Mentioned in This Article
1. What did IREN report for its first fiscal quarter?
IREN, the former Bitcoin miner that now earns most of its money from AI computing, reported total revenue of $228 million for the quarter ended September 30, 2026, up 66% from the previous quarter. The results, published for fiscal first quarter 2027, sent the stock up about 20% on Friday, November 6, to close near $21.64, even though the net loss widened to $263 million.
The headline number was the AI cloud business, which generated $200 million in revenue. That compares with $71 million in the final quarter of fiscal 2026, an increase of 182% in three months. The company said revenue beat analyst expectations, which helps explain why investors focused on growth rather than on the loss.
The quarter also confirmed how far the company has moved from its origins. Revenue from Bitcoin mining fell to $28 million from $67 million in the prior quarter, a drop that management describes as deliberate, because power and data center capacity are being redirected from mining to AI workloads.
2. How fast is the AI cloud business growing?
The AI cloud segment grew almost threefold from one quarter to the next. A segment that earned $71 million in the June quarter reached $200 million in the September quarter, and it did so at an AI cloud gross margin of 86%. Gross margin measures the share of revenue left after the direct costs of delivering a service, so a figure that high indicates the hardware is being rented out at prices well above what it costs to run.
IREN credited the jump to expanded contracts with major customers. These include a frontier AI lab, meaning one of the companies that develop the largest AI models, and a new five-year prepaid agreement for GPU capacity. GPUs, the graphics processors originally built for gaming, are the main chips used to train and run AI systems, and customers who prepay for several years of access give the provider a predictable stream of income.
The company's revenue backlog, which counts contracted work not yet delivered, stood at $20.3 billion, up 22% from the previous quarter. Operating annualized recurring revenue, or ARR, doubled to $1 billion. ARR takes the current run rate of contracted services and multiplies it over a year, giving investors a quick read on the size of the business at its current pace.
The growth also helps explain why the contract structure matters. When customers commit to multi-year terms and pay in advance, the provider can plan construction, financing and equipment orders with far more certainty than a business that sells capacity month to month. That predictability is one reason the backlog, rather than any single quarter's revenue, is the figure analysts watch most closely for companies in this sector.
3. Why did the net loss grow?
Despite the revenue surge, IREN recorded a net loss of $263 million, which was wider than analysts had projected. The quarter therefore combined strong top-line numbers with a bottom line that disappointed, a pattern common among companies that are spending heavily to build capacity before the full revenue from that capacity arrives.
A net loss that exceeds forecasts would normally weigh on a stock. The market reaction suggests investors gave more weight to the scale of the backlog, the doubling of operating ARR and the margin on AI services than to the loss itself. The figures show a company whose income is becoming larger and more concentrated in a single line of business, which tends to make revenue growth the metric that matters most to shareholders.
It is also worth separating a loss that stems from timing from one that stems from weak demand. The figures reported here point toward the former: revenue and backlog rose sharply, while the shortfall came from the cost of building and running the infrastructure that will serve those contracts. Whether that reading holds will depend on how quickly the new capacity begins producing revenue in the coming quarters.
4. What is the delay to the $4 billion ARR target?
IREN has set a goal of more than $4 billion in annualized recurring revenue, and it now expects to reach that level in early 2027, later than it had planned. The company attributes the slip to two causes: a design change requested by Microsoft, and delays in the delivery of equipment to its Childress, Texas facility.
Management says the target remains fully contracted and fully funded. In practical terms, that means the revenue is covered by signed agreements and the money to build the required infrastructure is already secured, so the delay concerns timing rather than demand. For investors, the distinction matters: a postponed start date pushes revenue into later quarters, but it does not remove the contracts behind it.
The episode also shows how exposed a builder of AI data centers is to the preferences of its largest customers. A single change in design requested by one client was enough to move a schedule by months, and equipment arriving late can have a similar effect even when the customer is satisfied with the plan.
