IREN CEO on $9.7B Microsoft deal: 'This is binding, this is real revenue'

you go. Reach millions of consumers who are using AI discover new products and brands. What do you want to talk about Jordy? We got to talk with uh Dan, CEO of coming in right now. Bring them in from the massive [music] deal. Dan, how you doing the show? Hi guys. Good to see you. Thanks for having us on.

Great to see you. Massive, massive day today. We saw you announced uh a pretty big deal and so we said what better day to have you on the show. Yeah. Yeah. You know, it's a big weekend, so a lot of uh sleepdeprived iron bodies around, but it's uh it's super exciting. What? Give us the headline number.

We like to ring the gong around here when there's a big deal that gets inked. All right. Iron and Microsoft 200 megaww 9. 7 billion in revenue 2 billion. Absolutely massive. Absolutely massive. Um break down break down the year for us. We we've uh we've covered uh Iron's rise a little bit.

Uh we kind of figured you had some type of deal like this in the works, but what what catch us up to speed uh on the last sort of six months? Yeah, look, it's been super busy.

Um so AI popped up really in force probably last April and since then we've been building out our um GPU cloud business which basically involves us buying the GPUs and providing access to those via cloud rather than an alternate model which is colllocation where you build the infrastructure and you allow one of these tech giants to lease it back.

So we believe in owning the compute and providing that uh to the end customers. Um it's been a journey talking to a number of the uh the large global technology companies but really uh happy to consummate it with Microsoft.

Yeah, I remember it was earlier this year Satia had a line that something to the effect of I'm happy to be a leaser and so uh I'm sure you guys you guys uh picked up on picked up on that. Can you give us the what? Sorry. So so I want the history. Yeah.

One thing that Satia's I guess been saying over the last couple years is maybe not a couple years but over the last years he feels like he's more uh energy constrained than he is uh GPU constrained. What kind of advantages has Iron had uh on the on the power side that that made a deal like this possible?

Yeah, that's absolutely right. And I think it's almost a little bit more nuanced than just power constrained. I think it's actually data center constrained as well. But let me come back to that.

I think when my brother and I set up this business 7 years ago, our underlying thesis was that the real world can't keep pace with the digital world. So the digital world is driving all these exponential demand curves. You know, adoption goes from zero to one overnight.

Whether it's Bitcoin, nothing 15 years ago to $15 trillion today. whether it was AI uh not really spoken about 2 years ago to the latest humanity defining technology and at the center of that is these exponential growth and demand but the ability to service that is constrained and limited by the real world.

So when we started this business 7 years ago, we explained to people, it sounds simple. You can't plug a computer into a high voltage transmission line. It takes years and years of permits, approvals, dealing with riskadverse utilities, giving you access to that network.

So yes, having access to power and having put in the groundwork very early is now paying dividends. But it's actually going a step further than that where you might have access to power, but do you have the framework to go and build the data center?

Do you have the governance regime and the flexibility of a of a corporate to give you that decision to go and build and start building ahead of the curve? Do you have the internal knowhow on how to build these new generation data centers? Because fundamentally they're a different asset class.

These are not metropolitan data centers providing corporate shared drives doing cat videos like they're fundamentally different. Um how do you think about the like the the predictability about the shape of compute workloads going forward? How important is it for you to project those out?

Uh I'd love some like history of like what was the first compute workload that you were doing because I imagine it wasn't LLM inference. uh there was probably crypto mining in the business. There's all these different workloads that were super valuable at certain times then changed. Now we're in the reasoning era.

We might be in a completely different paradigm. How much do you want to bake onto AS6? How much do you want to you know use Nvidia GPUs more generalized chips versus flexibility? Like how do you think about projecting out the future?

So, I would just bring it back to a really big picture and say as society, do we believe that we're becoming more digitized? Do we believe that we're going more online and that's going to grow?

And again, during our seed investment round, we quoted movies like The Matrix, like Ready Player One, Wreckit Ralph, and directionally I think humanity is heading that way, which is going to drive this appetite for high performance compute. None of us have a crystal ball. We don't know what's going to be next.

We had AI in our seed deck. We also had a whole heap of other stuff that hasn't eventuated. Sure. But what it means for us from a business perspective is a number of smallcaled bets along the way creating optionality. To go and pay an option fee on 500 hectares of land in West Texas is a really really small exposure.

To go and pay 10 20 million to get the grid connection is now a relatively small exposure for our business. So, it's right sizing your bets to a point where you get to sign a Microsoft deal underwriting $9. 7 billion of revenue. You go, "Okay, that justifies more capex. " Yep. Yep. That makes a lot of sense.

