RAPIDRATINGS
Even suppliers already considered financially healthy saw their position weaken under the scenario, with cash reserves declining and leverage rising. The net effect was a marked increase in the proportion of suppliers classified as high risk.
“ What we find, ultimately, is an increase in high-risk companies as a result,” Charlie adds.“ Even with explosive growth, we saw an increase from about 18.5 % to about 26 % of companies becoming high risk or very high risk – roughly a 37 % increase.
“ So, rapid expansion doesn’ t necessarily mean rapid improvement. It means there needs to be a way to fund that growth, because it doesn’ t happen on day one, and that puts companies into a much more vulnerable position than they’ re in today.”
The blind spot beneath chips and power Much of the public conversation about AI infrastructure has focused on chip availability and, more recently, power capacity.
Charlie argues that both concerns, while real, obscure a deeper structural risk sitting further down the supply chain:“ There’ s a very deep bench of companies contributing to those broad themes. When organisations are thinking about data centre build-outs, they don’ t often think about how the tier 2, 3, 4 and 5 suppliers are going to be able to deliver – not just on their own project, but on all the projects that are out there.”
That capacity constraint, he says, is not currently being reflected in how these
“ We saw an increase from about 18.5 % to about 26 % of companies becoming high risk or very high risk”
Charlie Minutella Chief Revenue Officer RapidRatings
projects are financed:“ You get to certain choke points where critical companies may not be able to do all of this at the same time, and that’ s typically not being priced into the infrastructure financing agreements. We think that’ s a huge blind spot for the wider AI infrastructure build-out.”
Underlining just how real that uncertainty is, Charlie points to a figure that illustrates the scale of exposure in the US market:“ We know there was US $ 130bn worth of data centre cancellations so far this year in the US.”
For the tier 2 and tier 3 suppliers who had scaled up to meet that demand, cancellations of that magnitude are a direct threat to revenue that had already been assumed and, in many cases, financed against.
Why hyperscalers are exposed Much of this exposure traces back to how hyperscalers are now building AI infrastructure.
6 October 2026