Rzkeller

Law partner, former financial advisor, investor

[Rzkeller @rzkeller]

$NBIS I anticipate a short squeeze at some point as the short interest has been rising with Chanos leading the charge. In my experience, this is normal volatility with it generally going down to no more than about 50% ($150) and ultimately rising to new highs above $300.

[Anders @AndersInvests]

$NBIS

We all talk about bottlenecks in the AI supercycle.
GPUs, power, data centres and so on.
But Nebius CTO Danila Shtan makes a more important point: the real constraint is qualitative growth.

Demand is enormous, but scaling a company without diluting its (elite) standards is much harder than simply adding hardware or headcount.

That is what stood out to me most from his appearance on the Beyond Coding podcast.
Danila comes across as a grounded, pragmatic CTO, someone shaped by operational excellence rather than the hype cycle surrounding AI infrastructure.

More broadly, this is the impression I repeatedly get from Nebius management: They're calm, objective, and highly performance-oriented.

They do not seem interested in inflating narratives. It is not their style. Instead, they communicate a level of realism and restraint that feels like an important counterweight to the market’s excitement.

Danila openly admits that he would like Nebius to move faster. But finding people who genuinely fit, particularly in rare, technically demanding fields such as GPU kernel work and deep infrastructure is a very hard task.

Nebius is deliberately hiring only around 10 to 15 engineers per month!!!

At first glance, that may appear low for a company operating in such a fast-moving market. But to me, it signals discipline.

One of my central concerns with Nebius has been whether extraordinary growth could dilute quality over time.
This is a familiar pattern in many organisations: demand accelerates, competition for talent becomes intense, and companies start hiring simply to fill seats.

The outcome is often predictable. People join who do not fit the technical bar, the culture, or the working ethic.
Execution suffers, accountability gets blurred, and the organisation gradually loses the qualities that made it successful in the first place.

Nebius appears aware of that risk. Its measured hiring pace suggests that it prioritises long-term organisational quality over short-term headcount optics.

Danila’s own definition of the CTO role reinforces this. He sees his job less as personally doing the deepest technical work and more as building the right teams, setting clear expectations, and connecting engineering, product, and business.

In his view, nearly all problems start with unclear expectations.

The same pragmatism shapes his technology outlook. Deploying the latest Nvidia hardware at scale for large customers is more complex and time-consuming than it looks from the outside. (which is why we see bitcoin miners like $IREN falling behind)

And on AI agents, he is equally direct: truly autonomous operations remain unrealistic for the foreseeable future. Critical production code that drives meaningful revenue still requires human ownership.

His longer-term prediction is telling: in five years, people may no longer speak about “AI clouds.” It will simply be cloud computing again, because everything will be AI.

That may ultimately be the Nebius thesis: once the hype fades, the companies that win will be those that possess an uncompromising focus on operational quality.

Long $NBIS

[Rzkeller @rzkeller]

$NBIS

AI summary of what is happening in AI over the last month up through today and its impact on Nebius:

The recent flurry of AI chip news heavily impacts Nebius Group (NASDAQ: NBIS), as the company sits directly at the intersection of Big Tech’s spending, Nvidia's supply chain, and the rapid shift toward model inference. Because Nebius operates as an AI-focused specialized cloud provider (renting out massive GPU clusters), these specific shifts act as double-edged swords for its business model.

1. The Inference Race Validation (Bullish)
•The News: OpenAI launching its custom Jalapeño chip and DeepSeek designing its own custom silicon both highlight a hyper-focus on lowering the cost of model inference.
•Impact on Nebius: This validates Nebius’s current strategic roadmap. Nebius just officially completed its acquisition of Eigen AI to power its Nebius Token Factory, which is a specialized, managed inference platform. The massive industry shift toward inference means highly optimized cloud platforms like Nebius are seeing explosive demand. Nebius posted an 841% YoY AI cloud revenue growth in Q1, largely driven by developers rushing to deploy open-source models (like DeepSeek R1 and Llama) through Nebius's efficient pipeline.

