The idea arrived as a small outrage. Andrew Webber was reading about California producing so much electricity that it paid Arizona $20 per megawatt-hour to take the excess. To a former Wall Street investor, the spectacle looked less like an unavoidable quirk of physics than a market begging for another buyer. His reaction, later recalled on The POWER Podcast, was wonderfully direct: he would take it. Put a bitcoin mine beside the unwanted power, let the computers consume what the grid could not, and stop pretending an electron without a customer had no value.
That flash became the operating thesis of Digital Power Optimization, the New York company Webber founded in 2020. DPO brings computing loads to power plants and other generation assets where electricity is stranded, curtailed or simply underpaid. The computers run when the economics make sense. When customers need the electricity more, or grid prices rise, a flexible bitcoin-mining load can shut down. The humble on-off switch becomes a financial instrument.
Webber prefers to describe the business as energy services, not a crypto wager. The distinction matters. A conventional miner begins with bitcoin and hunts for cheap power. DPO begins with the generator, studies its capacity, costs, price curve and operating limits, then designs a data center around the asset. Computing is the customer of last resort, available on site and capable of disappearing quickly. In a market usually obsessed with uninterrupted uptime, DPO found value in knowing when to leave.
Before the electrons, the capital stack
There was nothing inevitable about Webber ending up near turbines and substations. He grew up on the north shore of Maui, served four years as an enlisted active-duty U.S. Marine and spent several of those years at the Pentagon. He then attended the University of Iowa's Tippie College of Business from 2003 to 2007. New York came next, first through investment banking at Goldman Sachs, then equity investing at Surveyor Capital and Steadfast Financial, and finally credit investing at Fortress Investment Group.
Across roughly twelve years in finance, he analyzed companies, traded securities and structured bespoke private-credit agreements. The work taught him to look at an asset from several angles at once: cash flow, downside, control, timing and the terms that divide risk among partners. Those habits survived his move out of finance. DPO's pitch is full of hardware, but its skeleton is a deal. Who owns the computers? Who supplies the capital? At what electricity price should the machines stop? How does a generator share in the upside?
His bridge into the sector was searCH4power, a cryptocurrency-mining operation that used flared gas as an energy source, where Webber served as CFO. The experience introduced him to bitcoin mining as a movable industrial load. The newspaper story supplied the larger application. Electricity might be difficult to store and expensive to transmit, but computation travels. Instead of moving the electron across a crowded grid, move the machine toward the electron.
“It's sort of energy management infrastructure.”Andrew Webber, The POWER Podcast
The business of the off switch
DPO's first announced deployment began at a partner facility in Hennepin, Illinois, in November 2020. A larger public test followed at a hydro facility in Hatfield, Wisconsin, where DPO deployed and manages a 1.8 MW bitcoin-mining operation. Hydropower is a neat companion for this kind of load. Seasonal output is relatively predictable, Webber explained in a 2026 Hydro Leader interview, and a small hydro asset can support a mining operation at a scale far below what an AI campus usually demands.
The operating logic is almost theatrical. During ordinary periods, the data center buys power and produces computing revenue. On a hot day, when air conditioners roar and wholesale electricity prices spike, the miners can go dark. The generator sells its output to the grid instead. In Webber's account, every electron at the Hatfield facility remains available to the grid because the mining load can be turned off at a moment's notice.
When the grid calls, flexible compute steps back
Conceptual operating pattern - illustrative, not plant data
The arrangement is not frictionless. Regulations differ by state. Ownership structures matter. The plant, data center, interconnection and local utility all impose constraints. Nor are all computing jobs interchangeable. Bitcoin mining can tolerate remote geography, modest internet connections and abrupt interruption. AI computing generally wants steadier power, excellent fiber and far denser technical infrastructure. Webber's method is less about declaring a universal winner than matching each load to the physics and economics of a specific place.
Same electricity, more demanding customer
Bitcoin gave DPO its first tool and its first headlines, but the company gradually widened the proposition. In 2024, it announced a partnership with Schneider Electric to pursue modular, wind-powered AI data centers and signed a behind-the-meter wind power agreement for a proposed 100 MW green data center. In January 2025, DPO announced a $200 million high-performance computing development in Wisconsin Rapids, planned for up to 20 MW and supported by land, an energized substation and firm power through Billerud subsidiary Consolidated Water Power Company.
These projects preserve the order of operations: power first, computing second. Yet AI changes the bargain. A bitcoin miner can vanish when prices jump. An AI customer expects a more consistent supply and will pay more for it. Webber has called the two workloads different animals. The interesting work lies in the translation - sorting sites by generation pattern, interconnection, fiber, acreage and time to market, then deciding what kind of computation can honestly live there.
That thinking produced a new corporate branch in September 2025. Greenlight Data Centers launched as a separately focused AI-site development platform, with DPO co-founder Alex Stoewer as CEO and Webber as chairman. Greenlight's public strategy targets roughly 20-to-100 MW locations with usable fiber and existing substations, often below the enormous campuses that dominate industry announcements. Webber continued as DPO's CEO, keeping its flexible bitcoin work and related energy partnerships moving alongside the AI platform.
“There isn't one right solution for meeting data center energy demands.”Andrew Webber, on renewable power purchasing
The split is revealing. Bitcoin remains valuable precisely because it can be remote and interruptible. AI is attractive because demand for high-density rack space has grown, but it is picky. Treating both as generic “data centers” would erase the quality DPO sells: a close reading of place. The wind does not follow a model-release calendar. A river has wet and dry seasons. A substation has a real capacity, not a keynote number.
Optionality, rendered in steel and silicon
Webber's achievement is not a single mine, dam or proposed AI facility. It is a way of looking. Power generation was traditionally built around a limited menu: sell into the grid, sign a long-term contract, curtail production or wait for transmission. A colocated computing load adds another path. It gives the generator a local customer whose appetite can, in bitcoin's case, be shaped minute by minute.
That view has also moved from pitch deck to intellectual property. In August 2026, a U.S. patent assigned to DPO named Webber and John Potts as inventors of a system that tests available variable power and controls how many computing systems operate. The patent formalizes the premise that started with an annoyed newspaper reader: computers can follow electricity rather than demand that electricity follow them.
Webber is still candid about where he comes from. “I'm by and large a finance guy,” he told Hydro Leader. The sentence is modest, but it also explains his usefulness. He entered an engineering-heavy field fluent in pricing, contracts and capital. He can speak to a hydro owner about seasonal production, then to an investor about cash flow, then to a data-center audience about workloads. Translation is an operating skill when each tribe owns one part of the same project.
Away from the infrastructure circuit, his company biography offers a softer coordinate: he lives in the New York City area and escapes to the Adirondacks whenever possible. The contrast fits. His work lives between dense financial markets and remote physical assets, between Manhattan terms and water moving downhill. The machines are loud. The thesis is quiet. Find energy the market has overlooked, give it a useful second customer, and preserve the freedom to choose.