To The Moon

Pretty wild turnaround for AI infrastructure stocks. $CoreWeave, Inc.(CRWV)$ $NEBIUS(NBIS)$

Just a few days ago, everyone seemed worried about excessive AI spending, rising debt and whether data center investments could ever generate enough returns.

Then CoreWeave ($CRWV) reported earnings, and suddenly the market was buying AI infrastructure stocks again.

On November 11, CRWV jumped 24.7%. Nebius ($NBIS) gained 15.4%, Applied Digital ($APLD) rose 7.4%, and NVIDIA ($NVDA) was up 2.9%.

The Philadelphia Semiconductor Index also rallied 5.7%.

So what changed?

CoreWeave reported $3.66 billion in Q3 revenue, up 168% year over year, with revenue backlog reaching an incredible $132.5 billion.

Those numbers are impressive.

But here's the part I find more interesting.

The company also reported a $709 million net loss, while quarterly net interest expense reached $846 million.

Think about that for a second.

Revenue is growing at triple-digit rates, the company has over $130 billion in contracted future revenue, and yet financing costs are still eating up a huge amount of money.

So is this rally telling us that AI infrastructure demand is stronger than expected?

Or are investors getting excited about another capital-intensive growth story before the economics have really been proven?

I think there's a bit of truth on both sides.

The AI demand is real. That part is getting harder to argue with.

Let's start with the good news.

CoreWeave's revenue increased from $1.37 billion a year ago to $3.66 billion this quarter.

That's not just a strong earnings report. It's a massive expansion in the scale of its business.

More importantly, its revenue backlog grew from approximately $104 billion at the end of Q2 to $132.5 billion at the end of Q3.

That's a lot of contracted future demand for AI computing capacity.

And I think this matters because one of the biggest debates around AI spending has been whether companies are building infrastructure faster than actual customer demand.

CoreWeave's numbers suggest that at least some specialized AI cloud providers still have substantial demand ahead of them.

But there's an important distinction here.

A $132.5 billion backlog isn't the same as $132.5 billion in cash.

Those contracts need to be fulfilled over time. CoreWeave still has to build and operate the infrastructure, deliver the contracted computing capacity and collect payments.

And that's where the business becomes more complicated.

AI data centers aren't like traditional software products.

You can't just develop something once and sell it to millions of customers at almost no additional cost.

You need GPUs, servers, networking equipment, cooling systems, electricity and physical data center capacity.

All of that requires capital.

And a lot of it needs to be spent before the associated revenue is collected.

This is why I think looking at revenue growth alone can be misleading when evaluating companies like CoreWeave.

The demand may be real, but the cost of meeting that demand is just as important.

What really caught my attention was the financing story.

CoreWeave's debt had reached approximately $46.9 billion by the end of September.

Its Q3 net interest expense was $846 million, compared with just $207 million in GAAP operating income.

That's a pretty significant difference.

The company is generating operating profits, but those profits aren't enough to cover its financing costs.

And this has been one of my biggest concerns about the AI infrastructure trade.

Companies like CoreWeave need enormous amounts of capital to expand capacity, and their economics depend heavily on the cost of that capital.

A business can grow revenue rapidly and still struggle to generate attractive returns for shareholders if financing costs remain too high.

But management did share something encouraging this quarter.

CoreWeave expects to receive around $9 billion in customer prepayments and deposits over the next three months from existing contracts.

It also plans to rely more on financing backed by contracted customer revenue rather than expensive unsecured borrowing.

That's potentially a meaningful shift.

If customers are willing to provide cash upfront, CoreWeave may be able to fund more of its expansion without relying as heavily on traditional debt markets.

And if borrowing costs eventually decline, the company's financial position could improve.

But I wouldn't call this problem solved just yet.

Customer prepayments aren't profits. They come with future service obligations.

CoreWeave still needs to deliver the computing capacity customers are paying for.

And if construction costs rise, deployment schedules slip or future demand slows, the financing pressure could return.

Better financing terms make the business more sustainable. They don't automatically make it profitable.

The company is targeting profitability in 2028. But the spending isn't slowing down yet.

Management expects to reach adjusted net profitability in the first half of 2028, with a goal of delivering positive adjusted earnings for the full year.

That's encouraging, but it's still a forward-looking target.

And adjusted profitability isn't the same as GAAP profitability or positive free cash flow.

Meanwhile, CoreWeave expects full-year 2026 capital expenditures of approximately $37 billion to $39 billion.

Its Q4 net interest expense is also expected to reach $1.05 billion to $1.15 billion.

Those are enormous numbers.

I think the biggest question for CoreWeave is no longer whether customers want AI computing capacity.

It's whether the company can build that capacity at a cost that makes economic sense over the long run.

There's also the issue of hardware depreciation.

GPUs don't stay at the cutting edge forever.

New generations of chips are becoming more powerful and efficient, and the economics of older hardware could change as supply expands.

