Friday, September 11, 2026
AI 인프라 · 뉴스 & 분석
컴퓨트·클라우드리포트
컴퓨트·클라우드 · 리포트

IREN의 AI 가격 급등으로 차세대 클라우드의 26억 달러 계약 규모가 어떻게 크게 확대될 수 있을지 의문이 제기되고 있다.

아이렌의 급격한 가격 상승은 GPU 클라우드 공급 긴장을 시사하며, 뉴클라우드 섹터 전반의 공격적인 용량 확장 전략을 입증한다.
업계 전문지Slicast · September 8, 2026 · 미국 · 출처: 24/7 Wall St.
중요도 88

A single pricing metric buried in IREN’s latest earnings report suggests the company’s existing contract book may be significantly undervalued relative to current customer willingness to pay.

This publication may contain affiliate links; Flywheel Publishing may receive compensation for actions taken through them.

The pivotal figure in IREN’s (NASDAQ: IREN) Q4 FY26 report lies outside the income statement: approximately $25 million per megawatt of IT load. CEO Daniel Roberts noted this pricing is appearing “consistently across live conversations with customers at the moment.” Representing the run rate for three- to five-year AI cloud contracts currently under negotiation, this benchmark effectively reprices the entire portfolio IREN has already developed.

Tracking the trajectory clarifies the shift. IREN’s initial five-year agreements with Microsoft (NASDAQ: MSFT) and NVIDIA (NASDAQ: NVDA) were valued at $9.70 million per MW (November 2025) and $11.33 million per MW (May 2026), respectively. More recent three-year contracts executed in August 2026 exceeded $20 million per megawatt of IT load, with current negotiations centering near $25 million. Management reported that three-year contract pricing has increased by approximately 125% since November, while five-year pricing has risen roughly 70%.

Shorter contract durations are commanding premium pricing, primarily due to constrained compute capacity. Customers are also directly underwriting infrastructure expansion: recent advance payments are covering 45% to 55% of GPU capital expenditures, yielding an estimated two-year payback period on compute investments. Financially, IREN reported $1 billion in operating annual recurring revenue (ARR) as of August 26, 2026, with $4 billion in contracted ARR slated to become operational by December 31, 2026. Sequentially, the Q4 AI Cloud Services segment more than doubled to $70.5 million, while full-year AI Cloud revenue expanded approximately eightfold to $128.8 million.

The market responded swiftly. At the time of the Form 8-K filing on August 27, 2026, shares closed at $40.9469. Over the subsequent trading week, the stock advanced from $35.45 on August 28 to $44.68 on September 4, marking a 26.04% gain. Broader performance metrics show a one-month return of 14.89%, a year-to-date gain of 18.29%, and a one-year return of 70.99%. The company’s market capitalization now stands near $17.6 billion.

IREN’s combined annualized contract value with Microsoft and NVIDIA totals approximately $2.62 billion—a relatively modest footprint compared to peer AI infrastructure developers. (For context, we have profiled seven foundational infrastructure providers spanning power, cooling, and networking in a separate industry report.) This dynamic creates a compelling setup: IREN is already securing top-tier per-megawatt pricing while operating a contract book with substantial expansion potential. Crucially, the underlying physical assets required to scale are already secured.

The $4 billion in contracted ARR derives from less than 10% of IREN’s secured grid connection portfolio, which exceeds five gigawatts. Delivery milestones project approximately 0.3 GW of IT load in 2026 and 0.8 GW in 2027, positioning the platform at 1.2 GW of gross capacity by year-end 2027. Capital requirements are substantially covered: IREN holds $14 billion in committed GPU financing and customer prepayments, including $6.5 billion raised over the three months prior to the earnings call, alongside $5.8956 billion in cash on the balance sheet.

The partnership with NVIDIA extends beyond hardware procurement. IREN earned NVIDIA Exemplar Cloud certification for its GB300 NVL72 deployment serving Microsoft. Furthermore, its $3.4 billion five-year NVIDIA AI Cloud agreement is accompanied by up to $2.1 billion in NVIDIA equity investments, structured to vest as IREN scales toward 600,000 GPUs. Customer acquisition continues to accelerate. The fourth-quarter roster added Cohere, Prometheus, Perplexity, Figure AI, Fal AI, and Higgsfield AI, alongside a new multi-year agreement with a leading frontier AI laboratory. Incumbent clients Together AI and Fireworks AI both renewed and expanded their commitments. CEO Daniel Roberts summarized the momentum: “As our platform has scaled and our market position has strengthened, we have attracted leading customers and secured stronger pricing, more attractive contract terms and improved paybacks.”

For long-term investors, the central question remains whether IREN can successfully convert its secured power capacity into contracted revenue at these elevated pricing tiers. The immediate catalyst is the $4 billion contracted ARR target scheduled to go live by December 31, 2026, contingent on the rollout of Horizons 2 through 4. Management indicated that revenue recognition from this capacity will materialize primarily in the March fiscal quarter.

Governance presents a notable caveat: restricted stock unit awards granted in mid-2026 to co-CEOs Dan and Will Roberts, valued between $800 million and $1.1 billion with minimal performance conditions, sparked shareholder scrutiny. Ultimately, investors are gaining premium per-megawatt pricing and a multi-gigawatt growth runway, balanced against a founder-aligned leadership structure. Both dynamics must be weighed within the same investment thesis.

For inquiries or corrections, contact [email protected].

원문 보기