Capital Came Back in 2025 — but Only to a Handful of Postcodes

Alternative headlines:  The FDI Rebound Is Real, and Ruthlessly Concentrated   ·   Data Centres Up $235bn, Infrastructure Down $55bn: The Map of Money

Global foreign direct investment rose 6% to $1.6 trillion. The headline is a recovery. The detail is a reallocation — towards artificial intelligence and a shrinking club of destinations, away from the physical economy that shipping serves.

THE SET-UP

After two consecutive years of decline, global foreign direct investment returned to growth in 2025, rising 6% to $1.6 trillion. On the face of it, that is a clean recovery story: confidence back, capital moving, the worst behind us. UNCTAD's latest World Investment Report duly carries the cheerful headline. It also, if you read past it, carries the warning that the headline tells only part of the story — and the part it omits is the one that matters for anyone trying to read where the world's growth, and its cargo, will come from.

Because this is not a broad-based revival. It is a narrow one, and its narrowness runs along two axes at once — by geography and by sector. Capital came back, but it came back to fewer places and fewer kinds of project than before. The recovery in the aggregate masks a concentration in the composition. The question I want to put is not whether investment is rising. It is what it is being spent on, and what its choices reveal about the shape of the next cycle.

Chart 1 — FDI inflows by region, 2024 vs 2025 (US$ bn). NauticX original visualisation of reported figures.

THE READ

Take the geography first. The top twenty economies now hoover up more than 80% of all foreign direct investment, and the regional table shows how lopsided the rebound is. Europe surged 39% and developed economies as a whole rose 11%, whilst North America slipped 2% and — the figure that should trouble anyone thinking about future trade lanes — Africa fell 26%. Asia, still the largest single recipient, managed a modest 3%. Capital is not spreading out in search of the next frontier; it is huddling in the places it already trusts. For a global trading system that depends on new production capacity emerging in new regions, a rebound that bypasses Africa and barely lifts Asia is not the good news the top line implies.

“Capital is not spreading out in search of the next frontier; it is huddling in the places it already trusts.”

The sectoral swing is starker still, and it is where the real signal sits. Announced greenfield and project-finance value into data centres rose by a staggering $235 billion, dwarfing every other category. Oil and gas added $38 billion and semiconductors $13 billion. Now read the other side of the ledger: infrastructure fell $55 billion, so did the value into GVC-intensive industries, renewable energy dropped $37 billion, and real estate $31 billion. In one chart you can watch capital walk out of the physical, trade-generating economy — the factories, the logistics infrastructure, the manufacturing supply chains — and into the digital estate that underpins artificial intelligence. Money is voting, decisively, for silicon over steel.

Chart 2 — Change in announced greenfield & project value by sector (US$ bn). NauticX original visualisation.

This is the observation that should concern a shipping and logistics readership most, because it is a statement about the future composition of demand. A container ship earns its keep on GVC-intensive industry, on manufacturing supply chains, on the physical infrastructure that moves and makes goods — precisely the categories now shedding investment. A dollar committed to a data centre generates a very different, and much thinner, freight footprint than a dollar committed to a factory. If this composition holds, the world is building an economy whose growth increasingly bypasses the sea.

THE COUNTER-VIEW

The most serious rebuttal is that I am reading a snapshot as a trend, and reading announcements as reality. Greenfield and project-finance figures record intentions, not deployed capital, and FDI flows are famously lumpy and lagging. More substantively, the AI build-out is not as bloodless for the physical economy as the headline suggests: hyperscale data centres are enormous consumers of power equipment, transformers, construction materials, cabling and cooling systems, much of which is manufactured in Asia and shipped. The digital boom may yet reach back and pull hard on the very trade lanes it appears to be abandoning. That second-order demand is real, and I would not dismiss it. But it is downstream, uncertain in timing, and thinner per dollar than the primary industrial investment it is displacing. The near-term signal — capital choosing digital over physical, concentration over breadth — stands.

THE TAKEAWAY

Read the FDI headline as a recovery in level and a deterioration in mix. The frame to carry is that where capital goes today shapes what moves tomorrow, and today it is going to a narrow band of trusted geographies and to digital infrastructure over industrial capacity. For a trade-exposed reader that is a caution, not a catastrophe — but it argues for watching the composition of investment at least as closely as the total. The indicators to track are the greenfield sector mix in successive UNCTAD updates, the geographic concentration ratio, and, crucially, the second-order physical demand generated by the data-centre boom. If that last one fails to materialise at scale, the divergence between where money flows and where goods flow will only widen.