IREN's Transition Math: A Missed Quarter, a $9.7 Billion Contract, and a Market That Can't Decide

IREN's Q4 revenue missed estimates, but AI cloud revenue surged and Microsoft's $9.7B deal supports a $4B 2026 target, despite a $702.6M net loss.

The arithmetic of transformation rarely looks clean on a quarterly income statement, and IREN's fiscal fourth-quarter report was no exception. The company posted revenue of $137.2 million against the $157.14 million Wall Street had penciled in, sending shares down 8.23 percent in after-hours trading to $37.19. In the regular session, the stock had actually closed 2.4 percent higher at $40.53 — a split-screen reaction that captures just how divided investors remain on this story.

But fixating on the revenue miss means ignoring the more consequential numbers sitting elsewhere in the report. The company's annualized AI cloud revenue target for 2026 now stands at over $4 billion, with roughly 85 percent of that already contracted. Management is guiding toward approximately 150,000 GPUs deployed by the end of next year. Those figures, not the quarterly top line, are what actually measure whether the pivot from Bitcoin miner to AI infrastructure provider is working.

The Contract That Changes the Math

The guidance isn't aspirational — it's backed by signatures. Microsoft formally accepted the first of four data centers at IREN's Childress, Texas campus last week, a milestone that triggers the first billable invoices under a five-year agreement worth $9.7 billion. Days earlier, Nvidia had granted the same facility Exemplar Cloud status for its GB300 systems after testing. Two of the biggest names in AI infrastructure have now independently validated that IREN can deliver technically.

For a company mid-transition, the revenue miss is almost structurally inevitable. Shifting from Bitcoin mining to AI infrastructure means moving revenue recognition, capacity utilization, and contract ramp timelines — the Microsoft invoices only start flowing now, after the Horizon 1 acceptance. The more meaningful metric is the contracted ARR volume and the raised guidance, which suggest the operating story is ahead of where the market's expectations had drifted.

The stock's recent turbulence predates the earnings report. Shares came under pressure in mid-August as the market digested the Microsoft acceptance, broader weakness across neocloud names, and rising long-term bond yields — evidence that a meaningful portion of the volatility is sector-wide rather than company-specific.

Two Ledgers, One Company

The full-year numbers tell a story of deliberate destruction. IREN recorded a net loss of $702.6 million for the fiscal year, swinging from a $86.9 million profit the prior year, driven largely by $638.8 million in impairment charges — the accounting cost of writing down legacy mining assets. Yet the per-share loss of $0.41 came in 18 percent better than the $0.50 loss Zacks had projected, and revenue actually beat the Zacks consensus by 13.54 percent, underscoring how widely analyst expectations diverged.

The growth engine is unmistakable. AI cloud services revenue jumped from $16.4 million to $128.8 million year over year, while the legacy Bitcoin mining business — which IREN wants to retire — still grew from $484.6 million to $578.2 million. That coexistence of old and new revenue streams is precisely what makes the current phase so hard to value.

Management has set a definitive end date for the split personality: mining operations are slated to be effectively wound down by the end of December 2026. Until then, the investment continues — administrative and selling costs are expected to rise by $40 to $50 million sequentially in the first quarter as the company pours resources into sales, marketing, research, site operations, and cloud functions.

The Skeptic's Case

Not everyone is convinced the timing works. JPMorgan's Richard Choe reaffirmed his Sell rating with a $46 price target on the day of the release, pointing to significant risks in the AI infrastructure buildout — particularly uncontracted GPU capacity and elevated debt levels. It's a warning that carries weight precisely because it comes from a house that evaluated the numbers on the day they were published.

The central question is whether a company can wind down one business and scale another quickly enough to reach profitability before capital runs short. That answer won't come from any single quarter but from the durability of the new revenue base.

The market's ambivalence shows in the technicals. The stock closed Thursday at €34.85 in European trading, roughly 49 percent below its 52-week high of €68.61 set in early November, yet still 83 percent above its 52-week low. Thirty-day annualized volatility sits at 133 percent. Shares are trading about 4.2 percent below their 50-day average and nearly 12 percent below the 200-day average — signs that investors haven't fully embraced the narrative.

Bernstein's Buy recommendation from mid-August predates the Microsoft acceptance and the raised guidance, making it more of a historical data point than a current signal. The risks — extreme volatility, sector headwinds, a young business model — remain real. But for those who believe in the transition from miner to AI infrastructure provider, the current weakness looks less like a warning and more like the cost of watching a company rebuild itself in real time.