IREN’s AI Pivot Faces a Balance-Sheet Reckoning as Debt Climbs to $7.84 Billion

The market has already delivered its verdict on the sector. What it hasn't answered is whether IREN can fund its own transformation.

Shares of IREN fell 11% on the day, caught in a broader selloff of data-center and "neocloud" names. Persistent worries about the reliability of the company's older facilities and about industry-wide financing costs added to the pressure. But lumping all of that into a single doom narrative would be a mistake — the risks sit in different places, and they deserve to be judged separately.

The backdrop is a Gartner forecast that more than 80% of neocloud providers could lose their independence by 2030. IREN was not named in that report. Losing independence is not the same as going out of business, and reading a specific fate for IREN into a sector-wide projection goes well beyond what the forecast actually says. Dismissing the warning as mere market noise would be just as hasty.

A $25–30 Billion Bet on the AI Cloud

At the center of the story is IREN's planned shift from Bitcoin mining to AI cloud services. For fiscal 2027, the company projects capital spending of roughly $25–30 billion. That figure puts two questions on the table at once: whether demand for computing power will materialize, and — just as important — when contractual commitments actually turn into cash.

The scale of the ambition is visible in the announced early-stage plans for the Redbank campus: 900 MW near Wagga Wagga, a project that could involve more than AU$10 billion in investment, with IREN working toward a grid connection from 2029. Management has described the site as being in an early review phase. It is an expansion prospect, not proof of secured execution or future revenue.

The Funding Gap Is the Real Test

What matters most is the distance between customer payments and capital outlays. According to media reports, IREN's customers pay 45–55% of GPU costs upfront, and on some projects lenders finance up to around 90% of GPU spending. Those terms could limit the cash IREN itself has to put up.

The qualifier is essential: such financing ratios apply project by project and cannot simply be extrapolated across the entire buildout. Nor does a large upfront payment replace the proof that delivering the service later actually works economically. Investors should watch not only contract volume but its conversion into cash-generating revenue.

On that front, the reported numbers offer some support. IREN is said to hold $4 billion in contracted annual revenue, and management estimates GPU payback at about two years. Both underpin the financing thesis — provided capacity runs reliably and the expected customer payments arrive. The payback figure remains a management estimate: it describes expected economics, not cash already recovered.

Investor Infinity Curve has taken a positive view of IREN relative to CoreWeave, citing the upfront payments, project financing and the expected economics of the GPU contracts. That is a reasoned assessment, but it is not evidence that every expansion project can be financed on comparable terms.

Revenue forecasts sketch the size of the expected ramp: $2.72 billion for fiscal 2027, rising to $7.28 billion the following year. For a genuine re-rating, that growth would need to come with improving financial resilience. Higher revenue alone would not settle the central investor question. More convincing would be evidence that additional capacity gradually reduces cash burn rather than requiring ever more financing.

Debt, EBITDA and a Negative Cash Flow

The bear case starts where the buildout ties up more capital than expected, or where revenue arrives late. In the fiscal year through June 30, 2026, IREN's debt jumped from $964.23 million to $7.84 billion. Financing is already a core part of the investment case.

Operational risk adds to it. SemiAnalysis reported multi-day power outages, grid disruptions and storage problems at the Prince George and Mackenzie data centers — reports that bear directly on how reliably contracted capacity can be delivered.

The earnings and liquidity trend also warrants attention. EBITDA fell to $38.35 million in that fiscal year from $201.71 million, and free cash flow after financing costs came in at -$2.23 billion. Should technical difficulties persist or lenders demand stricter terms, IREN's financial room to maneuver could narrow — a serious risk even with AI demand still running high. Demand and fundable execution are not the same thing.

What the Next Report Has to Show

As long as upfront payments and project loans reliably carry the buildout, an operational recovery remains a plausible scenario. If cheap financing dries up or service delivery slips, the heavy investment burden would weigh more heavily — and a lower share price would not, by itself, justify a cheaper fundamental valuation.

The next concrete checkpoint is the upcoming earnings release, provisionally scheduled for December 1. The date is not yet final. Investors should focus there on liquidity, new financing terms and progress in bringing computing capacity online.

The decisive item will not be the next big growth number in isolation. IREN needs to show, in a traceable way, how its contract-backed business turns into dependable cash flow. Only that proof could put the financing concerns to rest for good.