A sharp market selloff is testing one of the most aggressively financed corners of the global technology economy: the race to build artificial intelligence data centres. Companies that have ridden the AI boom to secure large pools of capital are now confronting a more demanding funding environment, as investors reassess whether the economics of power-hungry computing infrastructure can hold up in a world of higher rates and tighter risk appetite.
Funding Gets Harder
The immediate pressure is showing up in capital markets. Firms planning to raise money for data centres, chips, cloud capacity, and related infrastructure are finding that investors are less willing to fund growth on the promise of future AI demand alone. Some companies are delaying fundraising plans altogether, while others are being asked to pay more for debt or offer more attractive terms to equity investors. The shift reflects a broader repricing of risk across markets, where the cost of capital has become a central test for businesses that depend on heavy upfront spending and long payback periods.
For the AI infrastructure trade, the timing is especially awkward. The sector has been built on expectations that demand for compute will remain strong enough to justify enormous investments in land, power, cooling systems, networking gear, and specialised chips. But those projects are capital intensive, and many are financed with a mix of debt and equity that becomes more expensive as interest rates rise. When markets turn volatile, lenders and investors tend to scrutinise cash flow, utilisation rates, and customer commitments far more closely than they do during a boom.
Power And Pricing Pressure
The challenge is not only financial. AI data centres consume vast amounts of electricity, and power costs have become a major line item in project economics. Rising rates amplify the burden by increasing financing costs just as operators face higher expenses for energy, grid access, and advanced cooling. That combination can compress margins and force developers to prove that their facilities will be used at high capacity and generate durable returns.
This is a significant shift from the earlier phase of the AI rally, when enthusiasm for the technology often outweighed concerns about near-term profitability. Investors were willing to back ambitious expansion plans on the assumption that AI adoption would accelerate quickly and that the winners would secure outsized market share. Now, with public markets more cautious and borrowing costs elevated, the burden of proof has moved back to fundamentals. Companies must show not only that AI demand is real, but that it can be monetised at a pace sufficient to service debt and justify new construction.
The selloff also exposes a structural tension in the AI buildout. The infrastructure race requires enormous scale, but scale itself can become a liability if financing conditions deteriorate. Projects that looked manageable when capital was cheap can become far more difficult to support when rates rise, especially if customers are still early in their own AI spending cycles. That makes fundraising more sensitive to market sentiment and increases the risk that some planned expansions will be delayed, resized, or restructured.
Investors Demand Proof
For investors, the new caution is less about abandoning AI than about insisting on discipline. The market is signalling that growth stories must now be backed by evidence of demand, pricing power, and efficient capital deployment. In practical terms, that means companies seeking funds for AI data centres may need longer customer contracts, stronger balance sheets, or clearer paths to profitability before capital is committed on favourable terms.
The broader macro backdrop matters as well. Higher interest rates have already changed the calculus for sectors that rely on debt-heavy expansion, from real estate to utilities and infrastructure. AI data centres sit at the intersection of all three: they are physical assets, energy consumers, and technology platforms. That makes them especially vulnerable when markets turn risk-averse. A selloff can quickly translate into a higher hurdle rate for new projects, even if the long-term strategic case for AI remains intact.
For now, the message from markets is clear: the AI infrastructure boom is not over, but it is entering a tougher phase. The companies best positioned to navigate it will be those that can demonstrate real demand, secure power at manageable cost, and raise capital without assuming that investor enthusiasm will remain unlimited. In a more expensive money environment, the winners may be the firms that can build less on hype and more on hard economics.