AI investment is bringing extraordinary profit forecasts, but some of the financing and accounting underneath the boom could be less sturdy than the headline numbers suggest. MoneyWeek’s analysis points to a growing split between highly leveraged data-centre businesses and cash-rich technology giants.
Watch Desk analysis
What happened
MoneyWeek says the cost of insuring debt issued by CoreWeave has risen to levels implying a greater-than-60 per cent chance of default within five years, according to Bloomberg. The article also says CoreWeave shares have fallen 67 per cent from their June 2025 peak. That is a warning about one company’s credit risk, not proof that the wider AI business is about to buckle.
The article contrasts that pressure with larger technology companies: Microsoft shares have risen 38 per cent since the start of July. It cites Columbia University’s Stijn Van Nieuwerburgh estimating that AI data centres, power systems, networking and specialised chips could attract $10.3 trillion in spending from 2025 to 2032.
Why it matters
The question is not just whether AI companies can generate revenue. It is whether the returns arrive before today’s infrastructure and hardware investments have to be written down. MoneyWeek notes that chip purchases can appear as immediate revenue for a supplier, while buyers depreciate the chips over several years. If the promised earnings do not materialise, that accounting timetable could leave a sizeable bill waiting down the road.
The article also points to forecasts of strong profits among AI-exposed firms, while warning that borrowing and more opaque financing arrangements make it harder to see where risk is building. That makes the boom a question of business models and balance sheets, not just share prices.
Our read
“AI bubble” is a useful phrase for a headline and a poor substitute for distinguishing companies, cash flows and financing. CoreWeave’s reported debt concerns deserve attention, but they do not establish that the hyperscalers face the same risk. The more revealing test is whether AI investments produce enough lasting earnings to justify the hardware and infrastructure spending, rather than merely moving impressive numbers between companies.
What to watch
- Whether data-centre operators can service their debts as capacity expands.
- Whether AI-related earnings growth persists after hardware depreciation is accounted for.
- How much of the investment boom is financed through less transparent arrangements.
Discussion spark: If AI revenue keeps growing but profits depend on costly hardware lasting for years, what evidence would convince you that the investment boom is durable rather than a bubble?
Sources and evidence
- Source update (30 October 2026, 10:30 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.