Iren's Earnings Weren't Groundbreaking, but 2027 Looks Very Promising

Iren posted FY26 Q4 revenue of $137.2M (-26.7% YoY) and a $684M net loss, but the key driver is delivery timing: Horizon 1 (50MW out of 200MW) was delivered Aug. 13, unlocking $485M in annual recurring revenue and expected meaningful sequential growth as it starts showing up in upcoming quarters. The company’s AI cloud deal economics are accelerating with new customers paying ~$20M per MW per year (more than double Microsoft’s implied ~$9.7M per MW), while management projects $25B–$30B capex for FY27 and says it can fund it with $14B cash plus ~$8B of GPU financing/prepayments—reducing dilution concerns. Overall, investor disappointment looks largely timing-related rather than thesis-breaking, with momentum expected into 2027 as Horizons 2–4 are delivered.

Analysis

The market is treating this like a dilution-and-loss story, but the more important mechanism is pricing power per megawatt. If contracted compute is really clearing at $20-25M/MW annually, the asset is behaving less like a software rollout and more like a scarce utility franchise with customer-funded capex, which shifts value capture away from vendors and toward operators with available power and balance-sheet capacity. That is a better setup for IREN than for earlier-stage AI cloud names still selling future capacity at lower implied rates.

Near term, the stock will trade on execution cadence, not on current-period earnings. The next 1-3 months matter because Horizon deliveries and financing headlines will either validate the bridge to 2027 or expose slippage; if sequential ARR does not inflect sharply next quarter, the market will re-open the capex/dilution debate. The real tail risk is funding quality: if prepayments slow or GPU financing comes with punitive terms, the equity starts to look like a levered infrastructure build instead of a scarce-asset compounder.

The contrarian miss is that this is not primarily a 2026 revenue problem; it is a timing problem inside a structurally better pricing environment. Consensus may be underweighting how much of the upside is deferred into 2027-28, but the move can still be too early because the equity is being asked to bridge a large capex cycle before the cash-flow inflection is visible. NVDA is a second-order beneficiary from GPU demand, but the bigger rerating potential sits with IREN if management keeps converting power into contracted revenue at current economics.

Key Decisions for Investors

• Initiate a starter long IREN only on continued post-earnings weakness; think 3-6 month horizon, with upside tied to the next two quarterly ARR updates. Use a 1/2 position until sequential cloud revenue proves the Horizon 1 ramp is real.
• Prefer 12-18 month call spreads on IREN over outright short-dated calls; the thesis is a 2027 cash-flow bridge, so optionality should match the catalyst window. Falsify if management delays Horizons 2-4 or next quarter fails to show meaningful cloud segment acceleration.
• Relative-value idea: long IREN / short NBIS on any spread widening if the market keeps rewarding near-term revenue acceleration over implied revenue per MW. The pair should work if investors rotate from headline growth to unit economics over the next 1-3 quarters.
• Set a hard alert on financing quality: if prepayments fall below roughly 40% of GPU capex or debt pricing tightens materially, cut exposure. That would signal the story is reverting from customer-funded growth to balance-sheet risk.
• Do not chase MSFT or NVDA off this print; they are second-order beneficiaries only. Use them as hedge context, not as the primary expression.