Yamna locks in Açu land for 250MW data center

Yamna, a London-based company that has focused on green ammonia and moved into data centers earlier this year, has signed a land reservation agreement at Brazil's Port of Açu that sets aside an initial 20 hectares for a hyperscale, AI-ready campus, according to Data Center Dynamics.

The plot, 280km northeast of Rio de Janeiro, comes with a right of first refusal on another 20 hectares, and first-phase capacity would be 250MW with room to scale; the company also says it aims to attach battery storage to its digital infrastructure sites.

Açu is an operating port, in use since 2014, with 22 companies and 11 private terminals on site, and its chief executive, Eugênio Figueiredo, says the complex offers land, water from multiple sources, clean energy tied to Brazil's National Interconnected System, and the potential for subsea cable landings — four attributes that check the key boxes for a data center site.

Yamna's chief executive, Abdelaziz Yatribi, calls the agreement a milestone for the firm's digital infrastructure platform and says a combination of development, energy and infrastructure expertise, together with local partnerships, will move the project from development toward delivery.

What the agreement does not include is just as important: no customer for the 250MW, no power purchase agreement, no construction timeline, no financing figure. The stated next step is pre-development work intended to reduce project risk — which is to say, nothing has been built and nothing has been sold.

That gap is where data center projects live and die, because interconnection for a 250MW load is a transmission process governed by queue rules, system studies, and utility agreements. The port's clean energy connection to the national grid is an asset only if that grid can actually deliver 250MW to a new user, and the announcement does not show a reserved capacity or a signed grid supply contract.

The site's physical advantages are real. Water from multiple sources is increasingly rare in the data center siting market, and a coastline with subsea cable potential gives Açu a foundation for serving international traffic. But cable landings require licenses, marine surveys, and commercial agreements with subsea operators, none of which appear in the coverage. The port is a candidate location, not a connected one.

This publication has argued that grid access is the binding constraint on data center development, more than capital, and Açu does not contradict that view; it reinforces it. The port can offer land, water, and cable-friendly geography, but the decision to build will turn on whether a hyperscale tenant signs for capacity, whether the grid studies clear, and whether the utility can actually supply the load. Those are the constraints a land reservation does nothing to resolve.

Yamna's move is best understood as an option: the reservation secures a position in a data center market that has already attracted plans for the same complex — Brazilian logistics firm Prumo has previously outlined data center ambitions at Açu — but it does not obligate construction. The company is placing a bet on the site's future value while the project's real merits are tested elsewhere, in procurement rooms and interconnection queues.

A developer without a data center operating record, entering the sector with an Oman project already on the table, is unlikely to win a hyperscale tenant without a site to show; the Açu land gives Yamna something to sell. Whether it is enough depends on the next announcements: a grid study, a power agreement, a construction partner, or a lease with a cloud customer. Until one appears, the plot is a coastal piece of industrial land with good coordinates and a long way to go.

The absence of an anchor tenant is not, by itself, a criticism. Development firms frequently secure land before locking in customers, and the current competition for suitable sites is fierce enough that a port with spare land and grid ambitions has real scarcity value. But the timeline between reserved land and delivered megawatts is measured in years, and the technical work between now and then — interconnection and power procurement above all — will separate the campus that gets built from the announcement that fades.

What would change the math is a power purchase agreement or a hyperscaler lease. With one of those in hand, Yamna's 20 hectares become the foundation of an infrastructure asset; without it, the reservation is a development-stage gamble, priced at the cost of a due-diligence step, far short of the value of a functioning data center. Investors watching this deal should watch for contracts before construction updates, because the first real proof of the project will be a signature on an offtake or a grid agreement, which matters more than the land title.