$NEBIUS(NBIS)$ $CoreWeave, Inc.(CRWV)$ $NVIDIA(NVDA)$
Something extraordinary is happening in AI infrastructure.
The demand is real.
The contracts are enormous.
But the share prices?
BRUTAL.
Nebius (NBIS) fell:
16.84% on Thursday.
Then another:
9.92% on Friday.
Closing at:
$115.88.
CoreWeave (CRWV)?
Friday close:
$53.13.
Yet Nebius just reported:
571% REVENUE GROWTH.
And CoreWeave has accumulated:
$132.5 BILLION IN BACKLOG.
So why are investors selling?
I think Wall Street has stopped asking how much AI computing companies can sell.
The question now is:
WHO WILL PAY TO BUILD IT ALL?
And I found something in Nebius's earnings that makes this question even more interesting.
$325 MILLION.
๐ฅ THE $325M CONTRADICTION IN NEBIUS'S EARNINGS
Nebius reported Q3 revenue of:
$980.8M.
Above the approximately $930M consensus highlighted in the market discussion.
Adjusted EBITDA?
$465.8M.
Excellent growth.
But its shareholder letter reveals something important.
Approximately:
$325M
of quarterly revenue came from non-recurring asset-light partnership activity.
That's roughly:
33% OF TOTAL REVENUE.
Without that contribution, reported revenue would have been approximately $656M.
Now, that doesn't mean the revenue was artificial.
The activity was real.
And management expects these partnerships to become a more regular part of the business as they scale.
But it raises a legitimate question.
HOW REPEATABLE IS THIS REVENUE?
A strong quarter isn't simply about beating expectations.
It's about how much of that revenue can be reproduced consistently.
And the market found another concern.
Nebius now expects 2026 revenue near the lower end of its:
$3.0Bโ$3.4B GUIDANCE.
Its previously targeted December annualised revenue run rate is now expected in February.
Why?
Some third-party data-centre deliveries have slipped into 2027.
DEMAND DIDN'T DISAPPEAR.
DELIVERY TIMING CHANGED.
But for a stock priced on extraordinary growth, even a delay matters.
๐ง HERE'S THE PART THAT MAKES ME MORE OPTIMISTIC
Nebius reported that customer prepayments on deals signed in Q3 covered:
MORE THAN 100% OF ASSOCIATED CAPEX.
That is a substantial improvement from the 50โ60% reported for Q2 deals.
Management also estimates payback on comparable new deals has improved to approximately:
1 YEAR AND 7 MONTHS.
And for 2027, Nebius expects customer prepayments and investment-grade asset-backed financing to cover:
75%โ85% OF PLANNED CAPEX.
That's encouraging.
Because Nebius plans to spend:
$40Bโ$45B IN 2027.
But here's the distinction.
That 75โ85% figure is management's expectation.
Not money already collected.
And the greater-than-100% prepayment coverage applies to the associated costs of newly signed Q3 deals, not all company expenditure.
I wouldn't describe Nebius as self-funding yet.
Its Q3 operating cash outflow was:
$273M.
Purchases of property, equipment and intangible assets were:
$7.30B.
The business is expanding.
The financing structure is becoming more interesting.
But the enormous spending commitment hasn't gone away.
๐จ COREWEAVE HAS A DIFFERENT FINANCING PROBLEM
CoreWeave reported Q3 revenue of:
$3.66B.
Its backlog reached:
$132.5B.
That's an extraordinary amount of committed future business.
But the company also reported:
Net interest expense: $846M.
GAAP net loss: $709M.
And management expects Q4 interest expense of approximately:
$1.05Bโ$1.15B.
Think about that.
More than $1 billion in expected quarterly financing costs.
That's the scale of the burden CoreWeave must manage.
Its 2026 capital expenditure guidance is:
$37Bโ$39B.
But there are positives.
CoreWeave expects approximately $9B in customer prepayments and deposits over the following three months.
It has also been moving toward more asset-backed financing, including loans secured against computing equipment and contracts.
That could improve the cost and structure of future funding.
Still, I want evidence that the financing burden becomes more manageable as revenue scales.
Because:
BACKLOG IS NOT CASH.
And $132.5B of CoreWeave backlog cannot be directly compared with Nebius's $38.7B of remaining performance obligations without recognising their different definitions.
