Date: 2026-02-22 Quarter: Q2 FY26 (Dec 2025) Task: Earnings Review Atlas Baseline Read: Yes — IREN_earnings-review_Q2_FY26.md
I want to begin by acknowledging the peculiar nature of this company, because it does not map cleanly onto the framework I developed watching Dow Chemical, Texas Instruments, and Motorola. IREN is not a company that iterates on products through research laboratories, nor does it serve a broad market of thousands of customers. It is something else entirely: a company that had the foresight to secure an extraordinarily scarce resource — 4.5 gigawatts of grid-connected power capacity — and then built the engineering organization capable of monetizing it at the precise moment history demanded it.
In my many years of investing, I have found that the greatest opportunities arise when a company does something that others recognize as obvious only in retrospect. The Bitcoin-to-AI-cloud pivot was not obvious to most observers. It was obvious to the Roberts brothers. That is what I wish to examine most carefully.
Before any number, I insist on understanding what competitors, customers, and close observers say about a company. I conducted the following research:
Competitive positioning: External analysts confirm that IREN's vertical integration — designing, building, and operating its own data centers — delivers a structural cost advantage of approximately $4 million per megawatt versus the industry standard of 10million.Thisisa6055 billion backlog) but carries substantially heavier debt obligations and does not own its own power infrastructure. (Sources: Nasdaq comparative analyses; BeyondSPX analysis; ainvest.com)
Management reputation: Daniel Roberts brings 20+ years of infrastructure and finance experience — Macquarie Capital, PricewaterhouseCoopers, and Palisade Investment Partners (where he was the second-largest individual shareholder in an infrastructure fund managing billions across 23 businesses). The leadership team collectively claims delivery of over $25 billion of projects globally. These are not technology promoters or cryptocurrency speculators; they are infrastructure builders who happened to find Bitcoin mining as the first application for their power expertise. That is a crucial distinction. (Source: IREN leadership team, Crunchbase)
Management candor under pressure: During the Q2 FY26 earnings call, multiple analysts pressed specifically on ERCOT interconnection risks for Sweetwater — a legitimate concern given that the large-load interconnection queue in Texas has nearly quadrupled in a single year and ERCOT's own analysis identifies administrative coordination as the dominant challenge. Daniel Roberts was direct: the 2023 interconnection agreement is signed and filed; Sweetwater is likely in "Batch Zero." He did not deflect or use euphemism. This is precisely the behavior I look for in management: candor on difficult questions, not only on triumphant announcements. (Sources: Motley Fool earnings transcript; Latitude Media; ERCOT LLWG meeting summary)
Employee signals: I found only approximately five Glassdoor reviews for Iris Energy specifically. The company has grown to 2,000+ employees, implying rapid hiring that has outpaced the review corpus. Forty-six open positions are currently listed on job boards, including network infrastructure roles at Sweetwater and Childress facilities. This is an active hiring posture consistent with a company mid-buildout, not one that has stalled. The absence of negative employee patterns is mildly positive — I found no evidence of the kind of cultural dysfunction or revolving-door executive departures that often precede operational failure. (Sources: Glassdoor; Indeed job listings)
Analyst reaction post-earnings: Despite an EPS miss of -0.52versusconsensusof−0.18, the stock rose approximately 5% in the session following earnings. The average analyst price target stands at approximately $79-80 per share, with Canaccord Genuity establishing a $70 target on February 10, 2026. The consensus is bullish with 10 buy ratings. Analysts who understand infrastructure execution recognized that the headline miss was driven by non-cash items and a deliberate strategic transition, not operational deterioration. (Sources: TipRanks; Benzinga; Canaccord Genuity)
Vertical integration moat confirmed externally: The 85% expected EBITDA margin on the Microsoft contract — cited repeatedly in external analysis — is possible only because IREN owns the land, the power, the substation, the data center design, and the operations. Each layer of the stack that a competitor must outsource represents margin leakage to a third party. IREN captures it all internally. (Sources: BeyondSPX; ainvest.com)
Fisher's Fifteen Points are a qualitative checklist, not a scoring rubric. Let me apply them with appropriate rigor.
