IREN — Earnings Review Q2 FY26 (Dec 2025)

Date: 2026-02-25 | GauchoRico


Headline

The $39.6M revenue "miss" is a red herring. IREN is burning down Bitcoin mining deliberately to free power capacity for AI cloud. The real question this quarter was never revenue — it was: is the capital structure solid enough to survive the transition valley? Answer: yes. Emphatically.

$9.2B in total committed funding. $2.8B cash. $3.6B GPU financing at under 6%. Microsoft prepayment covers 33% of GPU CapEx. Sweetwater 1 (1.4 GW) energizing April 2026. Microsoft revenue starts April 2026. IREN just transformed from "speculative infrastructure story" to "funded infrastructure story." That matters.


The Numbers That Matter

| | Q2_FY25 | Q3_FY25 | Q4_FY25 | Q1_FY26 | Q2_FY26 | | | Dec-2024 | Mar-2025 | Jun-2025 | Sep-2025 | Dec-2025 | |---|---|---|---|---|---| | Revenue ($m) | 116.1 | 144.8 | 187.3 | 240.3 | 184.7 | | YoY % | +173% | +166% | +229% | +355% | +59% | | BTC Mining ($m) | — | — | 180.3 | 233.0 | 167.4 | | AI Cloud ($m) | — | — | 7.0 | 7.3 | 17.3 | | AI Cloud % of rev | — | — | 3.7% | 3.0% | 9.4% | | Adj EBITDA ($m) | 62.4 | 82.9 | 121.9 | 91.7 | 75.3 | | Adj EBITDA margin | 53.7% | 57.3% | 65.1% | 38.2% | 40.8% | | AI Cloud GM% | — | — | — | — | 86.1% | | BTC Mining GM% | — | — | — | 65.7% | 62.1% | | Cash (m)|—|—|564.5|1, 032.3|3, 260.6||ARRcontracted(B) | — | — | — | — | 2.3 |


Six-Factor Take

Revenue growth: 59% YoY sounds like deceleration from 355% — but that's QoQ BTC noise. The forward trajectory is the only thing that matters here. Microsoft contract turns on in April. AI cloud was up 137% QoQ in Q2. The reported number is a lagging indicator of a business that is about to surge. Pass, conditionally.

Trajectory: Up. Massively. $184M reported revenue against $2.3B ARR already contracted. That divergence — 12.5x — closes starting Q3 FY26. Management reiterated $3.4B ARR by end CY26. If Sweetwater energizes on schedule and NVIDIA delivers GPUs, Q3 and Q4 FY26 will look nothing like this. Improving.

Margins: This is the sleeper signal of the quarter. AI cloud gross margin printed 86.1%. That is software-level margin on a capital-heavy asset. The 40.8% adj EBITDA margin is running well despite BTC wind-down drag. At scale — 3.4BARR, 852.9B of annual EBITDA from this business. The margin profile of the future business is exceptional. High.

Competitive dominance: 4.5 GW of secured power is a genuine moat. That took IREN 7 years to accumulate and costs billions to replicate. CoreWeave has GPUs but not power at this scale. Hyperscalers are building their own, but $9.7B contracts don't go to companies without a differentiated platform. Microsoft chose IREN — not CoreWeave, not DataBank — for their largest HPC outsourcing contract. That says something. Strong.

Valuation: $12.1B market cap. $13B EV. Against $2.3B contracted ARR — that's 5.6x. Against $3.4B target ARR — 3.8x. At 85% EBITDA margin on $3.4B ARR = $2.9B EBITDA. 20x EBITDA (fair for long-duration contracted revenue) = $58B EV, implying a 4.5x from here. Even at 15x (discount for execution risk) = $43B, implying 3.3x. That is exceptional upside if execution holds. The risk is not valuation — it's execution. Fair to cheap if thesis executes.

Special circumstances: $9.2B pre-funded infrastructure buildout with a hyperscaler anchor. Revenue recognition lag is not a thesis breach — it is the thesis. The company is deliberately destroying its own BTC mining revenue to build something much larger. The capital markets execution this quarter — $3.6B GPU financing, $2.3B converts, $1.9B prepayment — is what investors should be scoring. They delivered. Holy smokes, $9.2B of committed funding is not small.


The Bear Case

Single-customer concentration is real. Microsoft is ~84% of forward ARR. A contract amendment, delay, or hyperscaler captive build would cascade through the model. The $2.3B in convertible notes is dilution overhang — not existential given the asset value, but not nothing. GPU delivery timing from NVIDIA is not fully in IREN's control. And the CFO slip — "approximately $3 billion of CapEx for the Verizon data centers" — is either a brain malfunction mid-sentence or a tell on an undisclosed contract. Markets don't like ambiguity on that kind of thing.

The execution risk is front-loaded: Sweetwater 1 needs to energize on time, 140K GPUs need to deploy on schedule. One slip pushes the $3.4B ARR target by a quarter or more. At $13B EV you are paying now for future cashflows that haven't appeared yet.

I accept this risk at 11.5% allocation because the downside has a floor: $2.8B cash + $9.7B contracted revenue = the balance sheet is not going to zero. But I am not adding at current prices until I see at least one quarter of material Microsoft revenue recognition — Q3 FY26 (May 2026 report) is the test.


Prior Thesis vs This Quarter

Going in, I said: "Thesis intact but unproven. Moment of truth is Q3 FY26." This quarter did not prove the AI cloud business — that comes in April. But it de-risked the balance sheet substantially. The financing execution exceeded the target (3.6Bvs2.5B guided). That is a meaningful beat on the only metric that mattered this quarter.

Thesis status: Intact. No change in position. Watching for April 2026.


Promise Tracker Assessment

Management has delivered every closed promise so far. The open ones are due in the next two quarters.


Action

Hold 11.5%. Do not add before Q3 FY26 results (May 2026). Would add aggressively if Q3 FY26 shows $300M+ revenue with Microsoft ramp confirming.

We will see what happens...

GR