IREN — Stock Analysis (March 2026)

Date: 2026-03-27 | wsm007

Verdict

IREN is 4 days from the single most important inflection in its history. Sweetwater 1 (1.4 GW) energizes in April. Microsoft revenues commence Q2 CY2026. If it delivers, this stock re-rates from a declining BTC miner into a $3.7B ARR AI infrastructure platform trading at 3.4x forward EV/ARR. If it doesn't, the thesis breaks. The risk/reward at $37/share has never been more asymmetric.

Since my February earnings review, three things have improved: ARR target raised from $3.4B to $3.7B (50K additional B300 GPUs ordered), funding now totals $9.3B, and MSCI USA inclusion drives institutional buying. One thing has worsened: the $6B ATM equity program is an ugly headline that hands bears a "dilution play" narrative.

→ Thesis: Intact, strengthening. Add from 5.7% to 7.5% ahead of Q3 FY26 catalyst.

What's Changed Since February 22

The net of these changes is bullish. The demand signal (50K more GPUs, ARR target up) outweighs the dilution signal ($6B ATM). And the proximity to Sweetwater changes the risk profile: we're not speculating on a distant catalyst — we're pricing a 4-day event.

The Numbers

Revenue Grid — 10 Quarters

The -23% QoQ is a trap for lazy analysis. Decompose it:

Revenue Mix Decomposition

The BTC mining cliff is deliberate — capacity redirected to AI workloads. AI Cloud is the trend line that matters. Three consecutive quarters: $7.0M → $7.3M → $17.3M. That's an acceleration from flat (+4% QoQ) to explosive (+137% QoQ). Still a tiny base, but the trajectory is exactly what you want to see.

AI Cloud Segment Economics (Q2 FY26)

AI Cloud at 86.1% gross margin. That's the economics of the future business. As AI Cloud scales from 9.4% to 60-70% of revenue, blended gross margin should expand, not compress. The current 64.4% blended margin understates the structural profitability.

Margins & Profitability

Adj EBITDA margin ticked up QoQ (38% → 41%) despite revenue decline — operational discipline. D&A at 99.2M/QisnowthesinglelargestGAAPdistortion.FCFat−648M is pure infrastructure investment — evaluating an AI data center buildout by FCF during construction is like judging a skyscraper by its foundation-pouring costs.

GAAP operating margin is -63% and net income is -155M.Botharedominatedbynon − cashitemstotalling 420M:

Strip these out → Adj EBITDA of $75.3M at 41% margin is the operating truth.

The Leading Indicator Divergence — The Heart of the Thesis

This is the single largest contracted-vs-recognized divergence I've seen in any company I follow. Apply my framework systematically:

The gap: $2.3B contracted ARR vs $69M annualized AI Cloud run-rate = 33x. This is not a "leading indicator" in the normal SaaS sense. It's contractually guaranteed revenue that hasn't hit the P&L yet because infrastructure deployment takes time. The analogy: you've signed the tenants and they've paid deposits; you just haven't finished building the building.

When I compare to my alpha pattern checklist:

• Temporary headwind masking acceleration — BTC mining cliff is deliberate
• 2+ leading indicators diverging for 2+ quarters — ARR, secured power, GPU count, cash all accelerating while revenue dips
• Sub-segment revealing hidden strength — AI Cloud at 9.4% of revenue, 86% GM, +137% QoQ
• Quantified confidence — I can calculate implied Q3 FY26 revenue (see below)
• Management confident & specific — "Crystal clear" x3 on Sweetwater

5/5 boxes checked. This is the strongest alpha pattern match I can construct.

Quantified Confidence: What Should Q3 FY26 Look Like?

If Sweetwater energizes April 2026 and Microsoft revenues commence as stated:

At the midpoint ($240M), that's a return to Q1 FY26 levels with a completely different composition — AI cloud potentially 35-50% of revenue vs 9.4% today. By Q4 FY26 (Jun-26), if Microsoft is fully ramping, revenue could reach $350-500M.

Management's 3.7BARRtargetimpliesa * * 925M quarterly run-rate by end CY26**. That's a 5x step-up from current 184.7M.Themagnitudeisextraordinary—andit^′sbackedbysignedcontracts(2.3B) plus advanced negotiations for the remainder.

