The market just repriced IREN twice for the same event, and only one of the reprices was about IREN. The stock closed July 30 at $38.26 — down 50% from its November high of $76.41 — after printing $29.31 on July 29, the day Leopold Aschenbrenner’s Situational Awareness fund was margin-called out of its entire book, which included a $401 million IREN position. Twenty-four hours later, with that book absorbed by Citadel, IREN closed 30% above its low. That single-day round trip is the cleanest natural experiment you will get in separating what July did to IREN’s price from what it did to IREN’s business — and the answer frames this IREN stock prediction: $75 base, $100 bull, $29 bear.
The insight competing coverage misses: IREN’s July had two sellers with nothing in common. One was thesis-driven — the July 17 Meta Compute announcement raised a real question about hyperscalers competing with GPU clouds. The other was purely mechanical — a 4x-levered fund liquidating $401 million of stock into a falling tape because its prime brokers said so. The Bernstein desk that carries the $100 target was writing about the first seller; the July 29 low was made by the second. Buying decisions made without separating the two are pricing a margin call as if it were a business model.
Key facts
- IREN closed at $38.26 on July 30, 2026 — down 11.7% in July and 50% below its 12-month high of $76.41 — after a July 29 low of $29.31, per Yahoo Finance daily data
- Consensus 12-month target: $75.18 across 12 analysts (one Sell), with Bernstein at $100 — implying roughly +96% and +161% from spot, per Cryptonomist’s target roundup
- Management raised its end-2026 annualised AI cloud revenue target to more than $4 billion (from $3.7 billion), with roughly 85% underpinned by signed contracts including Microsoft and Nvidia deals, per TechTimes, which counts a record $2.8 billion AI contract haul
- Situational Awareness held $401 million of IREN per its May 18 13F; the position was sold to Citadel in the July 30 pre-open block, per BitMEX Research and CNBC reporting
- The July 17 Meta Compute announcement — Meta selling external AI compute — is the live competitive question; Bernstein also flags IREN’s enterprise-client concentration beyond Microsoft
- Next catalyst: September earnings
What actually happened in July
Three overlapping shocks. First, sector gravity: the AI-infrastructure complex de-rated after July 17, when Meta announced Meta Compute and the market began pricing hyperscalers as competitors rather than pure customers of the GPU-cloud build-out. Second, the crowding unwind: IREN sat inside the most famous concentrated book in markets — Situational Awareness owned $401 million of it alongside $8.5 billion in related AI-infrastructure longs — and when Goldman Sachs, JPMorgan and Bank of America margin-called the fund, IREN’s tape absorbed days of involuntary selling that peaked in the July 29 flush to $29.31. Third, macro: the July 29 FOMC held rates 9–3 with three dissents demanding a hike, the exact backdrop in which levered growth stories find no marginal buyer.
Then the seller finished. Citadel took the block before the July 30 open, and IREN — the most liquid proxy for the unwind — closed that session at $38.26, up 30% from the low. SpotGamma’s post-mortem of the episode is the operative frame: “forced liquidations of leveraged, crowded books have historically clustered near local extremes.”
The stretch of tape between July 24 and July 29 deserves its own line, because it shows how information asymmetry works in a modern margin spiral. On July 24, Aschenbrenner told investors his fund had “not been immune” to the turbulence while inviting fresh capital — language that, read against a public 13F, told every prop desk on the street exactly which names a wounded, 4x-levered whale might soon be forced to sell. IREN, as one of the book’s most liquid converted-miner positions, traded with unexplained heaviness for days before the block: the sellers’ identity was a secret, but the seller’s shopping list never was. That is why the July 29 low belongs in a different analytical category from an ordinary 52-week low — it was made by a countdown, not a valuation.
