Current Status: Full Port $NBIS Substack @nebulizerlabs Opinions are biased. Portfolio is undiversified.
New York / Vancouver
Joined November 2025
- Tweets7.5K
- Following187
- Followers12.3K
- Likes23.1K
As Mark said during $NBIS Live today, $NBIS will deliver 1GW+ of Vera Rubin capacity in 2027.
The $12B $META deal is entirely for VR200s, which reinforces where GPU mix is heading.
This is what I believe the market is most severely underpricing when modeling forward ARR.
During Q2 earnings management stated they were deliberately withholding contracting out all 2027 capacity. Additionally, Reflection deal, one of the $1b+ deals signed at $20-25MM/MW/YR, is already in the process of actively deploying per @BKad2005's talks at Builders & Brews.
This means the monetization rate a blended figure reflecting GB300s and potentially some early Vera Rubin deployments.
But 2027 is a different story. As the hardware mix shifts dominantly towards Vera Rubin, so does the revenue potential per MW.
Don't want to put a number on this yet, but know that $20-$25MM is the floor, and that 2027 capacity blend is 25% hyperscaler, 75% Enterprise / AI Natives.
xitter.cf/SMASIMHO/status/211252…
The biggest tailwind no one is baking into their $NBIS 2027 ARR estimates is $NVDA Vera Rubins.
The huge amount of 2027 compute will be for VR200s. So far, the numbers provided by $NBIS management during Q2 earnings in the $20-$25MM/MW/YR range are all for Blackwells, specifically B300s.
"We could sell today our entire 2027 capacity on these terms if we wanted to. We are not doing this."
- Arkady Q2 '26 Earnings
In other words, these numbers are the floor. Where is the upside? Well here's something to chew on.
@JensenHuang said during Q2 2026 $NVDA earnings that Rubin's will go for $40B/GW.
That's a 60% increase in hardware $/MW compared to GB300s, not annual cloud rental revenue. But is illustrative of the step function change in economics.
And I trust that $NBIS management will be conducting auctions galore to maximize producer surplus here.
xitter.cf/mernit/status/21108968…
Obviously guys, we aren't contracting out all this at Elon's risk-on short term contract rates so we aren't going to see a blended rate of $50MM/MW/YR HAHA (i wish).
$NBIS's core customer is enterprise because that's the cornerstone of longevity and stability. We may see an increase in number of auction and scale training deals, but by in far expect bulk of deals to be 1-3YR.
NVIDIA GTC PREGAME xitter.cf/i/broadcasts/1pKdRDpVq…
🚨 $NBIS Update: TD SYNNEX Partnership 2.0
$NBIS already had an existing partnership with TD in North America, and Europe is next.
TD is one of the world's largest IT middlemen. It connects $NBIS with thousands of tech sales and service firms that already help enterprises buy and deploy tech.
In this case we're talking about $NBIS AI Cloud, Token Factory, and Agents Blueprint.
This is essentially a play on operating leverage. Sales is expensive. $NBIS spent $173.9M on SG&A in Q2 alone.
Instead of building out its own enterprise salesforce entirely from scratch, $NBIS is leveraging TD's existing distribution network, reseller relationships, and GTM infra.
Much like the $PLTR partnership, this is another pathway to accelerate enterprise AI adoption through targeted outreach, customer education, and high-touch sales support.
xitter.cf/EndicottInvests/status…
$NBIS: TD Synnex and Nebius are now expanding Across Europe 🔥🔥
Just in time for GTC Berlin tomorrow 😉
"The agreement provides partners across TD SYNNEX's European footprint with access to Nebius AI Cloud, a purpose-built AI platform powered by NVIDIA accelerated computing technologies for AI training and inference workloads. Through a flexible consumption model, partners can rapidly provision enterprise-grade AI infrastructure and gain access to NVIDIA GPU capacity without the significant upfront investment, procurement challenges and operational complexity traditionally associated with deploying high-performance AI environments."
Huge shoutout to @AIMoneyTroels for calling this out
Arkady going on air in 7 hours to pump $NBIS.
3-5AM EST NVIDIA GTC Pregame.
Inspired by @EndicottInvests I'm going to attempt to stream this live on X for all my friends in the Eastern Hemisphere and insomniacs at home.
Expect PLENTY of technical difficulties.
Link to Official NVIDIA GTC Stream:
youtube.com/watch?v=PrA_ayyy…
Guys I found my soul mate. You all know I have my $NBIS Data Center Tracker.
Guess what @anthonybs created?
A $NBIS Customer Tracker. My favorite discovery? What a CRAZY CAC funnel Tavily is.
Where else do you see such a talented community of builders? Well, like father (Nebius) like son (Nebians).
This guy has more data pipelines and validation workflows running than a small business.
Seriously, check it out. You'll be surprised.
nebiusatlas.com
The biggest tailwind no one is baking into their $NBIS 2027 ARR estimates is $NVDA Vera Rubins.
The huge amount of 2027 compute will be for VR200s. So far, the numbers provided by $NBIS management during Q2 earnings in the $20-$25MM/MW/YR range are all for Blackwells, specifically B300s.
"We could sell today our entire 2027 capacity on these terms if we wanted to. We are not doing this."
- Arkady Q2 '26 Earnings
In other words, these numbers are the floor. Where is the upside? Well here's something to chew on.
@JensenHuang said during Q2 2026 $NVDA earnings that Rubin's will go for $40B/GW.
That's a 60% increase in hardware $/MW compared to GB300s, not annual cloud rental revenue. But is illustrative of the step function change in economics.
And I trust that $NBIS management will be conducting auctions galore to maximize producer surplus here.
xitter.cf/mernit/status/21108968…
the world isn't ready for vera rubin
the transition from B300 to Vera Rubin is a generational hardware upgrade where the architecture, cooling, and data center facilities are fundamentally different from the prior generation
consider a few things:
1. vera rubin deliveries are being prioritized to only the closest Nvidia partners and won't be widely available until 2027
2. less than 10% of data centers in the US will be equipped to run vera rubin *12 months* from now
3. of those that have the cooling and facilities to support vera rubin, only a small fraction of that group even has power contracted
combine this with the current demand for inference, and it's a perfect storm for GPU prices to continue climbing