Point 1 — Sufficient market potential for substantial further sales growth? PASS. The AI infrastructure demand cycle is in its early innings. IREN's $3.4 billion ARR target for end of CY2026 utilizes only 10% of its secured 4.5 gigawatt capacity. The remaining 90% represents the growth runway. Whether that demand persists through the decade is a question I cannot answer with certainty, but the current magnitude of hyperscaler capital commitments is unlike anything I observed in the semiconductor cycles of the 1960s and 1970s.
Point 2 — Determination to develop new products/services and pursue total sales growth? PASS. The pivot from Bitcoin mining to AI cloud was not inevitable — it required a deliberate decision to cannibalize an existing revenue stream (mining) in favor of a higher-value future (contracted cloud services). This is exactly the kind of forward-looking management behavior I admire. The acquisition of the Oklahoma 1.6 gigawatt site — adding jurisdictional diversity beyond ERCOT — demonstrates continued strategic initiative rather than mere execution of an existing plan.
Point 3 — Effective R&D relative to size? MODIFIED PASS. IREN does not have laboratories or patent portfolios in the traditional sense. What they have is an engineering organization that can convert a greenfield power site into a functioning data center faster and cheaper than competitors. The "R&D" is embedded in their construction methodology, cooling design expertise, and operational systems. External analysis confirms this through the $4M/MW cost metric. For an infrastructure company, this engineering excellence is the functional equivalent of research productivity.
Point 4 — Above-average sales organization? MODIFIED PASS. IREN does not employ a conventional sales force. Their "sales" function is direct relationship management with hyperscalers and large AI enterprises. The evidence here is the $9.7 billion Microsoft contract plus "multiple advanced negotiations underway for larger-scale deployments." Management noted that one pending contract would require IREN to bring a software solution — indicating that customers are actively pulling IREN toward expanded scope. This is the best evidence of a strong selling function: customers seeking you out rather than you seeking them.
Point 5 — Worthwhile profit margin? MIXED. Adjusted EBITDA margin of 40.8% [Non-GAAP] is respectable. GAAP gross margin of 64.4% is solid but declining from a peak of 72.2% in Q2 FY25, primarily because AI cloud revenue (still ramping) carries lower margins than the legacy mining operations during this transition period. The expected 85% EBITDA margin on the Microsoft contract at scale suggests the trajectory is favorable once the revenue mix completes its shift.
Point 6 — What is management doing to maintain or improve profit margins? PASS. The deliberate shift from commodity Bitcoin mining (subject to global hashrate competition and Bitcoin price cycles) to contracted AI cloud services (5-year agreements with prepayments) is the most decisive profit margin strategy I could ask management to execute. One is a commodity; the other is a contracted utility-like service. The margin improvement through this transition is structural, not cosmetic.
Point 7 — Outstanding labor relations? INCONCLUSIVE. Insufficient data. The rapid headcount expansion to 2,000+ employees with very few public complaints suggests no acute problem. The active hiring of specialized infrastructure engineers signals the organization can attract talent. I cannot draw a firm conclusion here.
Point 8 — Outstanding executive relations? PASS. The co-CEO structure (Daniel and William Roberts, brothers) is unusual, but the evidence suggests complementary function rather than conflict. Kent Draper (Chief Development Officer) presented with substance and precision on the earnings call — he clearly owns his domain. Anthony Lewis (CFO) provided straightforward financial commentary. The team appears to function with genuine depth rather than a single charismatic leader surrounded by order-takers.
Point 9 — Depth to management? PASS, with monitoring. Three distinct senior leaders — Daniel Roberts (strategic/commercial), William Roberts (not frequently quoted but co-founder), Kent Draper (construction/operations), Anthony Lewis (finance) — each appear to have genuine competence in their domain. A company executing a $9.7 billion infrastructure buildout across multiple geographies requires this depth. I would want to monitor succession — if key people depart, the construction execution thesis weakens.
