IREN은 6억 8,400만 달러의 손실을 보고하며 레거시 광산 시설을 AI 데이터센터로 전환하는 과정에서 막대한 자본 소모와 운영상 마찰이 발생했음을 강조했습니다.

IREN recently reported a milestone that should have propelled its stock higher: AI cloud revenue surpassed bitcoin mining for the first time in the company’s history. Instead, shares fell 12%. The $684 million loss accompanying this milestone explains the market reaction—and underscores a critical reality about the broader bitcoin mining industry’s rapid pivot toward artificial intelligence.

For fiscal Q4, ending June 30, 2026, IREN reported the following results: AI cloud revenue reached $70.5 million, up 110% quarter-over-quarter, while bitcoin mining revenue fell 40% to $66.7 million. Total revenue declined 5% to $137.2 million, and adjusted EBITDA dropped 68% to $19.2 million. The company posted a net loss of $684 million, driven largely by $450.4 million in non-cash impairments. On a full-year basis, revenue grew 41% to $707 million compared to fiscal 2025. However, the company swung from an $86.9 million profit to a $702.6 million loss. This is not a typical earnings miss; it reflects the financial cost of dismantling one business model while building another in its place.

The $450.4 million impairment charge warrants close attention. It represents the written-down value of bitcoin mining hardware being decommissioned—specifically ASICs (application-specific integrated circuits), which are specialized chips designed exclusively for mining. These units cannot be repurposed for AI workloads and must be scrapped. IREN has now quantified the exact cost of that transition.

IREN’s AI cloud revenue more than doubled in a single quarter, rising from $33.6 million to $70.5 million—a genuinely impressive expansion. For the first time, AI services accounted for 51.4% of total revenue, surpassing bitcoin mining at 48.6%. Yet Wall Street does not price milestones; it prices margins. Adjusted EBITDA, a standard proxy for operating profitability before accounting adjustments, fell 68% to $19.2 million, down from $59.5 million in the prior quarter. Margin compression stems from two primary factors. First, IREN is rapidly scaling headcount to support its AI cloud division, hiring costly data center engineers, cloud architects, and sales personnel. Second, the company is investing heavily in infrastructure that has not yet begun generating revenue. This creates the “valley of death” inherent to every mining-to-AI pivot: capital flows into AI operations before meaningful returns materialize, while legacy mining revenue declines as rigs are shut down. Consequently, the financial statements appear distressed precisely when the strategic transition is advancing.

IREN has secured $4 billion in annualized recurring revenue (ARR) from AI cloud contracts. The largest agreement is a $9.7 billion, five-year partnership with Microsoft, supported by $3.6 billion in GPU financing at a 6% weighted average interest rate. A secondary five-year agreement with Nvidia adds $3.4 billion. These are binding contracts with blue-chip counterparties, meaning the primary risk is not demand but execution. As of August 26, only $1 billion of the $4 billion ARR is operational. The remaining $3 billion hinges on completing data center construction, deploying GPUs, and delivering compute capacity to clients on schedule. Co-founder Daniel Roberts noted, “Exponential AI consumption growth has fueled demand for compute capacity well beyond available infrastructure supply.” While his assessment of demand is accurate, the critical challenge remains whether IREN can scale infrastructure rapidly enough to convert contracted commitments into realized revenue before cash burn becomes unsustainable.

IREN is not alone in pivoting from bitcoin mining to AI. The broader sector is pursuing similar strategies with varying execution profiles. IREN is executing a full pivot, winding down mining operations by approximately 2030, with a $9.7 billion Microsoft contract and execution timeline as its primary risk. Core Scientific is pursuing a hybrid mining and AI strategy backed by a 1.2 GW power portfolio, though it faces construction delays. TeraWulf is forming joint ventures with AI partners, including a $9.5 billion joint venture with Fluidstack, though geographic concentration remains a concern. Bernstein analysts project that IREN will fully exit bitcoin mining by 2030, replacing all legacy hardware with GPUs. While ambitious, the financial trajectory supports this outlook: mining revenue contracted 40% in a single quarter while AI revenue doubled.