2. Nvidia’s Pricing Compression & System Delays (Bearish)
•The News: DeepSeek’s chip pivot and delays in Nvidia’s latest server racks caused Nvidia’s stock to slide, signaling that rising chip competition is beginning to affect hardware pricing. Nvidia B200 chip rental costs fell from $6.11/hr to $4.22/hr over the last month.
•Impact on Nebius: This is a direct threat to Nebius’s top-line margins. Nebius is heavily exposed to Nvidia; Nvidia recently made a $2 billion equity investment into Nebius, and Nebius is slated to be one of the first providers to deploy Nvidia's next-gen Vera Rubin NVL72 platform later this year. If intense hardware competition forces the market rental rate of GPUs down, the premium prices Nebius charges to lease out its compute will drop sharply, potentially crimping its targeted 2026 adjusted EBITDA margin of 40%.

3. The Big Tech "Capex War" & Meta's Cloud Ambitions (Mixed)
•The News: Big Tech’s combined 2026 capex is crossing $700 billion(headlined by Amazon’s $25B bond sale). Concurrently, Meta launched Meta Compute to sell its excess chip capacity back to outside developers.
•Impact on Nebius:
◦The Bad: Meta is one of Nebius's major anchor tenants, having previously engaged in infrastructure agreements with them. Meta’s decision to enter the cloud compute market directly and sell its own excess chip capacity introduces a massive new competitor, creating demand uncertainty for specialized GPU clouds like Nebius.
◦The Good: The sheer scale of the Big Tech spending war has caused an acute global shortage of AI memory chips (HBM). Because Nebius has already aggressively secured its pipeline—expanding data center land to target 4 GW of power capacity (including a massive 1.2-GW site in Pennsylvania)—it holds valuable, scarce infrastructure that smaller AI labs cannot build on their own.

Summary Market Sentiment
The recent news highlights why Nebius’s stock has experienced intense volatility. While its infrastructure footprint is expanding rapidly, it is trapped in a race where AI compute output is deflating in cost.

[shoshin @zen_tropy]

$nbis, $meta, $crwv, $orcl Mstanley out tonight w/ clarifying comments re: bb report. Tldr: If meta does sell compute, it will be as a bare-metal offering of spare internal 1P capacity (not 3rd party-leased) to serve as an "eps bridge" while they develop their core products.

What's key is ms doesn't believe meta can or wants to compete as a full-service stack, since their models are limited & they don't have the expertise, software, or salespeople to service specialized, high-touch enterprise inference markets (precisely the markets $nbis, for instance, seeks to serve). MS suggests meta is not contractually-allowed to resell any of their 3P leased raw silicon from nbis, crwv, orcl, etc., Believes they will save their cutting-edge contracted capacity (e.g., the nebius 12B Vera Rubin order), for internal use.

Net/net: If meta does enter the compute market it will be as a stopgap, and in the bare-metal, older chip market. A minor supply addition, at best; least threatening, arguably, to nebius of all the neo's; and all this only assuming a backdrop where the compute supply were to materially loosen.

[Daniel Koss @daniel_koss]

Let me remind all $NBIS bulls that just a few weeks ago Nebius' CCO Tom Blackwell literally told me "demand from hyperscalers is INFINITE".

Nebius "only" makes 50% of their revenue from hyperscalers by choice.

The margin from selling to enterprises directly is higher.

Hyperscaler demand is definitely not weaking.

just "trust me bro" on this one.

That being said I have no idea how long market will panic irrationally over a nonissue. Sometimes it just does its thing 🤷🏻‍♂️

[Rzkeller @rzkeller]

$NBIS
As I’ve said before, my view is that the only material days are those with 150% or more average volume. This is one of those days. There are maybe only 10 of these days a year including the four days after earnings. Then I look up and see what you guys and any analysts have to say, which I appreciate, and then make my own opinion as to whether the thesis and demand are still intact.

I think it is best to keep it that simple.

I will continue to keep my 100% concentrated position in $NBIS.

Hope this helps someone.