Point 10 — Cost analysis and accounting controls? PASS, circumstantially. The $4 million per megawatt construction cost versus the $10 million industry standard cannot be achieved without rigorous cost discipline. The $3.6 billion GPU financing secured at sub-6% interest rates from Goldman Sachs and JPMorgan — with Microsoft's prepayments covering 95% of GPU CapEx — reflects sophisticated financial engineering. One does not achieve these terms without demonstrating tight controls to institutional lenders.
Point 11 — Aspects peculiar to the industry that give clues about competitive position? STRONGLY PASS. This is where IREN's story becomes genuinely interesting. The scarcest resource in the AI infrastructure buildout is not GPUs (which can be manufactured) and not capital (which is abundant) — it is power interconnection rights, specifically at large scale, with signed agreements, in renewable-rich regions. IREN's 2023 interconnection agreement at Sweetwater predates the quadrupling of the ERCOT large-load queue. They filed for interconnection when the queue was short; competitors filing today face years of delay. This temporal advantage cannot be replicated with money. It is the equivalent of having secured a patent on a key chemical process before the market recognized its value.
Point 12 — Short-range or long-range outlook in regard to profits? PASS. The willingness to accept a -23% QoQ revenue decline as mining capacity reallocates to AI cloud — forgoing near-term mining profit in favor of contracted cloud contracts — is precisely the long-range thinking I value above all else. Management explicitly stated they are selecting long-term partners carefully rather than maximizing immediate throughput. This is the language of stewardship, not speculation.
Point 13 — Foreseeable equity financing needs? CONCERN. This is the one area where I must register genuine reservation. The $3.7 billion in convertible notes could produce approximately 146 million additional shares — representing a 49% increase in basic share count of 298 million. I have generally been skeptical of companies that must constantly return to capital markets to fund growth. The convertibles are cleverly structured (sub-1% coupon, conversion features that only dilute if the stock performs well), and the 95% CapEx coverage from customer prepayments plus GPU financing means the equity dilution is not funding operations but rather the conversion mechanics of cheap debt. Nevertheless, a shareholder today who expects to own 5.7% of IREN may find themselves owning 3.8% at full dilution. This warrants ongoing attention.
Point 14 — Does management talk freely when things are going well, but clam up when troubles occur? PASS. The ERCOT question during the Q2 FY26 call was a genuine pressure point. Management did not deflect; they addressed it specifically, credibly, and with reference to the actual contractual agreement date. This is the behavior I look for. The CFO's characterization of GAAP losses as driven by non-cash derivative items was accurate and verifiable — not an attempt to obscure poor operating performance.
Point 15 — Unquestionable management integrity? PASS. No evidence of self-dealing, related-party transactions, or misleading disclosure. Founder-led with significant insider ownership. The Roberts brothers built this company and own a meaningful stake — their interests are aligned with shareholders. The GPU financing terms (secured at institutional-grade rates from Goldman Sachs and JPMorgan) imply that sophisticated counterparties with access to IREN's books have found the management trustworthy.
I have spent many years distinguishing between companies that are "fortunate and able" — in the right place at the right time, riding industry tailwinds — and companies that are "fortunate because they are able" — management that creates advantage through superior execution and foresight. The distinction matters enormously because only the latter reliably compounds over time.
IREN sits at the intersection of both descriptions, and one must think carefully about which force is dominant.
The AI infrastructure demand cycle is clearly a powerful tailwind — IREN benefits from a secular force they did not create. In this sense, they are "fortunate." But their ability to be positioned correctly — to have secured 4.5 gigawatts of interconnection rights before the queue quadrupled, to have built the engineering organization that can deliver at $4M/MW while competitors spend $10M/MW, to have navigated from a Bitcoin mining operation to a Microsoft infrastructure partner — this is management creating advantage. It is not luck.
I have observed in decades of investing that the companies that endure are those where management uses tailwinds as the opportunity to build durable structural advantages. The question for IREN is whether the 4.5 gigawatt power pipeline and the vertically integrated construction capability constitute a genuinely durable moat, or whether well-capitalized competitors can replicate it by waiting in interconnection queues and hiring construction teams.