Uh what other are are are you looking at other opportunities and sites in in other areas or is Texas the focus going forward?

Look, Texas was the focus early on because we did the lap of the world, worked out the best place to get power, low cost, excess renewables, lots of land, uh, cheap power, u ease of grid connections, lots of fiber backbones, and and now the world's there. Um, we do have additional sites.

We've publicly announced 3 gaw, but we don't talk about development sites that aren't 100% secured by virtue of a connection agreement. So there is a multi- gigawatt pipeline behind that um outside Texas and globally. Yeah.

How how do you think about the like the current playbook for companies that are doing partnerships with hyperscalers or open AAI? Like what are the dos and don'ts of announcing a big partnership?

Uh because we're hearing so many of them and I think a lot of people are having we're in somewhat of a press release economy right now. Yeah. People are having trouble handicapping them. Like you can just throw out a big number. You could you could date it a hundred years in the future and get really crazy numbers.

Like what does the market actually want to get out? This deal feels significant in some ways because Satia has been so much more conservative than other uh let's say like an oracle is. Totally. Totally. Totally. Yeah. Look, I think it's good to be skeptical, right?

The world works in memes and headlines and public markets. So I come from a private market background, infrastructure funds management and going into the public markets, there is a lot of narrative, there is a lot of hype. Um we've tried to keep it very limited to binding contractual deal announcements.

So this is binding. This is real revenue. Sure. Uh just like let's give it up for real. That's [applause] amazing. Yeah. I I honestly think that needs to be clarified, but thank you. Like that is very very helpful. Yeah.

we're in an uh there's so much incentive for for two parties to announce a the biggest possible number without any real underlying commitments. And so uh yeah, I was excited to see you guys uh get this done and announce it with such a great uh counterparty. Yep. Uh well, we'll let you go.

We know you have a busy day, but thank you so much for stopping by the show. Yeah, congrats to the whole team. Massive, massive milestone. We'll talk. Thanks, guys. Good.

Um Elon Musk says a large solarp powered AI satellite constellation would be able [laughter] fix him would be able to prevent global warming by making tiny adjustments in how much solar energy reached the earth. That would be I wonder how much how much capex does that cost?

Uh yeah, it it's interesting that Elon I I don't know. There's just like um when he says something like that, it hits just as like sci-fi techno optimism and it doesn't it doesn't hit as like I mean people still trade the stock off of this like news.

People will look at this and be like oh well like you know like like maybe Tesla will get a piece of that, right? Uh and it becomes a little bit of a narrative but people don't uh people don't dig in that far.

Um, we have one thing one thing that was uh super notable to me uh or standout kind of section of the uh Elon was on Allin Friday and he was talking about the the shareholder vote around his compensation but also the the voting power that he'll have and one of the things that he said was I don't want to create a robot army if I don't have if I can be like fired by ISS or any of these these voting groups.

Uh, and I thought that was [laughter] I mean it's just like the most Elon way way to to uh who should control the robot army? It's a good question. I don't know. He's like Yeah. He basically was saying I don't want to be fired for political reasons.

Typically typically armies are controlled by democratically elected leaders. Like we have a system for electing who is the commander-in-chief of the army. Uh very interesting question. if you actually wind up with a company that has control over a robot army. That's a that's an interesting philosophical question.

Uh also, you know, you're going to need to manage that army. You got to get on linear. Linear is a purpose-built tool for planning and building products. We need the system for modern software development. Streamline issues, projects, and product road mapaps. So, we have Mark German coming on in just a few minutes.

But, we had a question uh message to us. I thought it'd be good to get into kind of break it down. So, here is the question. Advice for my friend's company. He has around 400 million DAU, but he's been diluted by 90% since the IPO. He still has full control of the company, but he can't sell because of antitrust.

Should he bail? I I think no. Take it out. I don't know. There's nothing about that that's identifiable. Uh I think that you stick with it. I think it's got to be fun to run such a big company with that many DAU. Even if you've been diluted, you know, you're you're probably rich. You've sold a lot along the way.

You're you're post economic, but it's better to be post-economic and have control over an interesting organization than post-economic and just kind of retired. That's my take. My mind's just racing on what what company could this possibly be? And and it feels like at 400 million DAUs, it's got to be Snap.

Maybe could be something. Somebody should look into it and see if see if um see what the dilution has been like over there. I think Snap is a company uh that uh I think if you read the numbers it's like every 10 years they just give away the entire company to uh the team.

It was a billion in stockbased comp last year something like that