@danielisdizzy Toda la razón. Ese planteamiento es el correcto se puede comprobar que muchos países renuncian a tener centros de datos cerca. Problemas con la energía, la construcción a corto plazo, la tierra, normativas y legalidades. Todo eso favorecerá a Iren
$NVDA x $IREN
With the way AI & Compute are panning out, I think it’s highly possible that IREN will become Nvidia’s largest and most important partner in the future 🔥🔥🔥
https://t.co/pmTYweZ4lV
Horizon 1: delivered.
IREN has delivered Horizon 1 to Microsoft and achieved NVIDIA Exemplar Cloud status on NVIDIA GB300 NVL72.
Read more: https://t.co/ItVXSXK1V7
JUST IN: Horizon 1 is complete.
And the confirmation didn't come from $IREN.
It came from the people who actually built it.
Southwest Electrical Contracting Services, the electrical contractor on site, just held an employee appreciation dinner celebrating the completion of Horizon 1 in Childress, Texas.
Custom aprons and all.
The badge reads: "Horizon 1 Completion Celebration. Childress, TX. 7.16.26. Powering Progress Together".
Independent, ground-level, and dated. This is what a finished build looks like from the crew that poured the concrete.
Read the comments and it gets better. One of their team wrote it was "so amazing to actually see what our team does to build these data centers".
Another reply, and I love this one: "Thank you from a shareholder".
The contractors and the shareholders, celebrating in the same comment section.
Now line up the timeline, because this is not an isolated dot.
Days ago, Frans posted satellite imagery: the grounds paved, Microsoft's own security booth installed. Move-in details, not construction details.
RAISE attendees reported the handover date as July 19.
The contractor just celebrated completion on July 16.
And yesterday, IREN signed $2.8bn in new contracts and raised ARR past $4bn.
Every independent thread points at the same thing. Horizon 1, the first of four identical buildings, is done.
Not "on track".
Not "imminent".
Done, per the people holding the wrenches.
Remember what this building actually is.
Unit one of four.
The prototype that proves the repeatable model.
Once it hands over to Microsoft and starts billing, the other three stop being a question of "can they" and become a question of "when".
The miner-to-neocloud story had exactly one real risk: can they deliver a hyperscaler-grade facility, to a hyperscaler's standard, on time.
The crew just threw a party because the answer is yes.
This is not financial advice. Do your own research.
I'm long $IREN.
The $IREN bears’ only strong point was “massive amounts of dilution”. Today’s PR showed that companies are willing to pay up front for compute. The 45% prepayment shows they will not need to raise so much cash. With the delivery of Horizon 1, it opens the door to new financing options. The CAPEX needed to build out the 5GWs will come from prepayments, their existing cash, revenue from existing contracts, and financing. The dilution will be minimized. The bear case is fizzling out.
@chinoalemano Mejor que le gente venda. Así yo compro más barato. Lo fácil que es hoy en día meterle toda la información a Gemini o chat gpt y que te haga un enfoque de lo que pasa con Iren. Le gente no ve más allá de la noticia momentánea. Se mueven como pollos al pienso.
$IREN granting RSUs to Dan and Will Roberts isn’t a cash bonus – it’s a long-term commitment.
Shares vest over 4 years.
The vested shares cannot be sold for another 2 years.
Final restrictions extend into FY2033.
No additional equity grants until FY2031.
This means the founders are effectively locked in for the long haul and only win if shareholders win. They are betting on the future value of IREN’s AI infrastructure business and remain fully aligned with investors.
Long-term alignment > short-term compensation.👌👌👌
The biggest catalyst nobody has priced in:
$IREN is going to be the next AWS.
And it isn't doing it alone.
It's doing it with $NVDA and $DELL standing behind it.
Let me explain why, because once you see the structure, you can't unsee it.
NVIDIA needs IREN to survive.
Not as a nice-to-have.
As a matter of strategy.
Think about NVIDIA's position.
Its three biggest customers, Microsoft, Google and Amazon, are all building their own chips to cut it out.
Maia.
TPU.
Trainium.
Every one of them is trying to escape.
So NVIDIA needs a generation of clouds that will never build their own silicon, that stay pure NVIDIA, and that grow big enough to be a real alternative to the hyperscalers.
That's IREN.
That's why NVIDIA is funding it, feeding it allocation, and building it into a giant.