My assessment: the power interconnection timing advantage is real and multi-year in duration. The construction cost advantage ($4M vs $10M/MW) reflects accumulated organizational capability that cannot be instantly replicated. But I would not call this a permanent moat. It is a substantial lead in a footrace.
The headline numbers require decoding for anyone new to this company's story.
Revenue fell 23% sequentially to $184.7 million [GAAP]. This decline is entirely explained by the deliberate reduction of Bitcoin mining capacity — mining revenue fell from $232.9M to $167.4M (-28% QoQ) — as hashrate was reallocated to AI cloud workloads. Meanwhile, AI cloud revenue accelerated from $7.3 million to $17.3 million (+137% QoQ). The operational story is one of acceleration, not deceleration.
The GAAP net loss of $155.4 million is dominated by non-cash items: $219 million in unrealized derivative losses on convertible notes, $112 million in debt conversion inducement costs, and $32 million in mining hardware impairments. Adjusted EBITDA of $75.3 million at a 40.8% margin [Non-GAAP] is the more honest representation of operating performance — and even this margin figure is temporarily compressed by the transition, not structurally weak.
The truly important number in this quarter is not on the income statement. It is the contracted ARR figure: $2.3 billion. Against a trailing twelve-month revenue of $757 million, this means IREN has already contracted a revenue stream approximately three times its current run rate. The gap between contracted ARR and reported revenue represents future recognition — as Sweetwater energizes, as Prince George expands, as Microsoft contract GPU deployments ramp through H2 CY2026, this contracted revenue becomes reported revenue. The investor who exits because of a -23% QoQ decline is making a fundamental analytical error: confusing a transition quarter with a deteriorating business.
Promise tracking from prior periods: I note that the rolling earnings tracker records three consecutive "Miss" designations for Q3 FY25, Q4 FY25, and Q1 FY26. This warrants examination. IREN does not provide specific quarterly revenue guidance, so "miss" here refers to analyst consensus expectations, not management-set targets. The pattern of analyst misses likely reflects the difficulty of modeling a company mid-transition — analysts building models on Bitcoin mining trajectory when management was accelerating the AI cloud pivot. The more relevant accountability metric is management's own commitments: the Microsoft contract was delivered, the GPU financing was secured at better-than-expected rates (<6%), and Sweetwater remains on track. On management's own promises, the record is clean.
I want to be direct: precise valuation is not my method, and I resist pressure to name a price target. What I can say is the following.
At a current market capitalization of approximately $13.3 billion against trailing twelve-month revenue of $757 million, the stock appears expensive on any conventional trailing metric — 17.6x TTM revenue. Against the contracted ARR of $2.3 billion, enterprise value works out to approximately 7.5x — meaningfully more reasonable. Against the company's own target of $3.4 billion ARR by end of CY2026, the multiple falls to approximately 5x — inexpensive for a company with this quality of contracts and this degree of revenue visibility.
The dilution concern I raised in Point 13 is relevant here. On a fully diluted basis (including ~146 million conversion shares), the arithmetic worsens by approximately 49%. The investor must weigh the cheap funding (sub-1% coupon convertibles and <6% GPU financing) against the eventual dilution cost.
For a company of this caliber — Microsoft anchor customer, 4.5 GW power pipeline, vertically integrated cost advantage, founder management — I would not sell on the basis of a trailing revenue multiple. The time to sell is when the thesis breaks: if infrastructure delivery falls materially behind schedule, if Microsoft cancels or renegotiates, if management loses candor. None of these conditions are present today.