The financing architecture is equally notable. IREN has arranged $6.4 billion in GPU financing across its agreements. The Microsoft contract alone covers 96% of associated costs through favorable-rate GPU financing. This mirrors the “hyperscaler backstop” model fueling the wider AI infrastructure boom: technology giants guarantee payments, enabling former miners to secure institutional debt at rates previously unattainable. However, this structure introduces concentration risk. When a single customer dominates both revenue generation and equipment financing, contract renegotiation could destabilize the entire operation.

For crypto investors, the uncomfortable reality is that IREN’s pivot signals a loss of confidence in bitcoin mining margins. The company is not abandoning mining due to AI’s allure; it is responding to deteriorating mining economics amid surging AI infrastructure demand. With bitcoin trading volatile near $78,000, mining profitability faces pressure from rising network difficulty, post-halving revenue compression, and escalating energy costs. Firms possessing dedicated power infrastructure—which miners inherently do—can generate superior returns by leasing compute capacity to AI developers rather than selling hash power to the Bitcoin network. This represents a structural, not cyclical, realignment. IREN is the first major miner to quantify the true cost of this transition: a $684 million quarterly loss, with additional write-downs anticipated as legacy mining operations are systematically wound down.

Following the report, IREN’s shares fell 12%, trading in the $35–$36 range. The bear case is straightforward: a $684 million net loss, severe margin compression, and a $3 billion execution gap between contracted and operational revenue. The bull case is equally clear: $4 billion in committed ARR with Microsoft and Nvidia, $6.4 billion in secured financing, and a first-mover advantage in the mining-to-AI transition. If IREN realizes even 75% of its contracted revenue, the current valuation may appear undervalued. Portfolio decisions remain yours, but the mining-to-AI pivot stands as the most consequential strategic wager in crypto infrastructure since the 2021 mining cycle. IREN’s fourth-quarter results provide the first concrete data on the cost of that transition. Every competing miner evaluating a similar path is monitoring these figures closely.

**How does IREN’s loss compare to the broader AI infrastructure spending trend?**

IREN’s $684 million loss is substantial but reflects a one-time transition expense rather than chronic operational weakness. The wider AI infrastructure market is deploying capital at unprecedented scales, with OpenAI, Microsoft, and Google each committing billions annually to compute capacity. IREN’s loss functions as the entry fee for transitioning from mining into that ecosystem.

**What happens to Bitcoin’s network if major miners like IREN leave?**

Bitcoin’s mining difficulty automatically adjusts every 2,016 blocks, approximately every two weeks. When miners exit, difficulty decreases, improving profitability for remaining operators. Network security remains intact. The primary concern is hash rate concentration: fewer entities controlling a larger share of network security could eventually raise centralization risks.

**Could rising bond yields threaten IREN’s AI transition financing?**

Yes. IREN’s GPU financing carries a 6% weighted average interest rate, which correlates with broader credit market conditions. Significant increases in bond yields—plausible given the Federal Reserve’s current policy stance—could elevate refinancing expenses. However, the Microsoft revenue guarantee provides a buffer against worst-case scenarios, as the financing is directly tied to a guaranteed contractual payout.

**Is IREN’s story similar to what happened with Trump Media’s crypto pivot?**

Not substantially. Trump Media abandoned its cryptocurrency initiatives following losses and canceled agreements. IREN is executing a deliberate transition from one compute-intensive sector to another, backed by multi-billion-dollar contracts. Its losses represent expected retooling costs, not strategic failure. One scenario constitutes a retreat; the other is a managed industrial transition.

**How does the Fed’s inflation stance affect AI data center investments?**

Elevated inflation and interest rates increase borrowing costs for capital-intensive projects such as data centers. The Federal Reserve has cautioned that AI-driven expenditures could themselves exacerbate inflation through heightened energy demand and construction activity. While IREN’s locked-in financing rates offer partial insulation, the broader macroeconomic environment could delay AI infrastructure deployment across the industry.