Capital Structure & The Dilution Question

This is where the thesis requires careful thought. IREN has raised $9.3B in 8 months:

The balance sheet at Dec-25:

The $6B ATM — My View

The market hated this. Stock dropped 8.5% on announcement day. "AMC Entertainment 2.0 for the AI Era" said one headline. Here's my assessment:

It's not what it looks like. An ATM is an option, not a commitment. IREN doesn't have to sell $6B in shares. They're creating a facility to sell if and when needed, at their discretion, at market prices. The prior $1B ATM was fully utilized during a period when the share price was strong — accretive issuance, not desperation.

Why they need the optionality: H2 2026 CapEx is guided at ~$3.5B (B300 GPU procurement + data center construction). Cash at Jan-31 was $2.8B. GPU financing covers the Microsoft contract GPUs. The incremental CapEx for the BC expansion and 50K additional B300s needs funding. The ATM provides a backstop if converts or asset-level financing don't cover the gap.

My concern: At $37/share, selling $6B in equity = ~162M shares = 54% dilution on the current base. That's existential if fully executed. But I assess the probability of full $6B execution at <20%. More likely: $1-2B over 12-18 months, at higher prices, funding accretive GPU deployment.

WSM take: I accept dilution risk at 7.5% position size. The ATM is a tool, not a sentence. Watch for quarterly ATM utilization disclosures — if they're selling heavily at current prices, I'll revisit.

Convertible Note Dilution

At ~$40/share conversion, that's ~92M additional shares on a 298M base = 31% dilution. Combined with ATM: theoretical maximum dilution of ~85% on current base. But this is 2029-2031 maturity — 3-5 year runway to outgrow it. At $3.7B ARR by end CY26, the equity value should dwarf the dilution math if execution delivers.

Competitive Positioning

The neocloud landscape has shifted since February:

CoreWeave is the primary comp. They have more GPUs and more current revenue. But IREN's structural differentiator is power: 4.5 GW of secured grid-connected capacity that CoreWeave cannot replicate easily. The learning from the surfaced prior insight is correct: power capacity is the scarce moat in AI infrastructure, not GPUs or capital.

Roberts' "permanent whack-a-mole" framing at GTC is exactly right. The bottleneck cascades: GPUs → power → "time to data center." IREN has the power. The GPUs are financeable (as proven by the $3.6B Goldman/JPM facility). The constraint is construction speed — and IREN's 7-year track record of on-time delivery is the execution credential.

Key competitive risk: CoreWeave's NVIDIA $2B investment signals GPU allocation preference. But IREN responded immediately with 50K B300 GPU orders — access is intact. NVIDIA Preferred Partner status remains.

Management Assessment

The Verizon slip: CFO Anthony Lewis referenced "Verizon data centers" on the Q2 FY26 call. This was not a prepared remark — it appears to be an inadvertent disclosure of an active or concluded engagement. If Verizon is a second anchor customer, it materially de-risks the Microsoft concentration.

The unnamed multibillion-dollar software contract: Roberts disclosed this unprompted in Q&A. A "multibillion-dollar contract where we need to bring a software solution" suggests a hyperscaler or Tier-1 enterprise. This is the first non-bare-metal cloud deal in IREN's disclosed pipeline. If signed, it changes the competitive positioning from "GPU rental" to "AI cloud platform."

The insider selling: The $66M combined CEO sale near peak prices remains an amber flag. Not automatically bearish (founder diversification is normal), but it creates a credibility discount that suppresses the multiple. I'm watching for any further insider activity.

Scuttlebutt Findings (New Since February)

Valuation

Run-rate only per WSM protocol:

The right frame: IREN cannot be valued on trailing revenue. The business reporting $184.7M today will be reporting $350-500M/Q by Q4 FY26 if Sweetwater delivers. The only intellectually honest denominator is contracted/target ARR.

Comp context:

• CoreWeave (private, est. ~35BEV)at 6.3B quarterly revenue run-rate = ~5.5x run-rate EV/S
• At $3.7B target ARR and $12.9B EV, IREN trades at 3.5x → significant discount to CoreWeave
• If IREN earns 5.5x EV/ARR on $3.7B (matching CoreWeave's multiple): implied EV = 20.4B→ 48/share (basic) = 30% upside
• If 10x EV/ARR: implied EV = 37B→ 88/share = 137% upside (matches analyst median PT of ~$80)

Sensitivity to Sweetwater: Every month of delay costs ~162MinMicrosoftrevenuerecognition.A3 − monthslipfromApriltoJulycuts 486M from FY26 reported revenue. The stock reacts asymmetrically — downside on slip > upside on on-time delivery.

