IREN (IREN): AI Cloud Contracts Drive the Re-Rating

Thesis

IREN(IREN) is no longer a simple Bitcoin miner with an AI side project. The core investment case is that it is converting scarce power, owned data center infrastructure, and capital access into a contract-backed AI cloud platform at unusual speed. The hard evidence is substantial. In Q3 FY26, revenue was $144.8M. The company reported $3.1B of ARR under contract as of May 7, 2026 and targeted $3.7B of ARR by the end of CY26. Earlier, management said the business had secured a $9.7B AI contract with Microsoft, underwriting commitments for $3.6B of GPU financing at less than 6%, and more than 4.5 GW of secured power. That combination matters because in AI infrastructure, power and deliverable capacity are the choke points, not slide-deck ambition.

The bull case rests on three facts. First, FY25 revenue rose to $501.0M from $187.2M in FY24, while FY25 net income reached $86.9M versus a $28.9M loss a year earlier. Second, the company is moving from volatile Bitcoin-linked revenue toward longer-duration AI cloud contracts, including the Microsoft agreement and a 5-year, $3.4B AI Cloud contract with NVIDIA disclosed in the May 2026 earnings context. Third, IREN ended April 2026 with $2.6B cash according to the Q3 FY26 business update, giving it real funding capacity for expansion rather than a hope-and-pray capital plan.

The bear case is just as real. Quarterly results remain noisy because the company is decommissioning mining hardware ahead of GPU installations, and that transition can make reported revenue and earnings look messy. Q3 FY26 net loss was $(247.8)M even as adjusted EBITDA was $59.5M. The stock also carries a rich valuation on current numbers, with trailing P/E of 39.48, forward P/E of 62.5, EV/revenue of 22.45, and PEG of 4.05. Add a beta of 4.18 and a 52-week range of $7.29 to $76.87, and this is plainly not a sleep-well-at-night utility.

For a balanced, moderate-risk investor with a medium-term horizon, the right stance is constructive but disciplined. IREN has enough verified contracts, power assets, and financing to justify serious attention. It also has enough execution risk, customer concentration, and valuation stretch to rule out blind enthusiasm. The stock fits investors who want exposure to AI infrastructure with more upside torque than mature data center operators, and more operating substance than many AI-adjacent stories that are still selling dreams by the gigawatt.

Company Overview

IREN Limited operates a vertically integrated digital infrastructure platform spanning Bitcoin mining, AI cloud services, and data center development. The company was formerly known as Iris Energy Limited and changed its name to IREN Limited in November 2024. It is listed on NASDAQ under the ticker IREN, was incorporated in 2018, and is based in Sydney, Australia. Corporate data lists 257 employees, while management said on the Q2 FY26 call that it had assembled a team of over 2,000 people to execute across development and operations. The difference likely reflects reporting conventions around direct employees versus broader operating teams, but the key point is that this is no longer a tiny operator.

The historical business was Bitcoin mining. The FY25 10-K shows Bitcoin mining revenue of $484.6M and AI Cloud Services revenue of $16.4M, for total revenue of $501.0M. That split makes the current transition easy to see: the legacy engine still pays most of the bills, but management is redirecting capital and capacity toward AI workloads that carry larger contract values and, in theory, more durable economics. In plain English, IREN is trying to move from selling exposure to Bitcoin economics toward selling scarce compute infrastructure.

Leadership remains founder-led. William Roberts and Daniel Roberts serve as Co-Founders and Co-CEOs. Anthony Lewis is CFO, David Shaw is COO, and Denis Skrinnikoff is CTO. Founder-led structures can cut both ways, but here the operating record matters. Management has built out large-scale power access, energized data center capacity, and signed major AI contracts. That is more persuasive than polished language about optionality.

Market classification data labels IREN under Financial Services and Capital Markets, but that is a poor fit. The actual business belongs in digital infrastructure, Bitcoin mining, and AI compute. That matters for valuation and peer framing. Comparing IREN to custody banks would be like comparing a power turbine to a checking account. The right questions are about power, GPUs, data center delivery, contract quality, and capital intensity.

