Is Nebius a Buy After Q2 Earnings? Revenue Beat, Burry's Short Still Live

TL;DR

• Nebius beat and the stock ripped. Q2 2026 revenue hit $582.3 million, up 454% year over year and roughly $8 million above the Street's ~$574 million consensus. Shares traded as high as $234.65 around 10:50am ET Wednesday, up about 21% on the day against Tuesday's $193.23 close, though the gain narrowed and widened repeatedly through the morning.
• Adjusted EBITDA turned solidly positive: $236.2 million, against a $21.0 million loss a year ago. Adjusted net loss narrowed to $33.2 million.
• The GAAP net loss of $190.4 million looks worse than it is. A year ago the GAAP line showed $502.5 million of income, and Q1 2026 alone showed $621 million, both flattered by large non-cash gains on revaluing equity stakes. Strip those swings out and the operating trend, revenue and adjusted EBITDA, moved the right direction both quarters.
• Full-year guidance was reaffirmed, not raised: group revenue of $3-3.4 billion, annualized run-rate revenue of $7-9 billion by year-end, adjusted EBITDA margin near 40%, and capex of $20-25 billion. ARR itself hit $3 billion at the end of June, up 56% from $1.9 billion at the end of March.
• Burry's disclosed short at $211.77 is underwater at Wednesday's levels. That does not settle the argument. It raises the stakes on it: H1 capex already ran $8.13 billion against $981.3 million of H1 revenue, and hitting the $20-25 billion full-year capex guide means the back half spends more than the front half by a wide margin.

Is Nebius a Buy After Earnings?

The demand side of this argument is now proven for a fourth straight quarter, and I don't think that's seriously in dispute anymore. Revenue beat, EBITDA swung positive, and the backlog behind it, more than $40 billion of contracted revenue including the $17.4 billion Microsoft GPU supply deal, a $2.9 billion Meta contract and the $1 billion-plus Reflection AI agreement signed in July, is real paper with named, investment-grade counterparties, not a hand-wavy pipeline.

What the print did not resolve is the funding side, which is the specific wound Burry's disclosed short is aimed at. Capex ran $5.66 billion in Q2 alone, more than ten times the revenue it's building capacity to serve, and the full-year guide implies an even heavier back half. A company can have unimpeachable demand and still be a bad stock if the capital it needs to serve that demand comes at a cost the equity doesn't like. Nebius hasn't answered that yet in a way that would make me chase a stock up 21% in a morning.

The CoreWeave earnings the night before asked the identical question, and the pattern across AI infrastructure names this earnings season keeps repeating: revenue and EBITDA are believable, the balance sheet is the part reporters and short-sellers are still fighting over. I'm not convinced a 21% pop on a print that reaffirmed rather than raised full-year guidance is the entry point. The disclosed short is still on the tape and hasn't been covered, which tells me the fight isn't over just because the quarter was good.

What Actually Changed Since Tuesday

• Capex against contracted revenue, answered directionally, not fully. Q2 capex of $5.66 billion is a real number now, not a guess, and it lines up with the reaffirmed $20-25 billion full-year guide. What it costs to finance that spend is the piece still missing from Wednesday's release.
• Debt terms, partially disclosed. Nebius closed its first secured credit facility, roughly $775 million, on July 17, ahead of this print. That's a data point on funding cost, but it's a fraction of the $20-25 billion capex program, so it doesn't settle the covenant question Burry is short against.
• Utilisation, trending the right way. ARR of $3 billion at quarter-end, up 56% sequentially, is the clearest sign that contracted capacity is converting into running revenue rather than sitting idle.

Two of three questions moved in the bulls' favor. The financing-cost question, the one Burry actually named, is still open.

The One-Line Read

Nebius answered the demand question about as well as a company can, revenue up 454%, EBITDA positive, a reaffirmed $3-3.4 billion guide, and the market paid for it with a 21% pop; the funding math behind a $20-25 billion capex year is still unresolved, Burry's short is still live, and chasing the print is a worse trade than waiting for the next test of that thesis.