For readers unfamiliar with the metric, the reason ARR draws so much attention is that it translates a young, fast-changing business into a single comparable number. A move from $1 billion in operating ARR toward more than $4 billion would quadruple the run rate, which is why even a delay of a few months changes how analysts model the next several quarters.
5. Where does the Microsoft contract stand at Childress?
The Childress site in Texas anchors a five-year cloud services contract with Microsoft valued at $9.7 billion. During the quarter, the first Microsoft data hall at the site came online, running NVIDIA GB300 systems, which are among the newest generation of AI server platforms from NVIDIA.
A data hall is a large, self-contained room within a data center that houses racks of servers along with the power and cooling they need. Bringing the first one into service is a milestone because it converts a construction project into a revenue-producing asset, and it allows the remaining halls to follow on the same template.
The Microsoft relationship is therefore both the largest source of confidence and the largest source of exposure. The contract provides a long-dated income stream from a customer with substantial resources, while the design change that delayed the ARR goal came from the same customer.
NVIDIA's GB300 platform is relevant because customers of AI cloud services typically want the latest generation of hardware, and the company's ability to deploy it at scale is part of what makes its capacity attractive. Operators that can install new systems quickly and keep them running reliably are better placed to win and keep large contracts.
6. How much has the business mix changed?
The balance between mining and AI has reversed within a year. In the last quarter of fiscal 2026, Bitcoin mining brought in $67 million and AI cloud services $71 million, so the two were nearly even. In the latest quarter, AI cloud revenue of $200 million is roughly seven times mining revenue of $28 million.
Bitcoin miners and AI data center operators share the same raw ingredients: large amounts of electricity, land, cooling and grid connections. That overlap is why a number of mining companies have repurposed their sites for AI customers, who generally sign longer contracts and pay for guaranteed capacity rather than depending on the volatile price of a cryptocurrency.
IREN's own figures illustrate the trade-off. Mining revenue more than halved in a single quarter, but the AI revenue that replaced it was far larger and carried a published gross margin of 86%. The company has said it intends to keep shifting capacity toward AI, so mining revenue may continue to shrink.
Investors have watched this pattern across the mining sector, where companies with access to cheap, large-scale power have found that the same sites can earn more from AI customers than from producing coins. IREN's quarter is among the clearest examples so far of that shift showing up in reported revenue.
7. What risks remain for IREN shareholders?
The main risk is customer concentration. A large share of the contracted revenue depends on Microsoft and a small group of other major clients, so the loss, delay or renegotiation of any one agreement would have an outsized effect on results. The Childress delay is a concrete example of how one customer's request can alter the timeline.
A second risk is execution. Building data halls, securing equipment and bringing power online all have to proceed on schedule for the backlog to turn into revenue. If further setbacks occur, the early 2027 ARR target could slip again, which would test the patience of investors who rewarded the stock on Friday.
A third consideration is the size of the losses. Revenue grew quickly, but the net loss of $263 million shows that the cost of building the business is still high. The share price reaction implies the market expects that gap to narrow as more capacity goes live.
None of these risks was dismissed by the market on Friday so much as weighed against the scale of the opportunity. The stock's 20% gain reflects a judgment that the growth in AI cloud revenue, the size of the backlog and the funding behind the $4 billion goal outweigh a wider loss and a later date.
8. What should investors watch next?
Three developments will shape the next phase. The first is the activation of additional data halls at Childress, which will show whether the first hall's successful launch can be repeated at pace. The second is any update on the early 2027 timeline for the $4 billion ARR goal, including whether the company holds that date.
The third is the trajectory of mining revenue. If it keeps falling after the drop to $28 million, the company will be almost entirely an AI infrastructure provider, and its results will track demand for AI computing rather than the Bitcoin price. IREN's next quarterly report will provide the first full read on all three.
Readers tracking the story should also note that the company has described the target as fully contracted and fully funded, so any future change to that description would be an important signal. Until then, the main questions are about the pace of delivery rather than the existence of customers.