IREN — Earnings Review Q2 FY26 (Bear)

Date: 2026-02-22 Quarter: Q2 FY26 (Dec-25) Market cap: ~$12.1B | P/S (TTM): ~16x | Revenue growth: +59% YoY, -23% QoQ


Prior Beliefs / Updated Beliefs

Going into Q2 FY26, I had these expectations:

Metric Prior Belief Updated Belief Verdict
Revenue (QoQ) Continued growth ~$250M; BTC mining elevated $184.7M — down 23% QoQ Miss on headline; explained
AI cloud revenue Modest ramp from $7.3M — maybe $12-15M $17.3M (+137% QoQ) Beat — faster than expected
BTC mining Stable to slight decline $167.4M (-28% QoQ) Larger drop; intentional + price/hashrate
Adj EBITDA margin Stable ~55-60% 40.8% Compressed — mix shift eating margins
Capital position Progress on GPU financing $3.6B at <6% secured Positive surprise on rate
Contracted ARR Microsoft ($1.94B) + some additions $2.3B total (Prince George $400M added) Beat — new customer visibility
GAAP noise Meaningful non-cash charges $219M unrealized derivative loss + $112M debt conversion inducement Expected magnitude

Delta summary: Revenue soft but the signal underneath is better than the number. AI cloud accelerating faster than I expected. The GPU financing at sub-6% is a genuine positive — I'd have modeled 7-8%. What I underestimated: margin compression from the revenue mix shift would be this pronounced. Adj EBITDA margin fell from 65% in Q4 FY25 to 41% now. That's the price of the transition.


The Numbers

| | Q124 | Q224 | Q324 | Q424 | Q125 | Q225 | Q325 | Q425 | Q126 | Q226 | | | Sep-23 | Dec-23 | Mar-24 | Jun-24 | Sep-24 | Dec-24 | Mar-25 | Jun-25 | Sep-25 | Dec-25 | |---|---|---|---|---|---|---|---|---|---|---| | Revenue ($m) [GAAP] | 34.4 | 42.6 | 54.4 | 57.0 | 52.8 | 116.1 | 144.8 | 187.3 | 240.3 | 184.7 | | QoQ % | — | +24% | +28% | +5% | -7% | +120% | +25% | +29% | +28% | **-23%** | | YoY % | — | — | — | — | +54% | +173% | +166% | +229% | +355% | **+59%** | | Gross Margin % [GAAP] | — | — | — | — | 39.6% | 72.2% | — | 71.8% | 66.4% | 64.4% | | Adj EBITDA ($m) [Non-GAAP] | — | — | — | — | 2.5 | 62.4 | 82.9 | 121.9 | 91.7 | 75.3 | | Adj EBITDA Margin [Non-GAAP] | — | — | — | — | 5% | 54% | 57% | 65% | 38% | 41% | | GAAP Net Income ($m) | -5.3 | -5.2 | 8.6 | -27.0 | -51.7 | -21.9 | -16.3 | +176.9 | +384.6 | **-155.4** | | D&A ($m) | — | — | — | — | — | 36.1 | 47.3 | 63.8 | 85.2 | 99.2 | | FCF (m)|—|—|—|—|—|—|—|—|−138|* * −648 * *||AICloudRev(m) | — | — | — | — | — | — | — | 7.0 | 7.3 | 17.3 | | BTC Mining Rev ($m) | — | — | — | — | — | — | — | 180.3 | 232.9 | 167.4 |

Context for GAAP losses: Q126 and Q224 show massive GAAP swings driven by unrealized derivative gains/losses on convertible notes. Q1 FY26 had 384.6Mnetincomemostlyunrealizedgain.Q2FY26has155.4M — mostly $219M unrealized derivative loss + $112M debt conversion inducement. GAAP is nearly useless here. Adj EBITDA of $75.3M is the operational signal.


The Numbers Have to Match the Theory

Let me be clear about what theory is being underwritten here.

The theory: IREN deploys 140,000 GPUs by end CY2026, Sweetwater energizes Q2 CY2026, Microsoft contract revenues commence and ramp, taking ARR from $2.3B contracted to $3.4B by year-end. At 3.4BARRand 12B market cap, that's roughly 3.5x P/S on forward revenue — cheap if they execute.

What the numbers need to show by Q3 FY26 (March 2026 quarter):

What would break the theory:

The gap between contracted ARR (2.3B)andcurrentAIcloudquarterlyrevenue(17.3M = ~$69M annualized) is the largest leading-indicator divergence I can remember seeing. In SaaS terms, this would be like having $2.3B in signed contracts but only $69M in actual recognized revenue. That backlog is either the most compelling setup in the portfolio — or it never converts.

I don't invest based on hope. But I also recognize when a company has reduced the "hope" content through financing structure. The $3.6B GPU financing at <6% and the $1.9B Microsoft prepayment covering 95% of GPU CapEx materially de-risks the execution. The capital is there. The question is operational.


