Pinned Tweet
$IREN Breakout in progress?
The rounding bottom structure on the weekly chart is nearly complete after over 2 years of consolidation.
- Key zone: $16.80–$18.50
- Rising volume
- Precise rejection at $18.54 this week
- Long-term bullish pattern close to triggering
Technical target: $36
Parabolic (log scale) target at $335 👀
Pullback to $14.50–$15 could be a tactical opportunity.
Unai retweeted
Agree IREN is the next hyperscaler, @Joseph_Invests.
A challenge for the smart $IREN bulls:
IREN has >5GW of secured power and a >35GW pipeline, and it isn't slowing down. On Thursday's call Dan was asked about NVIDIA's modular DSX PODs. His answer: IREN has had teams on modular and prefab construction for nine months, is working with NVIDIA on designs, and is still deciding whether to run the DSX POD exactly as presented.
The goal he named: simpler, faster builds with less labour on site.
My read: that's the path to factory-built PODs that deploy in weeks and are quickly upgradeable at any time.
Remember who they bought in 2020: PodTech. The name wasn't an accident. See photo from 2019.
While most of the space faces headwinds, I think IREN has huge tailwinds. What else does this unlock? Add your tailwinds below.
(Disclosure: large IREN shareholder, PodTech co-founder.)
$IREN WILL BE THE NEXT HYPERSCALER, HERE'S WHY:
35 GW Acquired
20 M/MW Deals Long Term
35 GW at 20 M/MW Equals 700 Billion ARR
700 Billion X 4 P/S Is a 2.8 Trillion Market Cap
Yes I know this is crazy, yes I know this is a long term out but I am illustrating that $IREN could very well be the next hyper scaler
Unai retweeted
Software engineering is becoming a solved problem. The only bottleneck left is compute.
You can build, clone, or outperform any software. Just throw enough compute at it.
Unai retweeted
🚨 $IREN's >35 GW Pipeline
They FINALLY did it.... I can't believe $IREN finally revealed the size of their massive pipeline.
For years investors were begging management to reveal the size of their power pipeline.
It only took the stock to crash 58% YTD for them to reveal this very relevant piece of info.
But honestly, wow. No other neocloud, or colo provider for that matter, even comes close to 35 GW...
Unai retweeted
🚨 $IREN: $22m / MW on a 5-year contract including 90% prepayment of GPU capex.
This is MASSIVE, wow... the demand for compute is unreal.
Unai retweeted
AMD CEO Lisa Su on this AI moment we are seeing in markets:
"I think the market can sometimes be a little bit short-sighted. It is what it is. What I look at is in the long term, if you believe, which I do, I believe very deeply that computing, in many ways, equates to intelligence," Su added.
"So the more compute you have, the more intelligent your company is, the more problems you're going to be able to solve, the more you're going to accelerate your business. If you believe that foundation, then investing ahead of the curve is the right thing to do because we're always investing for the long term — long term being three to five years and beyond. And that means you have to make the investments now. So I think from our standpoint, I've always operated that way. We're always looking at what do we see over the horizon? And over the horizon, I see AI and super intelligence becoming more and more essential to every aspect of our lives and every aspect of our business. So I believe in this demand, and I very much believe that many of the investments that are being made throughout the ecosystem are also seeing some of these same trends."
Rewatch again MFs
Horizon 1: delivered.
IREN has delivered Horizon 1 to Microsoft and achieved NVIDIA Exemplar Cloud status on NVIDIA GB300 NVL72.
Read more: iren.gcs-web.com/static-file…
Unai retweeted
Hyperscalers: Builders Before Buybacks
In an instant‑gratification world, investors have grown used to technology giants that turn every spare dollar into buybacks and neatly managed earnings. Free cash flow was recycled, capex stayed lean, and multiple expansion did the heavy lifting. That pattern is breaking. The hyperscalers are now in a different business: pouring cash into the physical infrastructure of the AI age.
The numbers tell you why their shares have sagged. AI infrastructure spending is running at hundreds of billions a year and still accelerating. Capital intensity will not peak until around 2027–28, while truly material AI revenue only arrives on the same timetable. In the interim, buybacks disappear, depreciation climbs and reported margins look tired. Charts suggest the group has bottomed; they do not yet suggest it has healed.
Yet it is a category error to treat the clouds as ex‑growth value traps destined for single‑digit earnings multiples. They are not over‑capitalised telecoms. They are token factories. Each new data centre converts capex and power into tokens, the unit of billable output in an inference‑heavy AI economy, where value comes from answering, predicting and generating rather than just storing or routing data. As token volumes rise and token pricing strengthens, the return on capital is there. The problem, for now, is that fixed costs are so large and the build‑out so aggressive that they still dwarf near‑term revenue forecasts, leaving the equity story looking blurred in the short run. Markets still model AI as if it merely automates physical labour, when the real asset being built is “cognitive capital”: systems that think, infer and create at scale.