Business Segment Deep Dive

IREN’s business has two real economic engines today: Bitcoin mining and AI cloud infrastructure. A third layer, data center development and build-to-suit capacity, supports both. The FY25 10-K provides the cleanest reported split. Bitcoin mining generated $484.6M of FY25 revenue, while AI Cloud Services generated $16.4M. That means AI was still small in the reported base, but the company’s contract pipeline and capex allocation show where management expects the mix to go.

Bitcoin mining remains important because it monetizes existing power and data center assets while AI capacity ramps. In Q2 FY26, management said total revenue was $184.7M, down 23% sequentially, primarily due to lower Bitcoin mining revenue from reduced Bitcoin mined, lower operating hashrate during the AI transition, higher global hashrate, and lower average Bitcoin prices. In Q3 FY26, revenue declined again to $144.8M, with management citing lower average Bitcoin price plus decommissioning of mining hardware ahead of GPU installation and billing. That is the transition in one sentence: mining is being cannibalized to make room for AI.

AI cloud is the strategic growth segment. Management said on the Q2 FY26 call that it had approximately $2.3B of annualized revenue run rate under contract, including around $0.4B at Prince George, and was on track for $3.4B ARR by end-2026. By Q3 FY26, the reported figure had moved to $3.1B ARR under contract, with a target of $3.7B ARR by the end of CY26. That progression is the most important operating signal in the story. Reported AI revenue is still catching up, but contracted revenue is scaling much faster.

The infrastructure development segment is the hidden backbone. Management described the business through the '3 Cs' of capacity, customers, and capital. On capacity, it cited 810 MW of existing data centers and 3.6 GW of greenfield data center sites in Q2 FY26, later expanding the secured power narrative to 5 GW in the Q3 FY26 business update. On customers, it highlighted advanced negotiations with hyperscalers and enterprises. On capital, it pointed to GPU financing, prepayments, convertible notes, and future data center financing. This is not a software company with a server lease. It is an industrial buildout with software-adjacent economics.

For investors, the segment takeaway is straightforward. Bitcoin mining funds and validates the asset base, but AI cloud is the segment that can re-rate the stock. If AI contracts convert into recognized revenue on schedule, the business mix will look very different over the next 12 to 24 months. If deployment slips, the market will remember very quickly that mining remains cyclical and capital hungry.

Flagship Product Analysis

IREN’s flagship product is not a consumer-facing application. It is AI cloud capacity delivered through owned or controlled data centers, powered by NVIDIA GPU clusters, and sold primarily to hyperscalers and large enterprises. Management made clear that the bulk of demand is for bare-metal GPU access rather than a heavily differentiated software layer. That matters because the product is really a package of power, cooling, hardware availability, uptime, and delivery speed.

Prince George is the clearest operating example. On the Q2 FY26 call, Kent Draper said the data center fit-outs for air-cooled NVIDIA B200 and B300 GPUs were complete and awaiting delivery of the remaining GPUs on order. He also said around $0.4B of contracted ARR at Prince George was already largely operational, with additional negotiations underway for remaining capacity. In the Q3 FY26 update, management said revenue from the Microsoft contract would begin progressively over the year, while a decent proportion of the Prince George contracted revenue was already operational.

The Childress project is the other flagship asset because it underpins the $9.7B Microsoft contract. Management said construction across Horizons 1 through 4 was progressing on schedule in Q2 FY26, and the Q3 FY26 update said 2026 expansion to 480 MW remained on track with Horizon 1 through 4 expected for delivery by year-end 2026. That gives investors a concrete milestone set: energization, GPU deployment, contract commencement, and revenue recognition.

The product proposition is strongest where customers care most about speed and certainty. Management repeatedly emphasized that hyperscalers want partners with secured power, deliverable data center infrastructure, and the ability to scale over time. In a market where many operators advertise capacity that is still stuck in permitting, interconnection, or financing limbo, IREN is selling something more tangible: capacity that can actually be turned on.

The limitation is that this product is capital intensive and less differentiated on software. Kent Draper said software is more useful for smaller enterprise customers and that it is an area likely to commoditize quickly. That is refreshingly blunt. IREN’s flagship product wins on infrastructure, not on pretending to out-code hyperscalers. For investors, that means valuation should follow deployment execution and contract monetization, not dreams of a surprise software multiple.