๐ต SO WHICH BUSINESS MODEL DO I PREFER?
This is where I become selective.
NEBIUS
I like its improving contract pricing, customer prepayments and potential for asset-light expansion.
But I need better confidence in capacity delivery and repeatable revenue.
COREWEAVE
I like its scale, customer contracts and established financing channels.
But the interest burden is enormous.
So if I were evaluating the businesses today?
NEBIUS HAS THE MORE INTERESTING IMPROVEMENT IN CONTRACT ECONOMICS.
COREWEAVE HAS MORE REVENUE SCALE, BUT A HEAVIER VISIBLE INTEREST BURDEN.
Neither is risk-free.
And neither deserves to be bought blindly because it has fallen.
๐ฎ MY NBIS & CRWV PREDICTION: NOVEMBER 16โ20
These are my subjective scenarios for next week.
๐ข BULL CASE: 25%
AI sentiment improves.
Treasury yields stabilise.
Nvidia's November 17 earnings reassure investors about infrastructure demand.
Buyers return to beaten-down neocloud stocks.
Potential recovery:
NBIS: $125โ$140
CRWV: $58โ$65
๐ก BASE CASE: 50%
MY MOST LIKELY SCENARIO.
Demand remains strong.
But investors continue questioning capital spending, financing costs and delivery schedules.
Expect sharp rebounds followed by profit-taking.
Working ranges:
NBIS: $105โ$125
CRWV: $48โ$58
๐ด BEAR CASE: 25%
AI infrastructure selling resumes.
Bond yields remain elevated.
Financing concerns and construction delays dominate sentiment.
Potential downside:
NBIS: $95โ$108
CRWV: $42โ$48
These are conditional trading scenarios, not guaranteed forecasts.
๐ฏ MY TRADING MAP
NEBIUS (NBIS)
$109: Friday's intraday low, first support test.
$120โ$125: Initial recovery hurdle.
Above $128โ$130: Stronger evidence of recovery.
Below $109: Downside risk increases.
COREWEAVE (CRWV)
$49โ$50: Immediate support test.
$56โ$58: First recovery hurdle.
Above $60โ$62: Stronger bullish confirmation.
Below $49: Risk of a deeper retreat.
I wouldn't make a large entry in either company simply because prices have fallen.
My preference would be to see buyers defend support, followed by a breakout with sustained volume.
And I wouldn't forget that both remain exposed to changes in financing conditions.
๐ฅ MY CALL: I THINK THE WINNER WILL BE DECIDED BY WHO NEEDS LESS OUTSIDE MONEY TO DELIVER THE NEXT BILLION DOLLARS OF REVENUE.
Here's my central thesis.
The AI infrastructure boom isn't necessarily ending.
But investors are increasingly unwilling to finance growth at any price.
Nebius has demonstrated improving customer-contract economics.
CoreWeave has demonstrated enormous revenue scale and future demand.
Both are impressive.
Both are expensive businesses to expand.
So I won't judge them simply by their growth rates.
I want to see:
CONTRACTS โ PREPAYMENTS โ LIVE CAPACITY โ REVENUE โ CASH RETURNS.
For next week, I expect volatility before a sustained recovery.
My base cases remain:
NBIS: $105โ$125.
CRWV: $48โ$58.
If I had to pick one business to study more closely?
NEBIUS.
Because its improving customer prepayments could meaningfully change its financing economics if management delivers as promised.
But that doesn't make NBIS an automatic buy at $115.
I still want evidence that the delays are being resolved and that the next quarters produce increasingly repeatable revenue.
AI DEMAND IS NOT THE PROBLEM.
TURNING THAT DEMAND INTO FINANCIALLY SUSTAINABLE GROWTH IS THE TEST.
And that's where I think the next major AI winners will separate themselves.
I am not a financial advisor. Trade wisely, Comrades!
Would you buy NBIS or CRWV after the selloff, or wait for stronger proof that their financing models are sustainable?
If you find this post interesting, give it wings! ๐ชฝ Repost and share the insights.
Do consider โFollow meโ and get firsthand read of my daily new post. Thank you.
@TigerPM @TigerStars @Daily_Discussion @TigerEvents @Tiger_SG
Comments