A growing sense of unease is settling over the companies that were supposed to power the AI revolution. This week brought a flurry of warning signs: the chipmaker Solidigm pulled its IPO, smaller “neocloud” providers specialising in AI compute are coming under severe financial pressure, and the cost of insuring against defaults from some of these firms has jumped sharply. It is not a crash—not yet—but it is the kind of market stress that usually precedes a painful recalibration.
The neocloud model was built on a simple premise: with hyperscalers like Amazon and Microsoft gobbling up most GPU supply, there was room for leaner, more flexible cloud providers to serve AI startups hungry for compute. It worked for a time, but rising interest rates and an increasingly wary debt market are choking the growth that depended on cheap capital. The rising spreads on credit default swaps suggest bond investors are reassessing the risks these firms carry. When a company like Solidigm, a memory and storage maker with close ties to the AI supply chain, chooses to withdraw its public listing rather than price below expectations, it signals that the market is no longer willing to pay a premium for AI-linked stories.
What makes this development significant is not just the immediate pain, but what it reveals about the structure of the AI boom. A disproportionate share of the hundreds of billions poured into AI infrastructure has gone into facilities and hardware that may not generate the returns investors have been assuming. Some of the most ambitious buildout plans were underwritten by the promise of exponential demand growth. When that growth slows—even modestly—the debt taken on to build data centres and buy chips becomes a liability rather than an asset. The challenge is not whether AI itself is real; it is whether the financial architecture around it was built with enough prudence.
For the broader market, these tremors arrive alongside other sources of turbulence. Oil prices remain elevated, geopolitical tensions are unresolved, and the Federal Reserve is still navigating inflation risks. In such an environment, capital naturally flows toward what is proven rather than what is promised. That is not pessimism—it is the market performing its basic function of pricing risk. The question is whether the correction will be orderly or sharp. Either way, the companies most exposed to the AI narrative will need to show they can generate cash, not just buzz. The days of being rewarded for ambition alone are drawing to a close.