Valuation — The Math I Care About

Metric Value Notes
Market cap ~$12.1B As of 2026-02-22 [Ambiguous — verify]
TTM Revenue ~$757M Q3 FY25 + Q4 FY25 + Q1 FY26 + Q2 FY26
P/S (TTM) ~16x Rich
Contracted ARR $2.3B Forward revenue visibility
P/S (contracted ARR) ~5.3x More reasonable
$3.4B ARR target End CY2026 ~3.5x P/S if hit
Basic shares 298M
Fully diluted (incl. converts) ~444M 49% potential dilution
Fully diluted market cap ~$17.8B At $40/share — material difference
EV/Contracted ARR (fully diluted) ~7.7x After adding $3.7B debt, subtracting 3.3Bcash : EV 17.4B / $2.3B

The dilution math is the piece I keep coming back to. If I buy IREN today at $40/share on 298M shares, the true cost includes the 146M convertible shares sitting in the capital structure. Fully diluted, I'm paying 17.8BmarketcapandEVof 17.4B (adding $3.7B converts, subtracting $3.3B cash). Against $2.3B contracted ARR, that's 7.7x EV/contracted ARR. Against $3.4B target: 5.1x. That's not crazy for a platform AI infrastructure company with Microsoft as anchor, but it leaves limited margin of safety.

Atlas pegged conviction at 3.5/5. I'd call it 3/5 for my framework — the valuation math is tighter on a fully diluted basis than it looks at first glance, and I'm not comfortable with 84% customer concentration in Microsoft. Great customer, but a single relationship.


What I'm Watching — The Transition Tracker

Metric Q425 Q126 Q226 Target On Track?
AI Cloud Revenue (quarterly $m) 7.0 7.3 17.3 ~$200M+/q by Q4 CY2026 Too early to say
BTC Mining Revenue ($m) 180.3 232.9 167.4 Declining (intentional) Yes
Adj EBITDA Margin 65% 38% 41% Should recover as AI cloud scales Flat — watching
ARR Contracted ($B) ~1.94 ~1.94 2.3 3.4 by Dec 2026 Progress
Operating MW 810
Secured Power (GW) 3.0 4.5 Oklahoma adds diversity

The critical path: Sweetwater energizes Q2 CY2026 → Microsoft revenues commence → AI cloud quarterly revenue steps change from $17M to $100M+ → TTM P/S starts to look reasonable. Every quarter that slips is expensive at 16x trailing.


Red Flags and Green Flags

Green flags:

Red flags:


Management Credibility

This is a founder-led company (Daniel and William Roberts). The Microsoft deal is the clearest proof of credibility — that's not a deal a weak management team lands. The GPU financing at <6% from Goldman and JPMorgan also signals institutional confidence.

That said, the rolling tracker shows a pattern of quarterly EPS misses across Q3 FY25 through Q1 FY26. IREN doesn't provide quarterly guidance, which makes "miss" somewhat unfair — but the market's expectations have consistently been ahead of delivery. Management is building something big, and big things don't always stay on schedule.

Promise tracking:


Thesis Assessment

Haven't written about IREN before. This is my first look. Applying my framework:

IREN is not a software company. NRR, billings, customer count — the normal tools don't apply. What applies here is: does the contracted backlog convert to revenue? Is the capital structure sustainable? Is the secular position real?

On backlog conversion: the contracted ARR ($2.3B) provides unusual visibility, but actual revenue recognition depends on infrastructure delivery. One switch flip at Sweetwater changes the revenue math entirely. I respect that the financing structure (GPU loans + prepayments) has reduced execution risk meaningfully — this is not a company relying on capital markets staying open. But infrastructure build timelines have a history of slipping.

On capital structure: the $3.7B in convertible notes at sub-1% coupon is brilliant financing — cheap capital, deferred dilution. The 49% potential dilution is the price. I'd have preferred a simpler structure, but I can't argue with the economics.

On secular position: real. AI infrastructure scarcity is real. IREN's 4.5 GW pipeline is differentiated — you can't replicate that in 18 months. The "time to data center" comment resonates. This is a durable competitive position if they execute.

Thesis status: Intact, but unproven. The thesis requires Q2 and Q3 CY2026 proof points (Sweetwater energization, Microsoft revenue commencement, AI cloud quarterly revenue stepping to $50M+). Until those arrive, this is a high-conviction bet on execution, not a confirmation of execution.

I could be wrong — the Sweetwater concerns may be overdone, and the Microsoft relationship may be more locked in than the risk disclosures suggest. The 5.7% position size in wsm007 seems appropriate — meaningful exposure without bet-the-portfolio risk.


Action

Hold. At 16x trailing P/S, I would not be adding aggressively. The current price fairly reflects the contracted ARR trajectory but leaves limited margin of safety. The catalysts (Sweetwater, Microsoft revenue commencement) are 1-2 quarters away. I want to see Q3 FY26 numbers before adding — by then, the AI cloud revenue ramp should be clearly visible and the execution story either confirmed or challenged.

If Sweetwater energizes on schedule and Microsoft revenues commence meaningfully in Q2 CY2026, the position deserves to grow. If either slips, trim.

Bear


Sources: IREN Q2 FY26 earnings call transcript (Motley Fool, 2026-02-05); Scout brief IREN_earnings-review_2026-02-21; Atlas baseline IREN_earnings-review_Q2_FY26.md; companies/IREN.md; earnings/IREN/_ROLLING.md