If the loyal neoclouds fail, NVIDIA is left with nothing but the customers trying to replace it.
So NVIDIA will defend IREN to the death, because IREN is part of how NVIDIA stays the center of the entire ecosystem.
Now look at what IREN has stacked in a single year.
The density is staggering:
✓ NVIDIA Exemplar Cloud status, certified on the B300, NVIDIA's newest architecture. One of the few clouds on earth to earn it.
✓ Sweetwater 1, 1.4GW, energized on schedule. Disciplined execution on the scarcest input in AI: power.
✓ A direct $3.4B cloud contract with NVIDIA. The only software deal a neocloud has signed with a Mag7 company.
✓ NVIDIA's option to invest up to $2.1B in IREN, vesting as GPU infrastructure deploys. NVIDIA literally profits more the more IREN builds.
✓ Sweetwater chosen as the flagship deployment for NVIDIA's DSX AI factory architecture. Of all the neoclouds, IREN's site is the blueprint.
✓ A $3.4B to $3.7B ARR trajectory as the fleet scales toward 150,000 GPUs.
✓ The Mirantis acquisition, NVIDIA's own strategic partner, for the orchestration layer.
✓ European expansion through the Nostrum acquisition, adding 490MW.
✓ An 800MW data center announced in Australia. A 1.6GW campus announced in Oklahoma.
✓ A ~5GW secured power portfolio, owned, not leased, putting IREN among the largest in the entire space.
✓ A strategic partnership with NVIDIA to deploy up to 5GW of AI infrastructure.
✓ And a global pipeline still being disclosed.
Read that list again. Any single one of those would move a stock. IREN stacked all of them in a year, and the market is still pricing it like a former Bitcoin miner.
Now add the third leg.
Dell.
IREN signed a $1.6B deal with Dell for Blackwell systems.
So you have the full stack assembling: NVIDIA bringing the silicon and the ecosystem,
Dell bringing the systems and the hardware, and IREN bringing the power, the land, and the operation.
That's not a data center company.
That's the three layers of a hyperscaler being built in real time, by the companies that want to take AWS's place.
NVIDIA needs it to exist.
Dell is building it out.
And IREN owns the scarcest asset in the entire equation: the power.
The catalysts are real.
Most aren't priced in.
And the market is asleep.
You're not betting on a neocloud.
You're betting on the next AWS, with the most powerful company in tech making sure it gets built.
This is not financial advice. Do your own research.
I'm long IREN.
@danroberts0101 Justo en el momento de tweet por parte de @nvidia y junto con @AnthropicAI saca @danroberts0101 este tweet.
Entiendo que están testeando los centros de datos para dar el paso más grande de la historia. Esperaremos el mega contrato tomando una copa
NVIDIA independently validated our B300 training performance across their benchmark suite, a status only a few achieve. Design, networking, software and operations tuned as one system. Owning the data centers, not renting them, helps keep cloud performance repeatable at scale.
Esta semana ha pasado algo que, bien leído, es oro para la tesis de $IREN : Google ha tenido que alquilar GPUs a un competidor. SpaceX, que se quedó con la antigua xAI, le va a cobrar a Google 920 millones de dólares al mes desde octubre de 2026 hasta junio de 2029 por unas 110.000 GPUs de Nvidia. Cerca de 30.000 millones en total. Si lo bajas a lo que de verdad importa, sale a unos 11,5 dólares por GPU y hora. Y conviene parar aquí un segundo: Google, una de las empresas que más centros de datos construye del planeta, está pagando una millonada por alquilarle GPUs a un rival. ¿Por qué? Porque ni siquiera ella construye lo bastante rápido.
Y esa es, justo, la mesa en la que se sienta $IREN . Cuando hasta Google tiene que llamar a la puerta de un tercero y soltar 11,5 dólares la hora, el mercado te está diciendo una cosa muy clara: están desesperados. Y quien tiene megavatios, manda. $IREN tiene más de 5 gigavatios de potencia asegurada entre Texas, Oklahoma, Canadá y España. ¿Cuánto ha vendido? Solamente un 4%.