Innovation & Competitive Advantage

IREN’s moat is an infrastructure moat. Management’s case rests on vertical integration, secured power, internal development capability, and capital access. On the Q2 FY26 call, the company said it had secured more than 4.5 GW of power, 810 MW of operating data centers, and billions of dollars of AI customer contracts. In the Q3 FY26 business update, that power figure expanded to 5 GW secured power. Those are the raw ingredients that matter in AI infrastructure.

Vertical integration is central. Kent Draper said IREN designs, builds, and operates its own data centers with in-house engineering, procurement, construction, technology, and operations teams. That gives the company direct control over cost, timelines, and service quality. In a market where lead times and skilled labor can derail projects, internal execution is not a nice-to-have. It is the difference between a signed contract and a delayed one.

Power access is the second pillar. The company added a new 1.6 GW site in Oklahoma, taking secured power above 4.5 GW in Q2 FY26. Management said the site offers low-latency connectivity, jurisdictional diversity beyond ERCOT, and access to the Southwest Power Pool with low-cost power and renewable penetration. In AI infrastructure, power is the fuel, the permit, and the bottleneck all at once. Firms without it are often just renting optimism.

Capital access is the third pillar. The company secured underwriting commitments for $3.6B of GPU financing at less than 6%, and management said that, combined with customer prepayments, covered approximately 95% of GPU-related capex supporting the Microsoft contract. By January 2026, management said cash stood at $2.8B, and the Q3 FY26 update cited $2.6B cash at April 30, 2026. That funding base gives IREN more room than many peers to keep building without immediately leaning on dilutive equity.

There is also a credibility advantage from customer and partner validation. The Microsoft contract is the obvious anchor. The Q3 FY26 business update also cited a newly signed 5-year, $3.4B AI Cloud contract with NVIDIA and a broader strategic partnership with NVIDIA. Business context notes IREN as a Preferred NVIDIA Cloud Partner and highlights Blackwell-era deployments. Those relationships do not remove execution risk, but they do separate IREN from operators still pitching themselves as future beneficiaries of AI demand without equivalent counterparties.

Operations & Supply Chain

IREN’s operations revolve around converting secured power into functioning compute capacity. The company’s site portfolio includes Prince George, Mackenzie, Canal Flats, Childress, Sweetwater, and the new Oklahoma campus. Management said Prince George fit-outs for air-cooled NVIDIA B200 and B300 GPUs were complete in Q2 FY26, Mackenzie and Canal Flats were being prepared for AI cloud expansion, Childress Horizons 1 through 4 were progressing on schedule, and Sweetwater had procurement activities and civil works underway for its first phase.

The operating model is to remove ASIC mining hardware and replace it with GPUs where returns justify it. Kent Draper described the playbook plainly: 'ASICs are coming out of those data centers and GPUs are going in.' That is operationally efficient because it reuses existing infrastructure and power access, but it also creates near-term revenue disruption because mining revenue falls before AI billing fully ramps. Investors should expect lumpy reported numbers during that handoff.

Supply chain execution matters most around GPUs, cooling systems, and power infrastructure. Management said many industry constraints, including long lead-time procurement and skilled labor, were areas IREN had already addressed and were not disrupting execution. The company also highlighted air-cooled deployments as an advantage because they can be rolled out faster, and hyperscalers were increasingly focused on air-cooled GPUs for quicker deployment timelines. That aligns with IREN’s 810 MW of already operational air-cooled data centers.

Sweetwater is a useful test case for operational risk. In Q2 FY26, management said Sweetwater 1 was on track to energize in Q2 and that the full bulk substation was capable of 1.4 GW of power capacity. It also said Sweetwater 1 and 2 were likely to be included in ERCOT batch 0, which would mean the full 2 GW of power is secured. That does not eliminate grid and construction risk, but it does show the company is dealing with concrete milestones rather than vague future options.

The financing side of operations is just as important as the physical side. Management said financial year-to-date it had secured $9.2B from customer prepayments, convertible notes, GPU leasing arrangements, and dedicated GPU financing for the Microsoft contract. That breadth matters because AI infrastructure projects can die from capital mismatches even when demand is strong. IREN’s challenge now is not proving access to capital. It is proving that capital can be deployed into revenue-producing assets on schedule and at acceptable returns.

