I’m long NBIS but my biggest concern is they are hoovering up demand whilst the big players are supply constrained and haven’t fully optimised their AI hardware stack. Short-term we see spectacular revenue growth but when supply catches up in next 12-18 months, I worry they will eat into NBIS by offering much cheaper ‘electricity’. Mid-term, NBIS could also struggle to stay competitive on frontier GPUs without dilution, given the very short depreciation cycles and high replacement costs. There’s no winner takes all in this huge market but long-term the minnows will be squeezed by economises of scale unless there are significant switching costs/first mover advantages.
This is a legitimate concern but switching barriers are also high. You stick with the platform you know. I don’t think there will ever be mass exodus from one platform to another in this market. Pie is just growing and growing and I don’t know how it can reverse anytime soon.
Awesome analysis. I love nbis. Will be a bumpy ride but no real reasons to doubt their growth. Yeah they've missed their arr guidance past few earnings but lots of moving parts so it's understandable. Also I've read they will have no shortage of h200s due to the investment nvda made in nbis. If nvda sees something in nbis I'll defer to their judgement.
How do you factor in the supposed Microsoft comments about slowing spend at Coreweave and the impact it had on Corz last week? I am of difffernt mindsets: 1) no one at Microsoft has confirmed the rumor in the FT, 2) if true, the dissatisfaction with coreweave may drive business to Nebius, 3) it could signal a market shift away from data center spend. Or that it’s unrelated entirely. Thoughts?
I don’t think data center spending is going away, I simply think it’s shifting and it’s shifting to inference.
We have seen massive training but we haven’t seen even 1% of the demand come yet. Everything, and I mean everything will be automized and demand will skyrocket. This is not a forecast, this will happen the real question is when it will happen? I think we started that shift and it’ll continue increasingly in the next 1-2 decades.
I mean hyperscalers are mainly investing in compute to train models. But models require compute to handle live traffic too. At some point, gains from pre-training will flatten and investment will shift to inference. And I don’t have a coreweave article. Corz also isn’t in my radar, sorry not being able to help on those 🤷🏻♂️
Very convincing thesis, thanks Oguz. I noticed they there is high option premium on that stock (pricing a lot of volatility and upside at the same time?)
I didn’t actually check it Julien but it makes sense, it’s volatile. I wouldn’t start with options now and even if I did, I would rather buy buy a call or sell a put than otherwise.
Thank you for this. How do you view the supposed displeasure of Microsoft with Coreweave? I note that MSFT didn't confirm the report in the FT -- but I wonder if this could be a positive for NBIS?
Very interesting article! This is a company I have wanted to dive into for a while and still hope to do it in the future, your description of the origins and business model of the company are very clear.
I’m long NBIS but my biggest concern is they are hoovering up demand whilst the big players are supply constrained and haven’t fully optimised their AI hardware stack. Short-term we see spectacular revenue growth but when supply catches up in next 12-18 months, I worry they will eat into NBIS by offering much cheaper ‘electricity’. Mid-term, NBIS could also struggle to stay competitive on frontier GPUs without dilution, given the very short depreciation cycles and high replacement costs. There’s no winner takes all in this huge market but long-term the minnows will be squeezed by economises of scale unless there are significant switching costs/first mover advantages.
This is a legitimate concern but switching barriers are also high. You stick with the platform you know. I don’t think there will ever be mass exodus from one platform to another in this market. Pie is just growing and growing and I don’t know how it can reverse anytime soon.
Awesome analysis. I love nbis. Will be a bumpy ride but no real reasons to doubt their growth. Yeah they've missed their arr guidance past few earnings but lots of moving parts so it's understandable. Also I've read they will have no shortage of h200s due to the investment nvda made in nbis. If nvda sees something in nbis I'll defer to their judgement.
Indeed, plus they announced that GB200’s are coming soon.
How do you factor in the supposed Microsoft comments about slowing spend at Coreweave and the impact it had on Corz last week? I am of difffernt mindsets: 1) no one at Microsoft has confirmed the rumor in the FT, 2) if true, the dissatisfaction with coreweave may drive business to Nebius, 3) it could signal a market shift away from data center spend. Or that it’s unrelated entirely. Thoughts?
I don’t think data center spending is going away, I simply think it’s shifting and it’s shifting to inference.
We have seen massive training but we haven’t seen even 1% of the demand come yet. Everything, and I mean everything will be automized and demand will skyrocket. This is not a forecast, this will happen the real question is when it will happen? I think we started that shift and it’ll continue increasingly in the next 1-2 decades.
thank you. What do you mean by "shifting to inference"?
and any thoughts on CORZ here? down 30% since the coreweave article.
I mean hyperscalers are mainly investing in compute to train models. But models require compute to handle live traffic too. At some point, gains from pre-training will flatten and investment will shift to inference. And I don’t have a coreweave article. Corz also isn’t in my radar, sorry not being able to help on those 🤷🏻♂️
Very convincing thesis, thanks Oguz. I noticed they there is high option premium on that stock (pricing a lot of volatility and upside at the same time?)
I didn’t actually check it Julien but it makes sense, it’s volatile. I wouldn’t start with options now and even if I did, I would rather buy buy a call or sell a put than otherwise.
On NBIS, how do you value Avride?
I didn’t actually. They are generating a bit revenue but they are still too speculative. Just consider it as a factor increasing the option value.
Thank you for this. How do you view the supposed displeasure of Microsoft with Coreweave? I note that MSFT didn't confirm the report in the FT -- but I wonder if this could be a positive for NBIS?
Very interesting article! This is a company I have wanted to dive into for a while and still hope to do it in the future, your description of the origins and business model of the company are very clear.