The real issue is timing. Through 2026 and well into 2027, investors should expect a grind rather than a snap‑back: persistent capex, noisy AI revenues and little scope for old‑style cash returns. The inflection is a 2028 story. As capex growth finally slows, installed AI capacity is high and utilisation improves, free cash flow will turn. Financial engineering will return, this time layered on top of genuine AI‑driven growth rather than used as a substitute for it.
For day traders, that is an awkward horizon. For long‑term investors, it is the point. At today’s levels, hyperscalers are increasingly priced as capital‑intensive builders, not as perpetual buyback machines. If you can live without aggressive repurchases until the free‑cash‑flow turn, buying here should be considered.
The healing may be slow, but the franchise is not broken; impatience, not insolvency, is the real risk.
Unai retweeted
Databricks CEO Ali Ghodsi says if frontier models stopped advancing, it wouldn't matter for most orgs because enterprise AI is bottlenecked on context, not intelligence:
"There's no agentic work that's automated the whole enterprise. That has just not happened. Why is that? The real reason is the models are smart enough, but they just don't have the context that exists inside of any organization."
"They have not been in every meeting. They don't know what's in everybody's heads. They don't know all the processes. There's always a couple of employees who know everything in every organization. You go tap on their shoulder, and everybody's like, 'Oh my God, what would happen if he or she quits?'"
"If you just fused that and gave that context into the AI models, just the frontier today, I think there's so much productivity gains you could get for any organization on the planet."
"For that, we actually don't need smarter models... If the frontier doesn't advance, it doesn't actually matter for the vast majority of organizations on the planet. They're just so far behind in the adoption curve of actually automating things and getting value out of this stuff."
@alighodsi
Read the latest Profiles of the Future by @bhorowitz and @mikemcg0 on Databricks & the Future of the Firm: a16z.news/p/databricks-and-t…
"They're not going down" @danroberts0101.
$IREN & $NVDA DSX. Strategically, it makes sense for IREN and Nvidia to pursue a financial collaboration or JV on the first part of SW1.
$IREN
I’ve been thinking about Sweetwater and I believe the first phase (or at least a big chunk of it) will be developed and financed more closely with $NVDA than the market is currently pricing in.
The “up to 5GW” partnership isn’t a firm offtake, that’s clear. But Sweetwater is explicitly positioned as the flagship site for NVIDIA’s DSX architecture. NVIDIA already has a cloud contract with IREN, a warrant for up to $2.1B at $70, and a clear interest in having large-scale liquid-cooled capacity in ERCOT as a reference.
For $IREN, the CapEx for SW1 comes at a tough time: they’re dealing with the Horizons at Childress, liquid-cooling retrofits in Canada, and the 2027 ramp all at once. A large block of Sweetwater with NVIDIA as an anchor (whether through preferential offtake, stronger prepayments, or some form of JV/joint financing) would make sense for both sides. NVIDIA gets priority capacity and a real DSX showcase, while IREN unlocks part of the data center CapEx and reduces dilution or delay risk.
At the same time, I see IREN focused on actually building IREN Cloud, especially for smaller developers (AI-native startups and enterprise customers that aren’t hyperscalers). The Mirantis acquisition and management’s comments about diversifying the customer base point in that direction. The big offtakes (Microsoft, and potentially NVIDIA at Sweetwater) give them the scale and funding to build the infrastructure, but the longer-term goal seems to be having their own platform that captures more value with a broader mix of customers.
I’m not saying NVIDIA takes 100% of the campus, but the first meaningful phase of Sweetwater makes more sense as a joint project than as just another standard contract.
Thoughts?
Unai retweeted
Also, as management has emphasized several times, it is not primarily the cloud business that creates the need for dilution. It is the data center segment.
As I've said a number of times now, the cloud business is capital intensive in absolute total dollars, but ironically very capital light from IREN’s wallet perspective. Very little to no money is coming out of pocket for the Cloud business segment (GPUs).
For Microsoft, customer prepayments and GPU financing covered 96% of the GPU capex. For the recent AI developer contracts, the same combo provided funding equal to roughly 133% of GPU capex!
In other words, IREN is contributing very little out of pocket capital toward the GPUs themselves and in the most recent contracts the Cloud segment is actually generating SURPLUS capital!
The data center side is different.
