Summary
- Nebius is named as the first prospective offtaker at Volt’s proposed Rotterdam campus.
- The developer now describes six buildings and 600MW, with two 100MW buildings in the first phase.
- A letter of intent does not establish a commissioned facility or fully financed scheme.
Nebius has signed a letter of intent with Volt to take capacity at the developer’s planned AI data centre campus in Rotterdam. Volt describes the project as a six-building facility with an ultimate ambition of 600MW, and has identified an initial phase of two 100MW buildings. The agreement introduces a prospective customer to a development that remains at an early stage; neither a signed operational lease nor a commissioned power connection has been established in the announcement.
The revised scale is material. When Volt first outlined the Rotterdam project, it referred to as much as 800MW. Its current 600MW proposal should therefore be treated as the developer’s updated ambition, rather than a number already available to customers. The distinction also applies to the first phase: a 200MW development target is a proposed construction programme, not evidence that 200MW of IT equipment or grid capacity is in service. No timetable for placing the first halls into commercial operation was verified from the available reporting.
Volt has put the proposed total investment at approximately €7 billion, comprising €2.5 billion for the buildings and related physical infrastructure and €4.5 billion for computing equipment. Those amounts are project estimates, not reported completed expenditure or a committed financing package. A prospective offtaker can improve discussions with investors by demonstrating customer interest, but that does not settle construction risk, utility contracts, planning permission or the many individual procurement decisions involved in a campus at this scale.
Grid access is particularly important for a port-area scheme that would need large, reliable electricity supplies. A data centre’s quoted MW figure may refer to an intended campus power envelope rather than the IT demand at the racks. Facility cooling, electrical losses and ancillary systems add to the amount of power that must be delivered. The reporting available so far does not establish whether Volt has secured an energisation date or firm connection rights for the full campus, so those milestones should remain open questions.
Rotterdam’s industrial location could offer advantages in logistics and the availability of established energy infrastructure, but proximity to industrial electricity assets is not the same as usable spare grid capacity. Large developments also have to resolve standby power, cooling technology, physical security, fibre connectivity and the sequence in which buildings can be handed to a customer. These elements often advance on different timetables, with several years between an initial commercial agreement and the first revenue-producing operating hall.
Nebius is an established AI cloud infrastructure business, which makes the offtake announcement commercially more concrete than an unnamed-customer prospectus. Still, the contract is described as a letter of intent. Its binding obligations, precise capacity allocation, power-delivery terms and remedies for delay have not been publicly established. Volt’s intention to reserve part of its campus for European users likewise remains a development proposition until units are built, connected and leased.
The structure of the proposed investment will matter as much as its scale. Building shells, installing electrical infrastructure and purchasing accelerators each carry different risks and asset lives. If the development is phased, a first customer may commit to one building while later buildings wait for further commercial demand. Reporting on the proposed €7 billion envelope should therefore distinguish between development financing, tenant-funded computing equipment and obligations that have not yet become enforceable. There is no disclosed schedule showing when each portion of the quoted spending would occur or which organisation would own the equipment.
A cloud operator leasing wholesale power and space is also different from an enterprise buying individual cloud instances. The former needs defined electrical, cooling and network service levels from the landlord; the latter relies on the operator to deliver those services through its computing platform. The existence of a prospective wholesale tenant does not establish what prices or service guarantees might eventually be available to end customers. Nor does it demonstrate the amount of computing capacity that will be accessible to Dutch or European businesses rather than reserved for contractual workloads.
Construction sequencing is another test of credibility. Earthworks, foundation engineering, delivery of large transformers and commissioning of switchgear may have long procurement lead times. These can constrain the delivery of even a well-funded hall. Server installation follows acceptance of the physical environment, and stress testing must verify the response of the power and cooling systems under realistic load. The signing of a letter of intent is therefore commercially relevant without collapsing these distinct construction stages into a single promised launch.
The next useful milestones are a definitive customer contract, verified planning and grid approvals, financing completion, and the appointment of contractors for the first 100MW buildings. Each would reduce a different form of delivery uncertainty. Until then, the 600MW figure describes a possible future Rotterdam infrastructure asset and the letter of intent a step towards demand validation.