That doesn't mean existing GPUs suddenly become worthless.

Older chips may continue generating revenue from inference and other workloads for years.

But the prices customers are willing to pay for that capacity could decline.

If equipment depreciates faster than expected, or rental prices fall while debt obligations remain fixed, margins could come under pressure.

This is something I think investors need to watch much more closely.

Having a huge backlog is great.

But how much cash will CoreWeave actually keep after paying for infrastructure, financing and operations?

That's the number that will ultimately matter.

What does this mean for NVIDIA and the rest of the AI trade?

One thing I found interesting about the rally was how quickly sentiment spread across the AI infrastructure sector.

CRWV jumped nearly 25%, but NBIS and APLD also rallied sharply.

That tells me the market wasn't just reacting to one company's earnings.

Investors were reassessing the broader AI infrastructure story.

And I can understand why.

If AI cloud providers are still signing large contracts and growing revenue rapidly, it suggests demand for computing infrastructure remains strong.

That's potentially positive for several parts of the supply chain.

But I wouldn't treat all AI infrastructure companies as the same investment.

For example, $NVDA and $CRWV have very different business models.

NVIDIA supplies the computing platforms and technology that power AI infrastructure.

CoreWeave buys and operates that infrastructure, then sells access to computing capacity.

Both can benefit from growing AI demand.

But their capital requirements, margins, financing risks and competitive advantages are very different.

Personally, I still find that distinction important.

NVIDIA benefits from AI infrastructure spending without having to operate every data center that uses its technology.

CoreWeave, on the other hand, takes on the cost and operational risk of deploying that infrastructure.

There's also $AMD, which continues competing for a larger share of the AI accelerator market.

And further down the supply chain, companies like $VRT, $GEV and $MU provide exposure to data center power infrastructure, cooling systems and advanced memory demand.

But again, being exposed to AI growth doesn't automatically mean a stock is attractively valued.

That's especially true after a sharp rally.

My Take: AI Demand Can Be Real, and the Investment Risks Can Be Real Too.

I think the AI infrastructure debate has become a little too binary.

Some investors believe we're still at the beginning of a massive investment cycle, so almost any AI-related spending is justified.

Others see the huge debt levels and capital expenditures and immediately call it a bubble.

I don't think either view tells the whole story.

AI demand can be genuinely strong while some AI infrastructure investments still turn out to be poor financial decisions.

Both things can happen at the same time.

Imagine a company that signs billions of dollars in customer contracts.

That's great.

But if fulfilling those contracts requires massive upfront investment, expensive debt and hardware that loses value faster than expected, the returns may still disappoint.

This is why I'm becoming less interested in who announces the largest AI spending plan.

I'm more interested in who can actually turn that spending into sustainable cash flow.

For CoreWeave specifically, there are a few things I'll be watching over the next several quarters.

Can it bring down its financing costs?

Will the expected $9 billion in customer prepayments arrive as planned?

Can it keep expanding capacity without putting too much additional pressure on its balance sheet?

And most importantly, can it deliver on management's 2028 profitability target?

Those questions matter more to me than whether the next quarter beats revenue estimates by a few percentage points.

As for CRWV's 24.7% rally, I see it as a strong vote of confidence in the company's demand outlook and potential financing improvements.

But I wouldn't take one big green day as proof that all the risks have disappeared.

The company still has a lot to prove.

And with interest rates, hardware costs and competitive dynamics constantly changing, the economics of AI infrastructure could look very different a few years from now.

I'm still bullish on the long-term demand for AI computing.

But that doesn't mean I'm automatically bullish on every company building data centers.

The next phase of the AI trade might be less about who can spend the most money and more about who can actually earn a decent return on that spending.

That's the part of this story I'll be paying closest attention to.

Poll: After CRWV's 25% Rally, Where Would You Put Your Money?

A. AI Cloud Providers — CRWV, NBIS, APLD. Demand is still growing fast.

B. AI Chipmakers — NVDA, AMD. Prefer the suppliers over the operators.

C. Power & Memory — VRT, GEV, MU. Looking further down the infrastructure chain.

D. Staying on the Sidelines — Too much debt, too much spending and too much uncertainty.

Would you rather own $NVDA or $CRWV at this point?

And do you think CoreWeave's rally marks the beginning of another leg higher for AI infrastructure stocks, or just a sharp rebound after a difficult stretch?

Curious what everyone thinks.

#CRWV #CoreWeave #NVDA #NBIS #APLD #AMD #AI #AIInfrastructure #AIStocks #Investing

Disclaimer: For informational and discussion purposes only, not investment advice. CoreWeave's profitability targets, customer prepayments and capital expenditure guidance are forward-looking and subject to change. AI infrastructure investments involve significant financing, execution, valuation and market risks.

Sources: CoreWeave Q3 2026 earnings release (Nov. 10, 2026); Reuters (Nov. 10–11, 2026).