Those costs have historically been funded primarily from the company’s own balance sheet, including plenty of dilution as we all know. No one knows better than you that IREN must acquire land, substations, install electrical infrastructure and construct the data centers long before customer prepayments and GPU financing are even thought of.
IREN's set up the flywheel. Now it's about execution.
Unai retweeted
What Is $IREN Good At?
A friend who supports NBIS told me：
Iren is not a data center engineering specialist at all. IT data centers and centers for miners are very far from each other and that is exactly why Iren decided to use Nvidia's DSX platform, it gives you the design and since they are not good at IT/AI data centers they pay DELL to do some of that work too. Iren are good at finding land, building the halls and bringing the power shells on, which is closer to a construction company than to data center engineering company, so in reality it is a relationship of dependence for Iren.
This was my response. Does anyone have anything to add?
By the end of 2026, IREN’s Horizon 1–4 will be the world’s largest newly built, native liquid-cooled AI data center deployment designed for NVIDIA’s high-end GPUs, featuring direct-to-chip liquid cooling. By this standard, the number of data centers of comparable scale owned by NBIS at the end of 2026 is: 0.
The Horizon campuses are still only a small part of IREN’s GW-scale campuses, leaving enormous room for expansion. By this standard, NBIS will have 0 comparable facilities before 2029, and whether it can achieve this even after that remains uncertain.
IREN will also have multiple GW-scale campuses spread across different locations, interconnected to enable cross-campus connectivity. By this standard, NBIS’s comparable capacity in 2030 or even 2035 is: 0.
Multiple sites that IREN has announced, as well as others yet to be announced, are all GW-scale. This alone is something the vast majority of future competitors will be unable to match. The GPU clusters to be deployed will operate on a completely different physical infrastructure scale. Achieving this requires collaboration with top-tier experts such as Dell and NVIDIA; developing and operating such infrastructure independently is simply not feasible. The complexity of ultra-large-scale clusters increases exponentially with scale. By this standard, NBIS’s comparable capacity is: 0.
All of IREN’s sites have been carefully selected with strategic positioning and community acceptance in mind. These are scarce assets that cannot simply be replicated once the land is occupied. They also take green energy and environmental sustainability into account. By this standard, NBIS’s comparable assets are: virtually 0.
All of IREN’s sites are grid-connected, enabling future interaction with the power grid. This will significantly improve economics and stability while opening up multiple additional revenue streams. By this standard, before 2029, NBIS’s comparable capabilities are: 0.
In the future, IREN will also enable coordination across multiple sites and develop additional satellite sites to accommodate the specific demands of the inference era. By this standard, NBIS’s comparable capabilities are: 0.
Simply accumulating land and pre-purchasing power is already no easy feat. There is so much foresight about the future embedded in these decisions that they seem almost as if they were made by people who had traveled back from the future.
IREN could very well become the world’s leading blueprint for replicable next-generation liquid-cooled AI data centers—the printing press for AI factories.
This is also why so many leading data center experts have chosen IREN, and why an increasing number of outstanding professionals from Oracle, NVIDIA, Google, KKR, and Intel continue to join the company.
$IREN
I’ve been thinking about Sweetwater and I believe the first phase (or at least a big chunk of it) will be developed and financed more closely with $NVDA than the market is currently pricing in.
The “up to 5GW” partnership isn’t a firm offtake, that’s clear. But Sweetwater is explicitly positioned as the flagship site for NVIDIA’s DSX architecture. NVIDIA already has a cloud contract with IREN, a warrant for up to $2.1B at $70, and a clear interest in having large-scale liquid-cooled capacity in ERCOT as a reference.
For $IREN, the CapEx for SW1 comes at a tough time: they’re dealing with the Horizons at Childress, liquid-cooling retrofits in Canada, and the 2027 ramp all at once. A large block of Sweetwater with NVIDIA as an anchor (whether through preferential offtake, stronger prepayments, or some form of JV/joint financing) would make sense for both sides. NVIDIA gets priority capacity and a real DSX showcase, while IREN unlocks part of the data center CapEx and reduces dilution or delay risk.
At the same time, I see IREN focused on actually building IREN Cloud, especially for smaller developers (AI-native startups and enterprise customers that aren’t hyperscalers). The Mirantis acquisition and management’s comments about diversifying the customer base point in that direction. The big offtakes (Microsoft, and potentially NVIDIA at Sweetwater) give them the scale and funding to build the infrastructure, but the longer-term goal seems to be having their own platform that captures more value with a broader mix of customers.
I’m not saying NVIDIA takes 100% of the campus, but the first meaningful phase of Sweetwater makes more sense as a joint project than as just another standard contract.
Thoughts?