Y ahora, permíteme un poco de remember. Hace unos meses, mientras X se llenaba de subnormales diciendo "WEN DEAL, WEN DEAL" pidiendo un contrato a gritos, yo decía justo lo contrario: que cada día que pasaba sin acuerdo, brindaba. No por llevar la contraria. Por aritmética. Que firmar no era el problema, ojo. Vender unos pocos megavatios baratos a Microsoft fue la jugada maestra: con un 4% de su capacidad, IREN se ganó a un hiperescala de primer nivel, demostró que sabe construir y entregar a esa escala, y desbloqueó financiación a grado de inversión usando ese contrato como aval. Esos 2 USD por GPU y hora no fueron debilidad. Fueron el cebo barato que valida el modelo, demuestra la ejecución y paga todo lo demás.
Lo que no tenía ningún sentido era lo que pedía la grada: que $IREN siguiera malvendiendo el resto a la carrera. Porque el tiempo jugaba a su favor por partida doble. Por un lado, el precio del cómputo subía mes a mes, y eso lo reconoce hasta la competencia directa, por escrito, en sus informes a la SEC. Por otro, cada mes de espera era un mes más de potencia energizada, de ejecución demostrada y de volverse más imprescindible. Esperar no era estar parado de brazos cruzados. Era acumular precio y poder a la vez.
Y aquí está la regla que la grada nunca entendió: en una crisis de suministro como esta, el que tiene el recurso escaso pone las normas. $IREN ha dejado de ser el que busca que los gigantes lo validen, de hecho NVDIA lo ha elegido para llevar a la práctica sus últimos desarrollos. $IREN ahora es el que tiene la energía que esos gigantes necesitan para no quedarse atrás. Se acabó firmar con prisa y con descuento. De aquí en adelante, o el comprador acepta el precio, el plazo y las condiciones que marca $IREN, o se levanta de la mesa con las manos vacías. El que antes tenía que rogar un contrato es ahora el que elige cliente, plazo y tarifa. Ahora el precio lo pone $IREN.
$IREN is seeing record institutional inflows.
The smart money keeps piling in.
Institutions are aggressively adding to existing positions and opening new ones.
Triple digits no longer seem like a question of if — but when.
$IREN Does the stock have further to run ?
This chart answers that question for you, the answer is yes! ✅
Every bull run confirmed by my TD indicator hasn't peaked out until the weekly RSI has hit +70...
In May the stock flipped green on the TD indicator confirming the bull run, whilst the RSI remains at 64 which tells me that this rally has further to go 📈
$IREN Closes $3.65B Investment-Grade GPU Financing
This marks the highest publicly rated investment-grade GPU financing announced
• Blended cost of debt of 6%
• $2.10B U.S. private placement
• $1.55B delayed draw term loan
Goldman Sachs & J.P. Morgan as joint lead managers
$IREN's >4000% AI growth in the coming 15 months with 5 new deals announced in the coming 6 months. The path to >$12B ARR and >$100 share price by 2027 👇
I believe the upcoming 6-9 months will be highly compelling for Iren, and I expect the stock to significantly outperform the broader market. This outlook is driven by numerous positive catalysts poised to push the share price upward, which I will outline below:
1. Exceptional AI Revenue Growth (>4000% over the next 15 months)
Over the next five quarters, Iren is projected to achieve triple-digit quarter-over-quarter (QoQ) growth, scaling revenue from $34 million to over $1.4 billion within this timeframe. According to calculations by @bitcoinbutcher1 , Iren is on track to reach an annual Recurring Revenue (ARR) exceeding $12 billion by the end of 2027. (check his post)
2. Anticipated Contract Announcements (5 deals in the next 6 months)
Over the next three months, I anticipate Iren will announce several contracts related to their 2027 capacity. Iren currently has an ARR of $4.4 billion, and over the next six months, I expect management to increase contracted ARR from the current $3.1 billion (as of May 29, 2026) to over $10 billion. This growth will be driven by multiple agreements utilizing their 2027 capacity. Below is an overview of these expected agreements:
Mackenzie (80MW):
Iren has purchased approximately 36,000 B300 NVIDIA GPUs, scheduled for delivery in the second half of 2026. This data center will begin generating cash flow for Iren within the year. While the GPUs have been procured, a formal customer announcement for this facility has not yet been made. Market speculation suggests these resources may be allocated to existing clients such as Fireworks AI or togetherAI. However, I am optimistic that Anthropic could secure this capacity, given that Mackenzie is likely one of the data centers offering the fastest time-to-compute. Industry observers are keenly aware that Anthropic is in urgent need of immediate capacity. I expect an agreement for Mackenzie to materialize in the coming months, which would subsequently boost the projected EOY 2026 ARR from $3.7 billion to $4 billion.
