The AI Buildout Boom: Why Nebius and Friends Might Be Just Getting Started

Quick Read

Here are the big takeaways, explained like you’re five years old:

• Morgan Stanley (a big money advice company) thinks the total "brain power" of the biggest tech builders will almost quadruple — from 31 to 117 gigawatts by 2028. That could cost up to $8 trillion to build!
• Nebius (a company that rents out AI computer power) grew its money earned by 684% compared to last year. It hopes to make $7 billion–$9 billion a year by December, and $51 billion a year by 2030.
• Nebius made a new "light" deal: other people pay to build the buildings, Nebius brings the smarts, and gets paid again and again. This means Nebius doesn’t have to spend all its own piggy bank money.
• Act now: The analyst who predicted NVIDIA’s success back in 2010 just shared his top 10 AI stocks — and Nebius Group didn’t make the list. Grab the names FREE today.

What’s Been Happening With AI Stocks?

Imagine the stock market like a playground. This year, kids (investors) were super excited about companies building AI toys, then suddenly worried: "What if we build too many toys and nobody plays with them?"

That scared away money from smaller AI rental companies like:

• CoreWeave (NASDAQ: CRWV)
• Nebius Group (NASDAQ: NBIS)
• IREN (NASDAQ: IREN)

These companies rent out super-fast computer chips (called GPUs) to AI builders. The old worry was: if the giant tech companies build too much themselves, the small renters lose their edge.

Important Point: A new report from Morgan Stanley says the market was looking at the wrong boogeyman. Instead of "too many toys," we might see one of the biggest building sprees ever!

The AI Buildout Is Only Accelerating

Morgan Stanley counts the five biggest tech builders (hyperscalers). They say total computing power will go from 30.5 gigawatts in 2025 to 116.6 gigawatts by 2028. That’s nearly 4x bigger in 3 years!

Every extra bit of power must be planned, paid for, and built. That means $4 trillion to $8 trillion spent by 2028.

Here’s the simple table of who’s building what (gigawatts = GW, a measure of computer energy):

Source: Morgan Stanley bottom-up hyperscaler compute model.

Important Point: If demand is growing 4x, the "we have too much stuff" worry doesn’t make sense. The big sell-off in AI stocks earlier this year may have been a false alarm.

Act now: The analyst who called NVIDIA in 2010 just named his top 10 AI stocks — and Nebius Group didn’t make the cut. Grab the names FREE today.

Why Nebius Still Has Room To Grow

Nebius is like a kid with a lemonade stand in a neighborhood that’s getting way more thirsty.

• Its money earned in the first quarter jumped 684% vs last year.
• It expects to make $7B–$9B a year by end of this year.
• Long term (by 2030), it aims for $51B a year from stuff it owns.

The New "Asset-Light" Trick

Nebius announced a new way to grow without emptying its wallet:

1. A third-party (someone else) pays to build the data center building.
2. Nebius brings its AI cloud platform, friends (customers), and know-how.
3. Nebius gets paid over and over via fees and sharing the income.

This lets Nebius expand to new places fast, without tying up billions of its own dollars.

Key Takeaway

The worry against Nebius was:

1. AI demand will cool off.
2. Nebius can’t afford to build enough.

Morgan Stanley’s model says #1 is wrong (huge growth!). Nebius’s new partner model fixes #2 (others pay!).

Important Point: Nebius’s stock dropped 35% earlier. If Morgan Stanley is right, that drop was a mistake. But Nebius still must hit its big goals — nothing is guaranteed!

Act now: The analyst who called NVIDIA in 2010 just named his top 10 AI stocks — and Nebius Group didn’t make the cut. Grab the names FREE today.

Summary

• Morgan Stanley sees hyperscaler compute growing from 30.5 GW to 116.6 GW by 2028 (nearly 4x), needing up to $8 trillion.
• Nebius grew revenue 684%, targets $7B–$9B by year-end and $51B by 2030.
• Nebius’s asset-light model lets others fund buildings while it earns recurring fees.
• The old "too much AI supply" fear looks wrong; Nebius’s 35% drop may be a missed panic.
• Risks remain, but patient investors might see opportunity.

FAQ

Q1: What is a gigawatt (GW) in simple terms?
A: It’s a unit of power. Here, it measures how much energy the AI computer buildings use. More GW = bigger AI brain power.

Q2: What is a "neocloud" or "hyperscaler"?
A: A hyperscaler is a giant tech company (like Amazon or Google) that builds huge computer clouds. A neocloud (like Nebius) is a smaller specialist that rents AI chips to developers.

Q3: Why did Nebius stock fall 35%?
A: Investors feared AI builders were making too much capacity, which could hurt renters like Nebius. New data suggests that fear was overblown.

Q4: What does "asset-light" mean for Nebius?
A: It means Nebius doesn’t pay to build the data centers itself. Other owners pay, and Nebius supplies the platform and gets steady fees.

Q5: Is Nebius guaranteed to hit its goals?
A: No. The article says execution risk remains — they still must deliver on the big targets.