Summary
- A Dstream Group study commissioned by DE-CIX puts Iberia’s planned data centre pipeline close to 9GW.
- The study counts 35 submarine cables, 20 landing points, 13 Internet Exchanges, more than 100 data centres and more than 800 networks.
- DE-CIX positions Madrid, Barcelona and Lisbon as a distributed regional cluster, but the pipeline still has to convert into connected and operating capacity.
A study commissioned by DE-CIX has put the planned data centre pipeline across Spain and Portugal at close to 9GW, as Madrid, Barcelona and Lisbon become more closely connected parts of the same regional infrastructure market.
The research, produced by Dstream Group, describes the Iberian Peninsula as a distributed digital capital rather than a market dominated by one metropolitan hub. Its case rests on the combination of data centre development, submarine cable landings, terrestrial backbone networks and Internet Exchanges distributed across several cities.
The headline figure needs careful treatment. DE-CIX describes a data centre pipeline approaching 9GW, not 9GW of operating or contracted capacity. Projects inside a development pipeline can range from early land and planning positions to schemes with power, financing and construction under way, while some may never reach operation.
The scale nevertheless shows how far Iberia has moved beyond its earlier status as a secondary European market. Developers are now evaluating large projects around Madrid, Barcelona, Lisbon, Sines and other locations partly because established clusters such as London and Frankfurt face tighter land and power conditions.
Connectivity links several infrastructure markets
DE-CIX says the Iberian ecosystem now includes 35 submarine cables landing at 20 coastal points, 12 major terrestrial backbone networks, 13 Internet Exchanges, more than 100 data centres and more than 800 networks. The company reports typical latency of around seven to eight milliseconds between Madrid and the coastal hubs of Barcelona and Lisbon.
That network performance does not make the three cities interchangeable. Madrid remains the principal inland interconnection market, while Barcelona and Lisbon have different cable routes, power systems and development pipelines. The value of the distributed model is that workloads do not necessarily need every part of their compute, storage and network stack in one metropolitan area if the links between locations are sufficiently fast and resilient.
Different computing tasks can also place different demands on infrastructure. Large training workloads may prioritise access to substantial power blocks and land, while latency-sensitive inference or network services can place greater weight on proximity to users and interconnection. A regional cluster gives operators more options to place those functions where the physical conditions suit them.
Submarine connectivity strengthens that proposition because Iberia sits between routes serving Europe, Africa and the Americas. Cable landings alone do not create data centre demand, but they can reduce the network distance between international capacity and facilities built near the landing markets or connected inland through terrestrial fibre.
The pipeline still has to become connected capacity
The commissioned nature of the study remains relevant when interpreting its conclusions. DE-CIX operates interconnection platforms in Madrid, Lisbon and Barcelona, so it has a commercial interest in stronger regional connectivity and in presenting the three markets as complementary. The infrastructure counts are useful evidence, but claims that Iberia has become the world’s first distributed digital capital for AI are the sponsor’s market positioning rather than an independently established category.
Physical delivery will determine how much of the nearly 9GW pipeline becomes usable capacity. Each project still needs electricity, planning consent, finance, construction resources, cooling infrastructure and customer commitments. Those conditions vary substantially between Spanish and Portuguese regions and can change even where national renewable generation looks abundant.
Power availability is particularly easy to overstate. A market can have extensive wind and solar generation while a proposed data centre still faces a long connection timetable at its chosen substation. The relevant question for an individual scheme is whether the network can deliver the required firm capacity at the site and date specified, not the total renewable output available across the country.
Iberia also starts from a smaller installed base than Europe’s largest traditional clusters. Earlier DE-CIX research in 2024 counted around 1,000MW of installed data centre IT capacity across the peninsula. The current development pipeline is therefore several times larger than that earlier installed figure, making conversion from proposal to operation a major execution test rather than a simple continuation of existing supply.
The regional model could become more valuable if developers can use several locations to avoid concentrating every new megawatt around the same constrained city. It could equally leave a large paper pipeline if grid, permitting or commercial demand fails to keep pace with the project announcements.
Madrid, Barcelona and Lisbon now have enough interconnection, cable access and development activity to be assessed as parts of a broader Iberian infrastructure system. The next measure of that system will be less about the headline 9GW and more about how much power reaches commissioned halls, how quickly those halls fill and whether the network between the three hubs allows customers to use the region as one practical operating footprint.