Canal Flats (30MW):
As Iren's smallest data center, Canal Flats is set to be retrofitted from Bitcoin mining to air-cooled AI capacity. Consequently, an announcement from Investor Relations regarding additional B300 purchases for this site is highly probable in the near term. This facility is expected to be operational by the first half of 2027, contributing an estimated ARR of over $300 million.
Horizon 5-6 (Liquid-Cooled, VR200):
In their Q1 report, Iren announced plans to build an additional 100MW of IT load utilizing liquid-cooled data centers. I suspect this capacity will be contracted to Microsoft, which is already a client for Horizon 1-4. The key differentiator is that Horizon 5 and 6 will be designed for NVIDIA's latest Vera Rubin models. Estimating the ARR contribution is difficult, as comparable contracts are not currently present in the market, but I anticipate this contract will be secured shortly. With Horizon 1-4 slated for completion by EOY 2026, it is logical to commence groundwork and expansion soon—a process significantly aided by having a committed client.
Childress 250MW (Air-Cooled, Retrofitted):
Iren also announced an additional 250MW of air-cooled capacity for Childress. This involves retrofitting existing Bitcoin data centers into air-cooled AI facilities. The strategic decision to prioritize air cooling is driven by the ability to offer faster time-to-compute, a critical focal point in today's market (again, highly relevant to companies like Anthropic). We already know that 60MW of this capacity is contracted to NVIDIA, leaving 190MW currently available. This remaining 190MW could contribute an additional $2.1 billion in ARR (calculated as $700 million ARR from the NVIDIA contract multiplied by 3).
Sweetwater (300MW to 1.4GW):
This facility has been designated as NVIDIA's "flagship deployment for NVIDIA's DSX architecture." An initial 300MW is scheduled for development in 2027, scaling up to a massive 1400MW data center over time. The immediate 300MW phase currently requires a committed contract. Iren has already completed significant groundwork for the site and secured grid connection approval in early May. This contract will involve NVIDIA's VR200 GPUs, likely making it the largest data center globally with a VR200 installation. This is the contract I am most anticipating due to its sheer scale, potentially exceeding $20 billion. While it is a long-shot, there is a possibility that the entire 1400MW could be leased to a single client. Such a scenario would imply a potential contract size of over $100 billion—a staggering figure considering Iren's current market capitalization of $22 billion.
3. Strategic Rerating: From Bitcoin Miner to Neocloud:
Despite being a strong proponent of Bitcoin myself, its current association does more harm than good for the company's valuation. Bitcoin is generally viewed unfavorably by Wall Street, resulting in the stock trading at significantly lower multiples compared to pure-play AI companies. As Iren transitions away from Bitcoin over the next 6-9 months, I expect the stock to undergo a significant rerating as the perceived risk associated with cryptocurrency is eliminated. This pivot will serve as a major positive catalyst; the company will shed its label as a mining operation and be formally recognized as a premier data center provider.
$IREN: What to Monitor in 2026
Revenue
No doubt $IREN is rich in power, what $IREN investors need to focus on is revenue. The demand is certainly there, it's about getting GPUs online. Even the rate at which $IREN can sign contract is bottleneck by getting GPUs online because getting GPUs online derisk your ability to meet timelines to sign the next contract.
So what you see in $CRWV is that at the beginning they were really fast and then started slowing down. At first, a Neocloud will 2x revenue every quarter but then the ramp are bounded by the physical world. The benefit of having alot of power is that you'll be able to keep growing at a high rate until you ran out of power or your colocation provider(s) hits their limits. However, that's not $IREN's challenge. $IREN needs to get it's ramp -> cashflow -> ramp feedback loop going. Getting slow GPUs slowed down the whole ramp as you are doing theory based preparation for your datacenter until you get the GPUs and then problems can be uncovered sequentially.
