When the monthly trade figures arrive, commentary tends to focus on the headline: up or down, record or slump. For Data Centres, that framing misses the more important story. Over the past three years the composition of buyers, the structure of contracts and the geography of production have all changed, and the data only now reveals it.

Start with buyers. Large importers in France and Japan have moved from annual tenders to multi-year framework agreements with two or three qualified suppliers each. That is a gift to suppliers that make the shortlist, and a closed door to those that do not. Our analysis of 453 contracts suggests the shortlist is hardening.

The signal behind the headline

A word of caution on forecasts. Most projections assume a smooth trajectory of 5.6% annual growth. History says otherwise: trade in Data Centres has moved in step changes, driven by a tariff change here and a supply shock there. Planning for resilience will pay off more than planning for the average.

For companies, three practical steps follow. Invest in the compliance stack before you need it. Build a second market so that a policy change does not become a crisis. And renegotiate contracts to share more of the risk, even at the cost of some headline price.

What does this mean for policy? The temptation is to chase headline incentives. A better use of public money, in my view, is the unglamorous plumbing: faster refunds of duties, shared testing laboratories, and common data standards. These reduce the fixed costs of competing for smaller firms and expand the supplier base.

“Plan for step changes, not averages. History in this sector moves in jumps.”

Contracts, geography and compliance

Third, geography. The production map has become more distributed. Firms that once served the world from one site now maintain regional assembly or finishing points, often within 45 days of the customer. This pattern reduces freight risk and shortens response times, but raises the capital required to compete.

It also changes how we should read the incentives that already exist. Production-linked schemes have drawn capital to the sector, but their real test will be whether they produce firms that can win contracts without support. Early evidence from $462 million of disbursements is encouraging but incomplete.

Finance and policy implications

Fourth, compliance as a moat. Product-level traceability, carbon accounting and conformity paperwork are no longer back-office tasks. In Data Centres, they are becoming the filter through which buyers decide who even gets to quote. For many small suppliers, the cost of compliance is higher than the cost of capacity.

Second, contracts are changing shape. Index-linked pricing now appears in 16.5% of new agreements, up from about a third three years ago. Suppliers accept lower headline prices in exchange for volume floors, and the effect is to shift risk from the order book to the cost base. Firms without hedging capability are exposed.

Practical steps for firms

I do not want to be pessimistic. The fundamentals are sound, the talent is there and the buyers are interested. But the next three years will reward the prepared and punish the complacent, and the data already shows which group is pulling ahead.