I want to jump straight to my thought.
In the latest quarter, they acquired Mirantis, a cloud software service that help to manage the AI cloud. And from what CEO said, it seems like they want to chase the end customers to get higher profit margin. Here’s the quote:
Daniel Roberts (CEO): There are benefits in having hyperscale clients in terms of financability, contractual certainty, but there are also consequences in terms of price because you’re not servicing the end customer in many of those instances. The ability to service the end customer has been something we’ve focused on since day one. All of our early deployments have been very focused on non-hyperscale customers and getting as close to AI natives and enterprise as we can. The Mirantis acquisition certainly helps that. I’m not gonna sit here and say we’re going 100% hyperscale, we’re going 100% AI native end market. The reality is that blend will just emerge organically over time. … I think a combination of hyperscale, a hot combination of other is absolutely the goal.
Okay it sounds good for long term. But my issue here is that, in the past they always said bare-metal GPU is superior. It is what their main customers (hyperscaler) want. They want to just chase the hyperscaler which is the opposite direction of what NBIS is doing. But now they shift their direction 180 to chase the enterprise customer instead.
This makes me start to be question myself that can I trust the management ? It looks like the management is not telling the whole story and just try to change narrative to make the company look good rather than being honest.
14 Likes
Hi punkz
Yes, I know what you mean. If I were to guess, I would say they are not purists, rather they are hard nosed business people (as evidenced by their very successful bitcoin mining business).
They probably saw the worsening supply demand situation for compute, and realized it was a good idea to pivot sooner rather than later to more enterprise customers. I think it was a sensible move, but as you note they have the (only slightly) uncomfortable task of making it sound like this was their plan all along. It’s business, no?
6 Likes
I agree. The plan is good for long term, I don’t argue with that. But I’m just disappointed with how they communicate…
2 Likes
IREN’s management shows smart flexibility, pivoting to enterprise/AI-native customers via Mirantis without abandoning hyperscalers, which positions them better for a changing AI cloud market. This adaptability, rooted in their mining background, turns potential supply squeezes into margin opportunities.
This all makes sense to me.
This isn’t a 180° flip—it’s evolution: hyperscalers provide scale/financing, enterprises offer 2-3x higher margins via software layers.
10 Likes
The news I follow has the demand situation for compute getting stronger. The demand is growing significantly faster than they can build supply. The demand is so high that multiple of the hyper scalers had to limit compute to their customers.
Microsoft stopped signups for their compute recently. Anthropic created smaller usage during peak hours. Anthropic solved their issues by renting from xAI the only company with spare compute. And once anthropic had available compute they removed the peak hour limits and doubled the session limits.
Anthropic CEO said their demand for compute went up 80x in the first quarter of 2026 and they had planned for 10x
Iren has probably been following the news carefully. It’s easier to sell to hyperscalers as bare metal but I think it’s more valuable to sell to the medium and large companies that want compute but can’t afford to build datacenters. I think with productivity gains shown to those companies is well worth the cost and those that didn’t order compute are jumping in on it now and having to pay a premium.
Iren management has always been we will pivot to what will give us the best return. They first went bare metal because pivoting from Bitcoin to hyper scalers compute was enough of a challenge at one time.
Drew
13 Likes
Engineers are not always the best at that task. Another way of looking at it is that there is enough humility to adjust the mission slightly. My issue is that there are a ton of cloud and AI management tools. Why are theirs better?
2 Likes
I didn’t phrase that very clearly – I in fact meant that there’s a rapidly worsening balance of supply versus demand. Demand is outstripping supply at an increasing pace. cm
4 Likes
Iren has been a low conviction position for me. Not long ago, I sold all my holdings in this name at a profit and redeployed the funds elsewhere. Later, I bought back in, and after this most recent quarterly report, I sold again. I am usually not quite so fickle, but I don’t trust Iren’s management. An argument might be made that the embrace of offering a more full service product as opposed to Iren’s prior claims that their customer base wanted bare metal demonstrates “flexibility.” As for me, it’s indicative of inconsistency and a failure to truly understand the market they are attempting to penetrate.
It is my assessment that this management team does not understand which set of potential customers might provide the maximum profitability which leads them to a strategy of chasing rainbows. So, once again I have exited Iren.
It was not a large position, but I generally don’t sell without a target for the funds. I will put some of the $ in SIMO. Thank you @wpr101 for that pick. And I am looking at NET again. I’ve always liked the company. Years ago I made a barrel of money with NET leaps. But, the stock just got too expensive so I haven’t had a position for years. IMO, the recent sell off is way overdone and I think that for the first time in a long time it makes sense to start a position in NET.
Of course, contrary opinions are welcome. The beauty of this board is the collective intelligence shared here.
12 Likes
True. The problem is management’s expertise is in building barebones infrastructure, not providing AI-stack services. That’s where I believe NBIS has a natural advantage.
Not necessarily comparing the two, but it does seem like IREN’s been forced to pivot at a stage that likely wasn’t on the initial drawing board.
7 Likes
Were they forced, or did they come across an opportunity that was too good to pass up?
In one fell swoop, they get an AI software infrastructure business, engineers to further develop it, a customer support organization, and 1500 customers (including PayPal and Adobe, for example). That’s bad?
Nebius management touts their software infrastructure advantage. What percentage of their revenue comes from customers taking advantage of that, versus bare metal rentals?
16 Likes