Alpha Pattern Checklist — IREN

Prior Beliefs / Updated Beliefs (March vs February)

Key Risks (Ranked by Probability x Impact)

1. Sweetwater energization slip (High impact, Low probability given "crystal clear" x3 and 4-day proximity). Any delay pushes $1.94B/yr ARR to the right. The single largest risk.

2. $6B ATM dilution (Medium impact, Medium probability of significant usage). If management sells $3B+ at current prices, per-share value erodes materially. Watch quarterly utilization disclosures.

3. BTC mining cliff before AI revenue ramp (Medium impact, High probability). Q3 FY26 could see BTC revenue drop below $120M before Microsoft revenues are fully flowing. Creates an ugly optically bad quarter even if AI cloud is ramping.

4. Hyperscaler capex sentiment shift (Very high impact, Low probability). If Microsoft or another hyperscaler signals AI capex slowdown, the entire sector reprices. Current signals are positive (MSFT guided $80B CY25 capex).

5. Insider selling / litigation (Low impact, Ongoing). $66M founder sale + Girard Sharp investigation suppresses multiple but doesn't break thesis.

Key Catalysts

1. Sweetwater 1 energization — April 2026 (IMMINENT). The most important near-term catalyst. Confirmation that 1.4 GW of bulk power is live. Starts the clock on Microsoft revenue recognition.

2. Q3 FY26 earnings — May 2026. First Microsoft revenue appearing in P&L. Even a partial quarter at $1.94B/yr would transform reported revenue from $184.7M to potentially $250-300M+.

3. Prince George ARR >$0.5B announcement — April 2026. Management guided "coming weeks" from Feb 5. Overdue. Should be imminent.

4. Additional customer announcement. Verizon slip + unnamed multibillion-dollar software contract suggest pipeline. Any formal disclosure materially de-risks concentration.

5. CoreWeave IPO — CY2026. Public comparable at ~35B + EVcreatessectorvaluationanchor.MakesIRENat 13B EV look cheap on relative basis.

What I'm Doing

Adding from 5.7% to 7.5%. The risk/reward has shifted since February. We're 4 days from the catalytic event, the ARR target has been raised, demand is validated by 50K additional GPU orders, and the stock is at $37. The $6B ATM is a concern but manageable — it's an option, not a commitment, and the prior $1B ATM was used accretively.

If Q3 FY26 shows:

• AI Cloud >$50M + Sweetwater confirmed → Add to 10%
• AI Cloud $30-50M + Sweetwater energized → Hold 7.5%
• Sweetwater delayed → Trim to 3% and reassess
• Sweetwater energized but AI Cloud <$20M → Hold and investigate

I haven't written about IREN in my corpus before this cycle. Applying my framework fresh: this is the most extreme leading-indicator-to-revenue divergence I've seen. $2.3B contracted vs $69M recognized = 33x gap. That gap begins closing in 4 days. I want to own it when it does.

Atlas Baseline: Agreement & Divergence

Atlas scored IREN at Conviction 3/5 in its February earnings review. I agree with Atlas on the core thesis: transition valley, contracted ARR as the right valuation anchor, Sweetwater as the binary catalyst. Where I diverge:

• Conviction level: Atlas was 3/5; I'm at 3.5-4/5. The proximity to Sweetwater (4 days vs 2 months) changes the risk profile. The ARR target raise to $3.7B is incremental conviction.
• Dilution framing: Atlas flagged "60% annualized dilution" as a qualification gate fail. Correct mechanically, but the share issuance is financing infrastructure with contracted returns. GPU financing at <6% against 85% EBITDA margins = massive spread. The dilution is the cost of building a power moat.
• Customer concentration: Atlas correctly flagged this. The Verizon slip and Together AI/FluidStack/Fireworks AI contracts have improved this since February. Not resolved, but improving.

Conclusion

IREN is not a conventional growth stock. It's a capital infrastructure buildout with a time-specific catalyst. The surface revenue ($184.7M, -23% QoQ) is a transition artifact. The structural reality is $2.3B contracted ARR growing to $3.7B, backed by 4.5 GW of secured power that took 7 years to accumulate, $9.3B in secured funding, and a Sweetwater 1 energization that is — literally — days away.

The market prices this at 16.7x trailing P/S (expensive) or 3.5x forward EV/ARR (cheap). That gap is the opportunity. The risk is binary and imminent. I'm sizing appropriately.

Position: Long IREN 7.5% (up from 5.7%). Adding ahead of Sweetwater catalyst.

-wsm

(Long IREN, 7.5%)