How IREN got here: from Sydney hydro miner to Microsoft’s grid partner
The company the margin call mispriced spent five years building toward exactly this moment. IREN listed on Nasdaq in late 2021 as Iris Energy, a bitcoin miner differentiated by renewable power — hydro-heavy sites in British Columbia and a vast land-and-power position in Childress, Texas. The asset that mattered was never the mining fleet; it was the energised acreage and grid interconnection agreements, which is why the 2024–2025 pivot to AI data centres worked here while stranding weaker miners: the scarce input in AI infrastructure is not GPUs but powered shells to put them in. That pivot took the stock from a 52-week low of $15.40 last August to $76.41 by early November 2025 — and even after July’s carnage, IREN enters August up roughly 148% year-over-year, a detail the drawdown headlines uniformly omit.
The 2026 chop between roughly $32 and $68 reflects the market repeatedly re-asking one question: is IREN a contracted infrastructure business or leveraged crypto beta? Each answer has moved the stock 40%: the Microsoft cloud agreements and the Nvidia relationship argued infrastructure; the bitcoin drawdown and the miner cohort’s beta argued crypto. July’s forced selling settled nothing about that question — which is precisely why the September print, the first with the raised $4 billion run-rate target on the table, is the real referendum.
The Meta question, taken seriously
The July 17 announcement deserves more than a shrug, because it is the only genuinely new fundamental fact in the stock this month. If Meta sells external compute at scale, the bear logic runs, then every hyperscaler eventually rents out surplus capacity, GPU-cloud pricing compresses, and the independents — CoreWeave, Nebius, IREN’s AI cloud arm — get squeezed between their customers and their suppliers. The bull rejoinder has three parts. First, capacity reality: industry GPU capacity remains effectively sold out through 2027, and a new seller entering a shortage validates the price rather than breaking it. Second, counterparty logic: enterprises diversifying away from hyperscaler lock-in are not going to solve that problem by renting from Meta — neutrality is the independents’ product. Third, IREN specifically sells power-secured capacity under multi-year contracts, not spot GPU-hours; 85% contract coverage is insulation precisely against the pricing war the bears fear. The honest netting: Meta Compute trims the long-term multiple every independent deserves, and changes next year’s contracted revenue not at all.
The business the seller left behind
None of July’s flow news touched the operating story. IREN’s pivot from bitcoin mining to AI data centres has produced a record $2.8 billion in cumulative AI contracts, and management responded to the crash month by raising guidance: annualised AI cloud revenue is now targeted above $4 billion by end-2026, up from $3.7 billion, with roughly 85% of that covered by signed contracts — anchored by Microsoft, with Nvidia relationships on both the supply and customer side. Contracted revenue is the entire distinction between an AI-infrastructure story and an AI-infrastructure hope: at 85% coverage, next year’s revenue is mostly a delivery problem, not a demand problem.
The bear’s rejoinder is concentration, and it is fair. Bernstein’s own work flags that IREN’s enterprise pipeline beyond Microsoft is thinner than rivals’, and the Meta Compute question lands hardest on exactly this point: if the largest capex spenders begin renting out their own surplus compute, the customer pool for independent GPU clouds shrinks from the top. That is a real multi-year risk — it is also a risk shared by every name in the complex, and it does not compound at 4x leverage the way July’s price action implied.
The numbers: price vs targets
| Reference | Level | vs $38.26 close |
|---|---|---|
| July 29 liquidation low (bear line) | $29.31 | −23% |
| 2026 open | $42.70 | +12% |
| Consensus 12-month target (12 analysts) | $75.18 | +96% |
| 12-month high (Nov 5, 2025) | $76.41 | +100% |
| Bernstein target (bull line) | $100 | +161% |
Sources: Yahoo Finance daily closes to July 30, 2026; analyst targets per Cryptonomist and TechTimes, July 2026.
The macro overlay: rates, power and the regulatory seam
Two structural currents run under this chart. The hawkish Federal Reserve — three dissents wanted a hike at the July meeting — is a direct multiple compressor for pre-profit infrastructure growth, and a September hike would pressure every long-duration AI story’s discount rate; that is the single most important non-company variable in the bear path to $29. Pulling the other way is the physical bottleneck: US grid interconnection queues now run years, which converts IREN’s energised Texas and British Columbia acreage into a moat no amount of hyperscaler capex can quickly replicate. Regulators are becoming the sector’s gatekeepers — data-centre power allocation is turning into a state-level political issue — and companies holding signed interconnection agreements are on the right side of that seam. A business whose scarce asset is permitted electricity is, unusually for this market, hedged against its own hype cycle.