Market Analysis

IREN sits at the intersection of two large and very different markets: Bitcoin mining and AI infrastructure. The mining side remains cyclical and tied to Bitcoin price, network hashrate, and post-halving economics. The AI side is benefiting from a structural shortage of delivered compute capacity. Management’s framing in Q3 FY26 was direct: the world is structurally short compute, and the bottleneck is delivered data center and GPU capacity. That is the market backdrop supporting IREN’s pivot.

The demand evidence in IREN’s own numbers is strong. The company moved from $16.4M of AI Cloud Services revenue in FY25 to reporting $2.3B of ARR under contract in Q2 FY26 and $3.1B ARR under contract in Q3 FY26, with a target of $3.7B by year-end CY26. That gap between current revenue and contracted run-rate is exactly why the stock attracts growth investors. The market is valuing what the business can become if those contracts convert into recognized revenue and margins.

The AI infrastructure market also rewards operators with real power access and near-term delivery. Industry context identifies peers such as CoreWeave, Applied Digital, TeraWulf, Cipher Mining, Equinix, and Digital Realty for different slices of the opportunity. IREN’s edge is that it comes from the mining world with large power assets already in place, while many traditional data center operators move more slowly and many miners lack equivalent AI customer traction.

On the Bitcoin side, the market remains useful but less strategic. FY25 Bitcoin mining revenue was $484.6M, up from $184.1M in FY24, showing how meaningful the segment still is. But management has made clear that AI cloud offers materially better dollars per megawatt than colocation and a better value capture than pure mining. In effect, Bitcoin mining is becoming the bridge business, not the destination.

News sentiment adds another favorable signal. Across 71 data points, sentiment was 0.7043 over 7 days, 0.6706 over 30 days, and 0.7498 over 90 days, with the interpretation listed as strongly positive. Sentiment is not a substitute for cash flow, but in a story stock with heavy execution milestones, positive coverage can help sustain valuation while the income statement catches up.

Customer Profile

IREN’s customer base is shifting sharply up-market. Management said the bulk of current AI demand comes from hyperscalers, the largest enterprises, and highly advanced AI technology firms seeking bare-metal GPU access. That is an important detail because it implies customers with substantial budgets, technical sophistication, and long-duration compute needs. It also means customer concentration risk is real.

Microsoft is the anchor customer. The company’s $9.7B AI contract with Microsoft includes a 20% prepayment and is expected to contribute about $1.94B of annualized run-rate revenue once fully commissioned, according to business context. Management also said the related GPU financing and customer prepayments covered approximately 95% of GPU-related capex for Horizons 1 through 4. That kind of structure does more than validate demand. It lowers financing risk and improves project bankability.

NVIDIA is both a strategic partner and, based on the Q3 FY26 business update, a contract counterparty through a 5-year, $3.4B AI Cloud contract. That adds another marquee name to the customer and ecosystem list. For a company in transition, blue-chip counterparties matter because they reduce the market’s fear that revenue projections are built on fragile startup demand.

Customer preferences also shape the product mix. Kent Draper said most large customers want full control of the GPUs and their own software stack, which is why bare-metal access dominates. Smaller enterprise customers may care more about user interfaces and software layers, but management said that remains a small proportion of compute demand today. That means IREN’s near-term revenue is tied more to infrastructure delivery and less to software monetization.

The customer risk is concentration. A few large hyperscalers can transform revenue, but they also gain bargaining power and can create lumpiness in deployment schedules. For now, the evidence favors confidence over fear because the company has multiple advanced negotiations underway and reported strong customer demand. Still, investors should treat customer diversification as a medium-term requirement, not a cosmetic nice-to-have.

Competitive Landscape

IREN competes in two overlapping arenas. Against Bitcoin miners such as Riot Platforms(RIOT), CleanSpark(CLSK), Core Scientific(CORZ), and Hut 8(HUT), it competes on power access, hashrate scale, and operating efficiency. Against AI infrastructure players such as CoreWeave, Applied Digital(APLD), TeraWulf(WULF), Cipher Mining(CIFR), Equinix(EQIX), and Digital Realty(DLR), it competes on deliverable capacity, GPU access, contract wins, and time to revenue.

