Less than a year after securing one of the Frontier Coalition’s largest offtake agreements, carbon removal startup Arbor is inking its first deals to deliver turnkey power to data centers.
Starting in 2029, Arbor will build and develop up to five gigawatts of its 3D-printed, modular, fuel-agnostic turbines to GridMarket, a platform that matches clean energy projects with sites and buyers. The turbines will largely run on fossil gas, rather than the waste biomass that was central to the company’s initial carbon removal pitch. In both cases, the turbine turns the fuel into syngas, which burns with pure oxygen to heat carbon dioxide and spin the turbine blades, creating electricity. The CO2 then cools and is recompressed, before being captured and stored underground.
As part of the agreement with GridMarket, Arbor will both manufacture the turbines and build the initial projects, rather than passing things off to a traditional energy developer, said CEO and co-founder Brad Hartwig. “At the beginning, we’re working with data center developers…so that we can transact for power, rather than turbines,” Hartwig, a former SpaceX engineer, explained.
“We are the most invested in our success and bringing our technology to market, and also our engineers are the most intimately knowledgeable of the technology itself,” he added. “So we see Arbor at the beginning as a one-stop-shop for clean baseload power.” In the longer-term, however, the goal is for Arbor to become an OEM turbine manufacturer, which could allow the company to scale more quickly.
Arbor’s pitch is for a baseload data center power solution that is faster, cleaner, cheaper, and more flexible than traditional gas turbines. The company’s modular, fuel-agnostic turbines are small: 25 MW each, compared to the 500 MW of a standard large-frame unit. Adding additional systems for reliability is prohibitively expensive when using traditional turbines, but is a core element of Arbor’s approach to data centers.
For a 500 MW data center, Hartwig explained, adding a few additional 25 MW turbines for redundancy is a relatively small capital expenditure. And compared with how long it can take to secure a turbine from a major manufacturer like GE Vernova for example, Arbor can get power to its customers faster; that speed, according to Hartwig, is often their “first, second, and third priority.”
That said, it’s still early days. Arbor is just now ramping up its own manufacturing, and is a long way from scaled deployments. Late last year, the company raised a $55 million Series A to complete its 1-MW pilot and begin commercial deployments of its 25-MW turbines. Arbor is also building a demonstration plant in Louisiana, which will run on waste biomass and deliver carbon removal credits to its Frontier buyers.
The company’s first turbines are slated to come online in 2028, Hartwig told PowerMag earlier this year; by 2030, the company plans to ship more than 100 turbines annually, representing over 1 GW of new capacity each year.
The carbon removal angle
Arbor got its start as a carbon removal company that offered energy as a byproduct. Waste biomass, which it used to power the turbines, contains recently-absorbed carbon dioxide. Using it as fuel results in net-negative emissions after capturing the stream and storing it underground.
And while the company has certainly pivoted hard in light of the data center boom, it isn’t entirely turning away from the carbon removal market. The company recently hired Nishad Pai, formerly head of business development at direct air capture company Heirloom, to serve as Arbor’s chief commercial officer.
Pai’s primary role initially is to segment Arbor’s market and prioritize which verticals the company will target, and in which order. Hyperscalers and the AI boom are, of course, first up. But Pai explained that another priority segment is broader industrial power demand, including industrial and manufacturing processes.
Carbon removal is still an attractive offering for many potential customers, particularly those with ambitious sustainability goals, Pai said, like the major hyperscalers. And of course the Frontier deal will rely entirely on the waste biomass approach. But those deals are more likely to materialize on longer timelines — for example to help companies reach 2050 carbon negative goals, rather than 2030 carbon neutral goals.
In the near term, more of Arbor’s demand will come from the fossil gas configuration, Hartwig said. That’s partly due to the speed and scale of demand; the company’s demonstration facility will source biomass from thinnings of managed timber plantations, in line with Frontier’s sustainable biomass sourcing principles. That requires areas where the amount of biomass grown versus harvested is either stable or increasing, and where the pulp and paper mills that historically use the biomass have shut down.
That means lower fuel availability, Hartwig acknowledged, which would inevitably create a bottleneck for data center power deployments at scale.
“What we see is actually a real benefit in being fuel flexible on the customer side as well, because they have the option… In some cases, power from the biomass will actually be cheaper, but it’s more complex, and not everyone is interested in carbon removal,” he explained. “In either case you are storing the CO2 in a class six well. So in both cases, it’s a power plant without a smoke stack, effectively.”
Transition-AI 2026, Latitude Media’s flagship event at the center of the AI-energy infrastructure buildout, will convene developers, utilities, regulators, hyperscalers, and capital providers to move towards coordinated, sustainable solutions for AI-driven demand. Register now to join us in San Francisco April 13-14.
