Nebius Group is an emerging leader in AI cloud, data centers, robotaxis, and AI data infrastructure. It is building the foundation to become the AWS of AI.
Thank you for your deep dive on Nebius. I have entered a bullish setup on NBIS based on your analysis (and a few others), and have added a link to your deep dive in my recent post describing the trade. It will be great to get your feedback on my NBIS trade setup. Thank you.
2026-04-10 13:56 Data Centers NBIS Neutral Trade Reducing stake 20% "Taking a few chips off the table with NBIS
Reducing stake 20%
A $35–45B valuation range makes the company quite dependent on larger projects like Independence to sustain the upside
In the meantime, there can be approval issues, construction delays. We know how tight things are in the supply chain
Of course, we don’t know the specifications of the deal with Meta, and NBIS has other interesting projects, particularly across Europe, but I think this valuation at a $37B market cap requires much more executional perfection"
I tend to be quite skeptical that Nebius will be able to compete with the hyper scalers in the long run. This truly seems to be a business where scale drives the lowest cost.
They had a very compelling offer, which I believe may no longer be available. I don’t remember exactly which page of their site I saw it on, but the price really did drop that low for new users. While I wanted to highlight Nebius offering things that hyperscalers wouldn’t, you're right that in terms of overall pricing, it's around $2/hour without any discount.
Time will tell with hyperscalers. The need for compute is so massive that, at the moment, it’s not so much about who’s best, but about who can actually supply it. In the long run, I believe Nebius is well-positioned to scale up as needed. Their Jersey and Finland data centers are large-scale, and their competitive edge in engineering and software development will also work in their favor.
You just have to look at developer testimonials to see how much better the Nebius cloud is compared to AWS.
Also, I don’t think this is an industry where economies of scale are all that crucial. As long as you have enough capacity to power entire clusters and access to decently priced electricity, you’re all set. There aren’t many costs you can cut. In fact, in that regard, Nebius may be able to cut more than others; they don’t rely on Supermicro, Dell, or other server designers. They have their own, better product. And their data centers are all newly built and AI-purposed from the start. Nothing is repurposed. All of their data centers are also in cold climates, which helps with a crucial factor: heat.
I get that they design their own servers and do all this intricate cooling stuff, but I highly doubt that hyperscalers haven’t looked into this or made plans to do something like this in the future.
With regards to scale, scale allows for insane discounts when negotiating with suppliers for parts. Nvidia is for sure giving discounts to hyper scalers they are not giving to Nebius, and the same goes for servers or other pieces of the stack. Heck, these hyper scalers are even getting discounts on energy by producing their own electricity or partnering with other utilities (not sure if Nebius is doing this).
I get that with the current GPU shortage, compute is in such high demand. But you had said that you have a 5-10 year time horizon. In that time, a hyperscaler could have excess compute or stop pursuing their own foundational models (most likely Amazon). This would lead to them dropping the unused compute on the market and undercutting Nebius. I just don’t see what long term value Nebius has. It seems as though they only have pricing power in a compute starved world.
I recently wrote about WeRide - the robotaxi business that is in second place behind Waymo - nobody is talking about them and the valuation is under 3 B...
Thank you for your deep dive on Nebius. I have entered a bullish setup on NBIS based on your analysis (and a few others), and have added a link to your deep dive in my recent post describing the trade. It will be great to get your feedback on my NBIS trade setup. Thank you.
Thank you!
April’2026
Call was to sell 20 % of the portfolio is it true ? Or I am wrong
Where did you get that from?
On your portfolio Sheet name updates of the the three sheets of the portfolio,
Updated feed row number 13 and 14
2026-04-10 13:56 Data Centers NBIS Neutral Trade Reducing stake 20% "Taking a few chips off the table with NBIS
Reducing stake 20%
A $35–45B valuation range makes the company quite dependent on larger projects like Independence to sustain the upside
In the meantime, there can be approval issues, construction delays. We know how tight things are in the supply chain
Of course, we don’t know the specifications of the deal with Meta, and NBIS has other interesting projects, particularly across Europe, but I think this valuation at a $37B market cap requires much more executional perfection"
I reduced my stake solely in NBIS
Thanks Daniel , I follow your portfolio for my portfolio allocation
https://walkerrauschenberg.substack.com/p/nyse-four?r=5rdmrq
I’m not sure where you’re getting that H100 pricing number from. On demand H100 pricing is $2.95 and reserved H100s are $2/hr
https://nebius.com/prices
I tend to be quite skeptical that Nebius will be able to compete with the hyper scalers in the long run. This truly seems to be a business where scale drives the lowest cost.
They had a very compelling offer, which I believe may no longer be available. I don’t remember exactly which page of their site I saw it on, but the price really did drop that low for new users. While I wanted to highlight Nebius offering things that hyperscalers wouldn’t, you're right that in terms of overall pricing, it's around $2/hour without any discount.
Time will tell with hyperscalers. The need for compute is so massive that, at the moment, it’s not so much about who’s best, but about who can actually supply it. In the long run, I believe Nebius is well-positioned to scale up as needed. Their Jersey and Finland data centers are large-scale, and their competitive edge in engineering and software development will also work in their favor.
You just have to look at developer testimonials to see how much better the Nebius cloud is compared to AWS.
Also, I don’t think this is an industry where economies of scale are all that crucial. As long as you have enough capacity to power entire clusters and access to decently priced electricity, you’re all set. There aren’t many costs you can cut. In fact, in that regard, Nebius may be able to cut more than others; they don’t rely on Supermicro, Dell, or other server designers. They have their own, better product. And their data centers are all newly built and AI-purposed from the start. Nothing is repurposed. All of their data centers are also in cold climates, which helps with a crucial factor: heat.
I get that they design their own servers and do all this intricate cooling stuff, but I highly doubt that hyperscalers haven’t looked into this or made plans to do something like this in the future.
With regards to scale, scale allows for insane discounts when negotiating with suppliers for parts. Nvidia is for sure giving discounts to hyper scalers they are not giving to Nebius, and the same goes for servers or other pieces of the stack. Heck, these hyper scalers are even getting discounts on energy by producing their own electricity or partnering with other utilities (not sure if Nebius is doing this).
I get that with the current GPU shortage, compute is in such high demand. But you had said that you have a 5-10 year time horizon. In that time, a hyperscaler could have excess compute or stop pursuing their own foundational models (most likely Amazon). This would lead to them dropping the unused compute on the market and undercutting Nebius. I just don’t see what long term value Nebius has. It seems as though they only have pricing power in a compute starved world.
I recently wrote about WeRide - the robotaxi business that is in second place behind Waymo - nobody is talking about them and the valuation is under 3 B...
This aged very well.
Ive been in nbis for a year. Still learned a lot reading this article. Thanks. Institutions have to be buying right?
Excellent article !! The company has solid potential plus they have shareholders interest at heart !!
x10 in 5 years, 10^(1/5) = 58.5% CAGR is very good~! It is my high conviction name.