Its strongest relative position is versus miners trying to pivot into AI without equivalent customer traction. IREN has a signed Microsoft agreement, a reported NVIDIA contract, more than 4.5 GW of secured power in Q2 FY26, and 5 GW secured power in the Q3 FY26 update. That gives it a more credible AI transition than many peers whose AI strategy still reads like a future tense exercise.

Its weaker relative position is versus mature digital infrastructure operators with established enterprise ecosystems and steadier financial profiles. IREN’s revenue base is still mixed, earnings are volatile, and the business remains exposed to Bitcoin economics during the transition. Traditional data center operators may offer lower upside, but they also tend to offer lower drama. The market often pays up for lower drama for good reason.

Management argues that AI cloud generates materially better dollars per megawatt than colocation. Daniel Roberts said every incremental 200 MW can deliver roughly $300M through colocation or multiples of that under a cloud contract. If that proves durable, IREN’s competitive edge is not just owning power, but choosing the highest-value use for that power. That is a smart framing. In this business, megawatts are inventory.

Peer valuation data was not available from the peer screen, so the competitive valuation comparison must rely on IREN’s own multiples and Street targets rather than a clean peer median. Even with that limitation, the strategic position is clear: IREN is one of the more credible AI-transition names among former or current miners, but it has not yet earned the stability premium of established infrastructure incumbents.

Macro & Geopolitical Landscape

IREN’s macro exposure is unusually layered. First, it is exposed to Bitcoin price and mining difficulty. Management said Q2 FY26 revenue fell partly because of lower average Bitcoin prices and increasing global hashrate. Business context also notes that net electricity cost per Bitcoin mined rose to $25,642 in FY25 from $18,127 in FY24, mainly due to the April 2024 halving and higher global hashrate. That is the mining side of the macro picture: commodity-like revenue, rising network competition, and sensitivity to power economics.

Second, it is exposed to AI infrastructure demand and enterprise capex cycles. Here the macro is more favorable. Management said customer demand remained very strong, with multiple advanced negotiations underway for larger-scale deployments. The broader AI buildout by hyperscalers supports that view, and IREN’s own contract wins provide company-specific evidence that spending is real, not theoretical.

Third, the company is exposed to power market regulation and grid dynamics. ERCOT batch processing came up directly on the Q2 FY26 call. Management said Sweetwater 1 and 2 were likely to be included in batch 0, supporting the view that the full 2 GW of power is secured. It also argued that stricter processing could actually help by separating real megawatts from imaginary ones. Dry comment, but fair. In power-constrained markets, regulatory friction can punish weaker developers and reward those with genuine interconnection progress.

Geographically, IREN benefits from diversification across Australia, Canada, Texas, and now Oklahoma. The Oklahoma site adds exposure to the Southwest Power Pool rather than just ERCOT, which management described as attractive for low-cost power, renewables penetration, and hyperscaler interest. Jurisdictional diversity matters because local grid rules, permitting, and transmission constraints can change the economics of a site quickly.

Interest rates also matter because this is a capex-heavy business. The good news is that IREN secured GPU financing at less than 6%, and management said the effective rate was around 3% when factoring in prepayments. That is a meaningful advantage in a market where expensive capital can crush returns. The bad news is that any deterioration in credit conditions or project economics would hit a company like IREN harder than an asset-light operator.

Balance Sheet Health

Income Statement Strength

Estimates Outlook

Valuation Assessment

Target Prices & Recommendation

Closing

IREN(IREN) is one of the more credible AI transition stories in the former Bitcoin mining universe. The company has moved beyond vague ambition and assembled the pieces that matter: 5 GW secured power in the latest business update, major counterparties, substantial financing, and a contract base that is large enough to reshape the income statement if execution holds. Those are serious advantages.

The market, however, is not giving those advantages away for free. Current valuation already reflects meaningful optimism, and the reported financials remain volatile because the company is still shifting capacity from mining to AI cloud. That makes IREN a stock where operating milestones matter more than slogans. Energization dates, GPU deployments, contract commencements, and ARR conversion into reported revenue will decide the next leg.

For medium-term investors, the conclusion is balanced but positive. IREN has enough hard assets and signed business to justify a Buy rating, and enough execution risk to demand valuation discipline. The company is building something real in a market short on real capacity. That is a strong place to start. It is not the same thing as a guaranteed outcome, but it is far better than a story built on borrowed buzzwords and rented servers.