For most of the last century, Portugal saw itself as the western edge of Europe — the place where the map runs out. The age of artificial intelligence has quietly inverted that geography. What looked like the periphery has become, almost without anyone noticing, one of the continent's most strategically located thresholds: the point where Europe meets the Atlantic, the cables, and the energy that the new economy runs on.
This is the companion argument to an earlier piece, "O Novo Ouro do Atlântico." Where that essay traced the resource — clean, firm energy and the submarine cables already coming ashore — this one traces the architecture: what Portugal would have to build, and where, to turn that position into capability rather than rent.
The energy case, and its one real weakness
Artificial intelligence does not run on ambition; it runs on electricity, delivered without interruption. On the mainland, the scale is already real: the Start Campus data centre at Sines is building toward 1.2 GW of capacity, an €8.5bn investment cooled by seawater, sitting at the Atlantic landing point of the great transatlantic cables — EllaLink to Brazil, Olisipo, Medusa, and Nuvem to the United States. Portugal's grid is also among the cleanest in Europe, heavy with hydro, wind and solar.
But honesty requires naming the weakness, because it is the counterargument every serious reader will raise. The Iberian Peninsula remains an electrical near-island: interconnection with the rest of Europe stands at roughly 3%, against the EU's target of 15% by 2030. And on 28 April 2025, the peninsula suffered a major blackout. It would be easy — and wrong — to read that as proof that renewables or isolation make Portugal unreliable. The ENTSO-E final report of March 2026 attributed the event to a failure of voltage control, not to renewable generation and not to the interconnection deficit as such. The lesson is not that the model is broken; it is that the grid needs voltage-management investment and more interconnection — a fixable engineering problem, not a structural curse.
It was a failure of voltage control, not of renewables. The fix is engineering, not retreat.
The Azores: measured in rent per unit of heat
The temptation, having established the energy case, is to imagine the Azores as a second Sines. They are not, and pretending otherwise weakens the argument. São Miguel draws 40–50% of its electricity from high-enthalpy geothermal energy, around the clock — a genuine advantage — but the island's entire peak demand is roughly 79 MW, against Sines' 1.2 GW. Total Azorean geothermal capacity sits near 33 MW, heading toward 40 MW through a €75m nine-well programme. These are not hyperscale numbers, and they were never meant to be.
The Azorean vocation is different in kind: specialised, sovereign, secure computing, where the value lies not in raw scale but in firm clean energy per unit, in latency on the transatlantic routes, and in strategic location. Iceland — volcanic and Atlantic, as we are — reached that niche first, which is precisely why the comparison should sharpen Portuguese ambition rather than flatter it.
Terceira: the anchor — a proposal, not a fact
Here I must be explicit, because the distinction matters and the earlier draft of this argument blurred it. What follows is my own proposal, not an existing facility. I have called it the ADMH-T — a hybrid, dual-use, civil-and-defence cluster anchored on Terceira, designed to convert the island's Atlantic position into productive and strategic capability rather than rent collected from a foreign base.
The shape of it: an envelope of around €980m over seven years, mobilising an estimated €700–950m in private investment, generating between 2,000 and 4,400 direct jobs and up to roughly 10,000 in total once induced employment is counted, and projecting €450–700m in cumulative dual-use and technology exports. The reference markets are not speculative — EU defence procurement runs at roughly €250bn a year, and the European Defence Fund committed around €8bn for 2021–2027. Terceira's triple tectonic-plate junction makes it a natural site for strategic redundancy alongside São Miguel's primary node.
The cluster would not begin from nothing. Portugal already has the firms: Tekever, the maritime-surveillance drone unicorn backed by the NATO Innovation Fund, valued above €1bn and pointedly ITAR-free; Thales Edisoft in secure communications, naval and space systems, already operating a teleport on Santa Maria; CEiiA in aerospace, autonomous systems and ocean technology, building Portugal's first licensed satellite; and a space cluster around GMV, Deimos and Neuraspace. The pieces exist; what is missing is the architecture that joins them.
The open field at Santa Maria
One further possibility deserves mention without overstatement. Santa Maria's spaceport was licensed in August 2025, and in May 2026 SpaceX confirmed it was actively searching for an international launch site, with no location yet chosen. It would be dishonest to present this as a Portuguese opportunity already won — Santa Maria's latitude and small-satellite niche differ from Starship's requirements, and the search may well end elsewhere. But the field is open, and a country that has prepared the ground negotiates from a different position than one that has not.
The choice beneath the choice
There is a gap that no foreign investor will close for Portugal: the protection of the very cables and infrastructure that make the country valuable. That is sovereign work, and it is also the connective tissue of the cluster — the reason the civil and the defence dimensions belong together rather than apart.
The deeper point is the one that closes the earlier essay too. Holding the resource has never been the same as capturing the value. Portugal can let others land the cables, install the servers, fly the drones built elsewhere, and carry off the added value — or it can build the capability to keep it. Value does not live in the position. It lives in the capacity to transform it.
The frontier is no longer where the map runs out. It is where the capability begins.