O-Green's 200MW target is demand for its own electrons

Oman's O-Green says it will have more than 200MW of AI-ready data center capacity ready for service by 2028 across the sultanate and European markets that include Spain, and the project book behind that target shows how the same number can be a credible power plan and an unproven compute one. Founded last year, the firm holds a secured portfolio topping 3.3GW of wind and solar plus more than 2.3GWh of battery storage across Oman and Botswana, while its power division describes a pipeline spanning 12 countries and exceeding 11GW.

Shareholders include state-owned Naqaa Sustainable Energy and OQ Alternative Energy, and the group runs the chain from nacelle to rack: O-Green Power develops green energy projects abroad, O-Green Technologies manufactures wind turbines, and O-Green Compute & Data Centers builds the AI-ready, cloud and hyperscale capacity that the firm says will be powered by low-cost renewable energy in Oman and Europe. That structure is why the 200MW number is more than a capacity target; a developer that owns generation, storage and the turbine factory can finance a data hall against assets it controls, which is a different conversation with a lender than the one a merchant developer has.

That structure puts O-Green inside the trade this publication has argued is now central to digital infrastructure: the AI buildout's bottleneck has shifted from land and chips to electrons. Firms that spent the past two years racing for land and interconnection are now negotiating for firm power, and the ones that already control generation have found a cheaper way to monetize it: build the load behind the meter, keep the generation margin, and sell compute in place of kilowatt-hours. O-Green is among the few developers in this market positioned to attempt the whole stack without buying power from a third party.

The evidence of intent stands at Duqm, where the company has installed what it claims is the largest onshore wind turbine outside mainland China, close to the OQ8 refinery and presented as a landmark of Vision 2040, the sultanate's long-term plan to move its economy away from oil. What remains unsaid is whose turbine design O-Green Technologies manufactures under license, if any, and how much of the 11GW pipeline sits at financial close rather than in development.

What the announcement does not carry is nearly everything a lender or a tenant would want to see: no capital cost, no site list, no named counterparty, no power price, and no indication of what share of the 200MW is contracted. The target arrived as a LinkedIn post from a company founded last year, which makes it a capacity aspiration with a date attached. A target with a blank where the price belongs is a financing flag rather than a construction schedule, and 2028 is near enough that the flag gets tested on a fixed clock.

Raising construction debt against uncontracted capacity in 2028 is a harder pitch than it was in 2024, and the silence on offtake is the single largest thing standing between the Omani leg and a financing. Oracle's fixed 433MW RWE contract, which swapped procurement optionality for an obligation, is the opposite posture; the distance between the two is the distance between an infrastructure asset and a stated ambition.

Oman is not empty ground: DataCenterMap counts 15 data centers in the country, two of them Equinix sites in Muscat and Salalah, and Oracle launched its second OCI Dedicated Region there last year. The only other project named at hyperscale is a 150MW AI development agreed in May by Omani, Emirati and Italian companies, which would make O-Green's target the larger of the two named builds, and the announcement gives no capacity figures for the 15 existing sites and no schedule for either new project.

Europe is the harder half. In Oman, O-Green has a state shareholder, a domestic generation fleet, a turbine factory and a national strategy pointing the same direction; in Spain it would be a late entrant bidding for interconnection capacity and permits against developers who have held queue positions for years, and the announcement does not say whether it holds either. In emerging markets the scarce input is a local partner who has already cleared ground, power and permits, and an entitled site is a product in its own right. Those advantages do not travel; a Spanish campus is won on queue position, and the announcement does not address it.

State ownership deserves a higher price than Western investors habitually assign it in this sector, and Oman is the case for that view. The sultanate has assembled inputs that consent-constrained markets cannot: submarine cable networks, a national digital strategy, and state shareholders whose standing likely shortens the path through land and power approvals. Where the constraint is consent rather than capital, a state balance sheet is the cheaper instrument, which is why the Omani leg of this plan should be underwritten at a different discount than the European one.

Oman's government has built most of the conditions that make a domestic AI campus financeable; what the 200MW target has not produced is a buyer. The next disclosure on this program is the one that matters: a named counterparty and a price would turn O-Green's compute division into an infrastructure asset with captive generation behind it, while a 2028 capacity figure with no offtake attached leaves it as a place to put the company's own electrons.