[Mark Hogan @MB_Hogan]

$NBIS JUST IN : New Northwise Model projects Nebius ~$70B in revenue in 2030 🔥

A 2030 PRICE TARGET Extrapolation:

Using Northwise’s model assumptions with their estimated 339M shares outstanding (their base case) and MY OWN 10X revenue multiple applied to their $70B rev projection for 2030, I would arrive at a rough valuation of $700B with a 2030 Price Target of ~$2,065 a share 🔥🔥🔥

[Northwise Project @InvestNorthwise]

Our full $NBIS model is officially here!

Check out the full report in the first comment.

Since our first model Nebius received a significant $NVDA investment, a very underrated $META backstop as a financing lynchpin, and is nearing the 20 site mark globally.

We continue to build our Nebius model from the ground up using a site by site energization method that we now have visibility to run out through 2030.

Thanks to interviews and events with @daniel_koss @mvcinvesting @romanchernin Tom Blackwell and many other Nebius members and contributors, we gained immense insight into how Nebius continues to execute faster that even our aggressive bull case could image.

While we were directionally correct that mix shift would continue to weigh more towards ai cloud contracts with enterprise and ai natives through the end of the decade, we couldn't imagine that it would be near achieved in 2026.

Our 2030 Base case below illustrates just how forward our expectations for growth have shifted at this rate of execution.

Capacity — Connected MW (base case)
2026: 905
2027: 2,142
2028: 3,964
2029: 4,646
2030: 5,200

Undisclosed data center expansion bucket — Connected MW (base case)
2027: 175
2028: 425
2029: 600
2030: 739

ARR per MW (M, base case)
2026: 9.9
2027: 11.3
2028: 12.8
2029: 13.8
2030: 14.5

Exit ARR (B, base case)
2026: 9.0
2027: 24.2
2028: 50.7
2029: 64.1
2030: 75.4

Recognized revenue (B, base case)
2026: 3.4
2027: 15.8
2028: 36.1
2029: 58.1
2030: 70.3

Gross CapEx (B, base case)
2026: 25.0
2027: 39.4
2028: 59.6
2029: 26.2
2030: 25.2

Cumulative 2026–2030: ~$175B

Funding assumptions (base case)

1. Prepayments, % of CapEx: 55%
2. Core OCF, % of EBITDA: 70%
3. External gap, debt/equity: 85/15
4. Blended interest cost: 5.5%

Funding outcomes (B, cumulative 2026–2030, base case)

1. Prepayments: ~95
2. Core OCF ex-prepayments: ~49
3. Debt raised: ~34
4. Equity raised: ~6
5. Ending debt: ~43
6. Ending cash: ~20

Adjusted EBITDA margin (base case)
2026: 40%
2027: 42%
2028: 44%
2029: 45%
2030: 45%

Implied 2030 adj. EBITDA: ~$32B

D&A (B, base case)
2026: 2.9
2027: 8.1
2028: 16.2
2029: 23.1
2030: 27.3

Share count (base case)

Ending diluted shares: ~339M

Base case scenario probability weight: 55%

Thank you to our premium members for your massive support in bringing this refresh so quickly.

Price targets, our portfolio allocation, present value calculations, and our buy/hold/trim/sell zones are now live.

[amit @amitisinvesting]

Today was obviously weird for many reasons.

The Nasdaq $QQQ fell 5.3% intraday.

The last time that even got close to happening? Last Friday.

The time before that? April 2025.

In fact, as per Goldman, today's NDX intraday move was the LARGEST move we have seen in the last 3 years.

You have $NVDA that can't seem to get a bid even after the best earnings the company could have ever delivered. You have $MRVL gaining $100B in a week just because Jensen likes them. You have IPOs coming that are probably not actually worth anything near what they are being justified to go out at, but people are finding the funds to pay for them.

However, you also have the best earnings growth we have ever seen. You have massive compute constraints, creating the market place for the picks and shovels to continue thriving. You have equity dilution from the Mag 7s which feels like it's bearish but in reality its those companies using their equity to drive even stronger growth and set up the next stage for them in an AI driven world.

I don't know if there's a proper conclusion from any of this price action.

It seems like there are absolutely more jitters taking place. More people that are nervous.