| Point | Description | Assessment |
|---|---|---|
| 1 | Market potential for substantial growth | PASS |
| 2 | Management determination to develop new opportunities | PASS |
| 3 | Effective R&D relative to size | MODIFIED PASS |
| 4 | Above-average sales organization | MODIFIED PASS |
| 5 | Worthwhile profit margin | MIXED |
| 6 | Maintaining/improving profit margins | PASS |
| 7 | Outstanding labor relations | INCONCLUSIVE |
| 8 | Outstanding executive relations | PASS |
| 9 | Management depth | PASS (monitor) |
| 10 | Cost analysis and accounting controls | PASS |
| 11 | Industry-specific competitive clues | STRONGLY PASS |
| 12 | Long-range profit outlook | PASS |
| 13 | Equity financing needs | CONCERN |
| 14 | Management candor under adversity | PASS |
| 15 | Unquestionable management integrity | PASS |
Summary: 12 PASS, 1 MODIFIED PASS (×2), 1 MIXED, 1 INCONCLUSIVE, 1 CONCERN.
Under my framework, a company must pass on a strong majority of the Fifteen Points to warrant serious consideration. IREN does so. The concern on Point 13 (dilution) is real but mitigated by the structural nature of the convertibles. The MIXED on Point 5 (margins) is a transition artifact, not a structural deficiency. No automatic disqualifiers — Point 15 (integrity) passes cleanly.
Execution risk on infrastructure delivery. This is the dominant risk. A company that has committed to 140,000 GPU deployments by end of CY2026, has accepted $1.9 billion in Microsoft prepayments, and has borrowed $3.6 billion in GPU financing cannot afford material construction delays. If Sweetwater misses its Q2 CY2026 energization date, the revenue ramp slows, covenants come under pressure, and management credibility erodes. I will watch Q3 FY26 earnings (approximately May 2026) for any language shift on this milestone.
Customer concentration. Microsoft represents $1.94 billion of $2.3 billion contracted ARR — 84%. This is a degree of customer concentration I have historically been uncomfortable with. Microsoft is a superb customer in terms of creditworthiness and contractual commitment, but any single customer dependency creates vulnerability. Management mentioned "multiple advanced negotiations underway" for multibillion-dollar deployments. Every additional contract signed reduces this risk.
ERCOT regulatory environment. The ERCOT large-load integration queue has quadrupled. While IREN's 2023 interconnection agreements pre-date this surge and management is confident about "Batch Zero" designation, regulatory environments can change. Texas political dynamics around data center power consumption deserve monitoring.
AI demand durability. If the hyperscaler AI infrastructure buildout moderates — whether from technology efficiency improvements (DeepSeek-style), economic slowdown, or changed strategic priorities — IREN's growth assumptions weaken significantly. The company will have committed to massive capital deployments predicated on sustained demand.
IREN is not a textbook Fisher growth stock in the mold of Dow Chemical or Texas Instruments. It is an infrastructure company that has positioned itself at the most valuable intersection in the current technology cycle — AI compute demand meeting power scarcity — through a combination of foresight, engineering excellence, and founder conviction.
The Q2 FY26 earnings report, superficially disappointing, is in substance an affirmation of the thesis. The deliberate mining-to-cloud transition is accelerating. The contracted ARR of 2.3billiondwarfsthecurrentrevenuerunrate.Thefinancingstructure(3.6 billion at sub-6%) was secured at better terms than the market expected. Management demonstrated candor on the most difficult question (ERCOT risk). The Fifteen Points are largely satisfied.
The critical variable for the next twelve months is infrastructure execution: do the GPUs deploy on schedule, does Sweetwater energize in Q2 CY2026, do Microsoft contract revenues commence as promised? If yes, the revenue trajectory inflects sharply upward in H2 CY2026, and the current valuation will look cheap in retrospect.
My view: IREN passes the Fisher qualitative screen convincingly. For a position already held (as in the wsm007 portfolio at 5.7%), I would hold and watch the May 2026 earnings call for execution confirmation. I would not add on thesis alone — the valuation is not a bargain on trailing metrics, and I prefer to add to great companies at reasonable prices rather than momentum chases. But I would certainly not sell. The time to sell is almost never when the business remains sound and management remains candid. Neither condition has changed.
Thesis status: INTACT — Strengthening. The Q2 FY26 quarter confirms the strategic logic. The key monitoring point is Sweetwater energization and Microsoft contract revenue commencement in Q2 CY2026.