CoreWeave Surges Over 18% Pre-Market; Super Micro and Nvidia Rally in Tandem
Nashnova编辑部
CoreWeave's Q2 revenue doubled year-over-year to $2.58 billion with margins beating expectations, sending shares up over 18% pre-market; Super Micro's guidance blew past estimates, and Nvidia's CEO personally addressed AI debt concerns — three threads converging on the critical test for AI infrastructure investment logic.
What exactly made CoreWeave's quarter so strong?
Revenue hit $2.58 billion, up 112% year-over-year, topping the Wall Street consensus of $2.56 billion. Adjusted loss per share came in at $1.03, better than the expected $1.20 loss.
The headline number: adjusted operating margin reached 5%, nearly double the FactSet consensus of 2.7%. This means → CoreWeave is no longer just burning cash for scale — it is starting to prove the business can make money.
Total backlog stands at $104 billion, plus over $25 billion in new customer commitments signed early this quarter — roughly $129 billion combined. In plain terms = years of future revenue are already locked in.
Live data-center power capacity reached 1.5 GW, adding 500 MW quarter-over-quarter; total contracted capacity expanded to 4.2 GW. The company raised full-year revenue guidance and issued a stronger-than-expected Q3 outlook.
Why did Super Micro's guidance catch the market off guard?
Super Micro guided Q1 adjusted EPS at $1.01–$1.10, far above the LSEG consensus of $0.76. Revenue guidance of $14.5–$15.5 billion topped the expected $11.68 billion by roughly 30%.
This means → the market had severely underestimated near-term AI server demand. Super Micro is one of Nvidia's largest GPU system assemblers — its guidance directly reflects downstream purchasing pace.
Last quarter's results also beat expectations. Combined with the raised outlook, shares jumped over 7.5% pre-market.
Why did Nvidia's CEO step in personally on the debt question?
A key drag on Nvidia's stock recently: concern over debt sustainability in the AI buildout. Alphabet, Amazon, Meta, and Microsoft have borrowed heavily for AI infrastructure, and their credit default swap (CDS — essentially the price of insuring against default) costs have been climbing.
Nvidia is directly exposed too — it provides financing guarantees to chip buyers and is reportedly negotiating up to $250 billion in financing with OpenAI. Its own CDS spread has risen to a year-high.
On Monday Nvidia announced a joint AI infrastructure financing platform with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR, targeting over $500 billion in capital mobilization. Rather than calming markets, the move sparked fresh fears of a "circular financing" spiral.
CEO Jensen Huang responded directly on X: Nvidia may offer "residual value support," but capped at 25% of any single project, evaluated case by case. This means → Nvidia drew a clear ceiling on its risk exposure, signaling to the bond market that it will not backstop everything indefinitely.
How does Wall Street see CoreWeave? Where is the disagreement?
The bull camp: Evercore holds outperform at $160; Deutsche Bank holds buy at $150, citing "consistent strong execution and early profitability improvement"; Citi holds buy at $142, noting SKU prices up roughly 25%; Bank of America holds buy at $140, projecting Q3 operating margin at 6.6% and Q4 at 14.6%.
Even the bear raised the target: Bernstein analyst Madison Rezaei lifted the price target from $67 to $74 but kept an underperform rating, calling it "CoreWeave's strongest quarter to date" while adding "this is still not enough to make us believe in the new cloud model dream."
JPMorgan raised its target from $110 to $120, held neutral, and said the quarter's highlights point more to the long-term story than near-term. This reflects that even neutral-stance firms acknowledge the data is improving — they just question how long "improving" lasts.
What does the market watch next?
Two key checkpoints ahead: first, whether CoreWeave's margins can keep climbing as it scales — the road from 5% to Bank of America's projected Q4 14.6% is long; second, whether Nvidia's actual financing exposure stays within the bounds Jensen Huang described.
In plain terms = the AI infrastructure narrative is shifting from "is demand big enough?" to "is the money safe enough?" Revenue growth has proven the demand case; now the market needs to see margins and risk controls.
Content is for reference only, not financial advice.