Benchmarks
2024 was $CRWV's ramp year:
Q4 2023: 116m
Q1: 188.7m (+62.7% QoQ)
Q2: 395.4m (+109.5% QoQ)
Q3: 583.9m (+47.7% QoQ)
Q4: 747m (+27.9% QoQ)
Q1 2025: 981m (+31.4% QoQ)
2025 was $NBIS ramp year:
Q4 2024: 37.9m
Q1: 50.9m (+34.3% QoQ)
Q2: 105.1m (+106.5% QoQ)
Q3: 146.1m (+39% QoQ)
Q4: 227.7m (+55.9% QoQ)
Q1 2026: 399m (+75.2% QoQ)
What Will Move the Stock
To look what will move Neocloud stocks, I admit that $NBIS has done a fantastic job this year. What $NBIS executed well objectively was:
1. Sign a 12B Meta Contract with +15B Extension Option
2. Back up their capability to fulfill the contract by hitting revenue numbers and critically showing acceleration in revenue growth. Observe how $NBIS Q1 2026 earnings show an acceleration to 75.2% revenue growth.
I monitor the whole industry to figure out what's going on and for $NBIS, I got to give credit where credit is due, $NBIS put up the GPUs and in this market it doesn't matter if you pay colocation or whatever, getting the GPUs up and showing revenue growth is what the market wants to see from early stage Neoclouds.
IREN's Revenue
IREN's ramp was suppose to start in Q4 2025 but really it's Q1 2026 because we couldn't get GPUs delivered on time due to HBM shortgage which snowballed the whole ramp process back.
Using currently delivery guidance, here are my calculations for the next few quarters revenue:
Q1: 33.6m
Q2: 100.8m (200% QoQ)
Q3: 207.6m (106% QoQ)
Q4: 385.4m (85.6% QoQ)
Q1: 843.8m (118.9% QoQ)
Q2: 1454.7m (72.4% QoQ)
Q1 Calculations: Reported in Q1 earnings
Q2 Calculations: PG exited Q1 with 307m run rate (page 19 of 10Q in source 1 - also screenshotted 1st picture) from the financial digging that @_Sgr_A_Star did and should be at 500m run rate by end of quarter. With linear ramp, (307+500/2)/4-qtr = 100m.
Q3 Calculations: 125m from PG and H1 Handoff stated by Dan to be in Q3 which I will take to be July. I'll take August + Sept of 124m from H1 so 124m * 2/3 = 82.6m.
Q4 Calculations: 125m from PG; 124m from H1; assuming we get Mackenzie handed half way through the quarter = 432.8/4qtr/2-halfway = 54.1m; assuming we get 2/3 duration of H2 and 1/3 duration of H3 and H4 get's handed over at the very end of the quarter, we only count the 2/3 H2+ 1/3 H3 = 124m. Sum = 427.1m
Q1 2027 Calculations: 125m from PG, 485m from H1-4, 108.2m from Mackenzie, CF = 40.6m, half duration of Nvidia Childress site = 85m.
Q2 2027 Calculations: 125m from PG, 485m from H1-4, 108.2m from Mackenzie, 40.6m from CF, 170m from Nvidia Childress, 1/2 duration of Block 7-9 is 330.9m, SW1 50MW IT is 195m
Share Price
IREN's ramp started late but having the power supply abundantly clear, means that the ramp can sustain high % growth for longer because power is not the bottleneck.
If IREN can have report 385.4m quarterly revenue for Q4 2026 it will mirror NBIS 399m quarter where most of its revenue was either H100/H200s and bare metal to MSFT with colocation payments so margin are similar. With an SW1 contract in hand, it would match the Meta 12B with potential 15B extension contract NBIS has. In this case, market is giving NBIS an 55B valuation.
Let's give 5B for NBIS 25% Clickhouse stake at 20B next round valuation even though at 15B valuation now. The rest of NBIS's subsidiary + the power they secured is rough equal to value of IREN's power portfolio (I know must IREN investors wouldn't make this trade off let's just call this even to make comparisons easy, you can do whatever adjustments you want). IREN Q4 2026 report will be which would be 50B (current NBIS market cap) / 22.89B (current IREN MCap) * 64.07 (current IREN stock price) = $139.95/share. Q4 earnings is early Feb 2027 but IREN also has higher sustain revenue growth rates due to it's power abundance but since Q4 earnings is Feb 2027, let's take 20% of for the time delta between EOY 2026 and earnings report for Q4 2026 to have a target of $111.96 share price for EOY 2026.
The really strong year for IREN will be 2027 as it sustains high growth rate and not be stuck in early ramp pains.
Sources
(1) https://t.co/Cd3HLFXzzY