japanese companies.

Franklin 46, was born into energy. He grew up in Bakersfield, the center of California’s once thriving oil industry. Grandpa and dad worked in the Kern River oil field, which used to be one of the biggest producers in the nation. At one point they owned wells producing 1,000 barrels per day, and their home was near the Panorama Bluffs, which overlooked the operations. “At night it looks like a city,” he says.

After earning an economics degree from the University of California, Berkeley and an MBA from UCLA, he worked at Edison Mission Energy, Jaco Oil, and BP’s solar company. Then the earthquake and tsunami that caused a meltdown at Japan’s Fukushima nuclear reactor changed his trajectory. The disaster led to a huge shift in Japan’s energy policy as many reactors were shuttered and the government introduced a new law in July 2012 that required utilities to buy from renewable sources at fixed, above-market prices for long-term contracts.

Sensing an opportunity, Franklin moved with his wife and three young kids to Tokyo in 2012, and started Pacifico Energy, initially operating out of a two-person office. Soon he was developing projects like Kumenan Mega Solar, an installation in mountainous terrain, with backing from the likes of General Electric, Bank of Tokyo-Mitsubishi UFJ, and a syndicate led by Sumitomo Mitsui Trust Bank. With the help of Nomura and Goldman Sachs, he raised a $142 million solar fund in 2017 and another for $266 million in 2019. With more than 20 solar projects, Pacifico grew into Japan’s biggest developer.

Franklin moved back to the U.S. in 2019, but kept up with his contacts, convincing Sumitomo and Mitsubishi UFJ to fund $90 million in equity and loans for 27 MW of solar and battery installations in California and Massachusetts (sold last year to independent power producer CleanCapital). “By 2023 it was clear that AI data center growth was accelerating,” says Franklin. That’s when he turned his attention to the Texas site.

Not everything has gone smoothly for Franklin and Pacifico. He’s still looking for backing for a 3 gigawatt windfarm project approved offshore of South Korea. And Pacifico has lowered expectations for a sale of Japanese solar assets after potential buyers including KKR and Macquarie reportedly balked at the initial $1 billion price tag.

Franklin says Pacifico doesn't need to sell assets to fund early work at GW Ranch. Nor is he presumptuous enough to assume that his Japanese friends will step up to back the orders-of-magnitude bigger GW Ranch. But he's making the case to them—and there’s reason to think they might be interested.

Masayuki Nagano of Daiwa Securities points out that Japanese energy giants like Jera, Tokyo Gas and Japex have already spent more than $15 billion acquiring U.S. assets in the natural gas value chain. Most recently, Mitsubishi agreed to buy private gas producer Aethon Energy for $7.5 billion. Moreover, last July, in tariff talks with the Trump administration, diplomats agreed that Japanese companies would invest $550 billion into the U.S. by the end of Trump’s second term. Just this week Trump announced $36 billion of preliminary investments into energy and mineral projects in Ohio, Texas and Georgia.

Then there’s the unusual circumstances of the GW Ranch, with its emissions permits, gas rich location and lack of local opposition. “There's not many places you can do these kinds of projects,” says Franklin. “To be able to have that much energy in one place is rare.”

Franklin won’t specify exactly how much Pacifico has invested in the project so far, or exactly what his “significant” stake is in privately held Pacifico. He has been successful enough that his family foundation funded the construction of a $13 million aquatics center in his hometown of Bakersfield and a community beautification in San Juan Capistrano in Orange County, California, where he lives now.

A key that just might make this work: Franklin’s decision at the beginning to make GW Ranch into its own private power grid, disconnected from the ERCOT grid of Texas. Smart, says Ed Hirs, a lecturer in energy economics at the University of Houston, who figures that Pacifico will be able to offer GW Ranch customers electricity at 10 cents per kilowatt hour cheaper than on the grid by avoiding ERCOT's transmission fees and local utilities’ fixed distribution charges. At a time when AI’s energy needs are massive and the build out is rapid, Pacifico will also save a year or so by not having to wait in ERCOT's crowded interconnection queue filled with other data center projects. The Texas grid manager forecasts electricity demand from new data centers across the state will grow tenfold by 2031 to 24 gigawatts.

Other big off-grid projects aiming to power Texas data centers include Chevron’s planned 2.5 GW gas plant in the Permian basin, the 1.2 GW Sandow Lakes plant in central Texas and the Hays Energy 1 GW plant in San Marcos (each just a fraction of what Pacifico has planned).

The biggest competitor to watch is being built by Dallas-based Fermi America in Amarillo, Texas. Dubbed the Donald J. Trump Generating Station, it is adjacent to the Department of Energy's Pantex plant for manufacturing plutonium pits for nuclear warheads. Fermi raised $680 million in an IPO last year then shot up to $20 billion in market cap, briefly turning two of its founders, Toby Neugebauer and Griffin Perry, into billionaires. Its market cap is now down to $5 billion.

Fermi generated a lot of hype with its plans to erect 11 GW of new nuclear reactors, but those won’t be built for a decade or more. In the near term Fermi has broken ground on a more modest plan for 6 GW of natural gas turbines that would cost more than $10 billion. In early February it announced that its first Siemens turbines had arrived at the port of Houston. But just days later it admitted it was halting construction work at Amarillo because it hadn’t yet finalized its permits from the Texas Commission on Environmental Quality (the same agency that has approved permits for the GW ranch).

Here’s another strand from the Fermi story that offers hope for Franklin and Pacifico: Showing its interest in Texas data centers, Japan’s Mitsubishi-UFG has so far loaned Fermi $500 million. Franklin says Fermi isn’t a competitor because the U.S. will need many more such projects.

In contrast to Fermi, Franklin isn't looking at nuclear power right now. Another contrast: He waited until Pacifico had all its emissions permits to even announce GW Ranch. Others “claim they can scale through future phases. But those are never guaranteed,” he says.

Indeed, there’s a move afoot to slow down the data center advances. Six states are considering bans or moratoriums; New York’s would halt new builds for three years. Senator Bernie Sanders is pushing for a national moratorium, though Franklin considers such a sweeping ban unlikely. He’s focused on demand, instead.

Franklin likens the current state of AI adoption to the America Online dial-up-modem era of the Internet. “This is an unprecedented load growth story,” he says, “and it's going to take a lot more than one or two projects to keep up with the scale and speed of what's coming.”