The FinanceFeeds call: $75 base, $100 bull, $29 bear
Base case $75 by mid-2027 — the analyst consensus, which effectively prices the contracted $4 billion revenue path executing while the Meta question stays unresolved. Bull case $100 — Bernstein’s number — requires two things: September earnings confirming the AI cloud ramp on schedule, and at least one new named enterprise customer that answers the concentration critique; in that world IREN re-rates as contracted infrastructure rather than crypto-adjacent beta. Bear case $29 — the July 29 liquidation low — is the invalidation line: that price has now been tested by the most violent involuntary selling imaginable and held for exactly one day, so a fundamental return to it (a weekly close below $29) would mean the market is pricing contract slippage or a Meta-driven demand break, not flow. Between here and September, the reference points are mechanical: above $42.70 (the 2026 open) the recovery is complete; below $34, the bounce has failed and the tape is telling you Citadel is still distributing.
What would change this call
Four triggers. A September earnings miss on AI cloud revenue or contract coverage — the 85% figure is the load-bearing number in the whole thesis. A second hyperscaler following Meta into external compute sales, which would convert a single-competitor worry into an industry structure change. A weekly close below $29, per above. And on the upside, evidence that the Citadel inventory has been absorbed — several weeks of IREN outperforming the miner-converts complex would be the tell that the technical overhang is gone, the pattern this desk is also tracking in Bloom Energy, the unwound fund’s largest long.
A final word on how to read the next month’s tape, because the ordinary rules are suspended when a block this size changes hands. Citadel did not buy $401 million of IREN to hold it as a conviction position; it bought a distressed book at a negotiated price and will distribute it as liquidity allows. That means rallies can be capped by supply for weeks without any fundamental meaning, and it means volume tells you more than price: heavy-volume flat days are absorption, light-volume drift is apathy, and a high-volume push through the $42.70 level would be the market saying the inventory is gone. Watch the tape’s texture, not just its direction.
FAQ
Why did IREN stock crash in July 2026?
Two reasons stacked: the July 17 Meta Compute announcement de-rated all GPU-cloud names, and IREN was a $401 million position in Situational Awareness’s 4x-levered book, which was forced into liquidation — IREN printed its $29.31 low on July 29, margin-call day, and rebounded 30% once Citadel bought the book.
What is the IREN stock price prediction for 2026?
Our call: $75 base (the 12-analyst consensus at $75.18), $100 bull (Bernstein’s target, on a clean September quarter plus new enterprise names), $29 bear (the liquidation low as invalidation). Spot reference: $38.26 at the July 30 close.
Is IREN a buy after the Aschenbrenner unwind?
The forced seller is gone and the business raised guidance into the crash — $4 billion targeted annualised AI cloud revenue, 85% contracted. The open risks are Meta-driven competition and Microsoft concentration. Every published analyst target sits far above spot; the July 29 low defines the downside line. This is analysis, not advice.
How much IREN did Aschenbrenner’s fund own?
The Q1 2026 13F disclosed $401 million of IREN — one of several converted-miner positions alongside Core Scientific ($389m) and Applied Digital ($320m) — all sold to Citadel in the July 30 pre-open block trade.
What is the next catalyst for IREN?
September earnings, where the market will test the raised $4 billion AI cloud run-rate target and its 85% contract coverage. Before that, watch the tape against $34 and $42.70 for evidence the post-liquidation recovery is holding.
This article is informational analysis only and is not financial or investment advice. Equities are volatile and can lose substantial value rapidly. Past performance and analyst targets do not guarantee outcomes. Do your own research and consult a regulated financial adviser before making any investment decision.