Retail activity is similar to October 2025, and in some cases, beyond October 2025 levels. There's obviously more and more people participating which means more and more margin, leverage, etc.

I don't know if that means we take a pause over the coming weeks or that nervousness goes away once these companies again put up massive earnings, but it does feel like 2 sessions in 3 days of historically large Nasdaq intra-day moves means the market is more fearful than not, and that fear should not stop the broader AI thesis from becoming real.

It just means we consolidate the gains and levels we have been able to achieve in the midst of a war until the next major catalyst upwards. AKA, buying the dip short term might be painful if the market keeps testing bulls but long term thesis around AI very much intact and if there is no epic blow up (which again, earnings don't signal that to be the case) then dip buys will be rewarded. Short term...so many parabolic moves have happened that maybe the market really is just trying to catch its breath and that will inevitably shake investors out.

[Rzkeller @rzkeller]

$NBIS it’s interesting to watch the psychology of how people make this complicated:

“Ignore that the business is doing well, the stock is too expensive now.”

“Let me make a few more hundred dollars with a short-term trade. Low chance it could go against me or that there will be significant positive news while I’m out of the stock.”

“I use options; I’m OK with capping my upside; and I’ll look the other way at adverse spreads, thin markets, and the need to have perfect timing.”

“You don’t understand accounting, look at the company’s costs. Recognizing book value is where all the money has been made in the last 30 years.”

“It can’t go up forever. Forget the top all-time investors telling me to keep the winners.”

Nah, I will just keeping holding, while all in, as the team continues to teach a master class in business, in the perfect space.

[Rzkeller @rzkeller]

@RJCcapital Haha; well said

[Rzkeller @rzkeller]

@daniel_koss Hahaha. The last line is the best.

[Rzkeller @rzkeller]

$NBIS - $NVDA rolling out open source looks to be the reason we are up big pre-market

[Roman Chernin @romanchernin]

we badly need strong open-source models, and who is better positioned to deliver them in the "western" world?

[Rzkeller @rzkeller]

$nbis went all in a couple weeks ago after being all in on $nvda the last couple years. You X Twitter folks made me aware of it with great analysis. Thank you.

[Northwise Project @InvestNorthwise]

We are currently updating our $NBIS model.

The direction of travel is becoming pretty clear:

$NBIS may be scaling faster than our current framework gives it credit for.

The old model treated 2 variables conservatively:

Revenue per MW
Enterprise mix shift

Both may need to move higher.

Nebius is not just adding power. It is trying to turn scarce connected capacity into higher-value AI throughput.

That means the real question is not only how many MW come online.

It is how much revenue each MW can support as density improves, Aether scales, and more enterprise workloads move onto the platform.

Our current framework already assumed rising ARR per MW and a gradual shift away from hyperscaler concentration.

Based on the latest operating signals, that shift may be happening faster than expected.

That would matter for revenue quality, margin durability, and the long-term profile of the business.

We are not publishing updated valuation work here.

But the operating scoreboard is straightforward:

Power secured.
MW connected.
Revenue per MW rising.
Enterprise mix improving.

That is the $NBIS thesis in one sequence.

[Rzkeller @rzkeller]

@EndicottInvests Videos that show the cake layers in a way that us non- tech people can appreciate. Help us all understand exactly what we are buying.

[Nate Endicott @EndicottInvests]

$NBIS is the next AWS

[CAB @CABeyney]

It’s okay everyone has the right to their own opinion, and honestly, I don’t really mind.

At the end of the day, our customer is most certainly the next AWS. They are miles ahead of the competition.

That being said, I’m also convinced the market opportunity is so massive that, there is room for everyone to succeed.

[Shay Boloor @StockSavvyShay]

$NBIS is becoming one of the biggest winners in the new AI economy because it can deliver compute efficiently at scale.

The number that mattered most to me was Nebius seeing four or more customers competing for every GPU it brings online (huge for AI cloud market).

Management also said they raised prices again and are still selling out across both old and new GPU generations which is one of the cleanest pricing power signals you can get.

This is a $100B company in the making.

[Rzkeller @rzkeller]

@Sandeman52 Agreed. All in (still)