Michael Burry's Nebius Short Is Underwater After a 454% Revenue Quarter

Nebius reported Q2 revenue of $582.3M (+454% YoY), with AI cloud revenue growing 514% to $575M and a 50% EBITDA margin at the unit level (up from 45% in Q1). However, depreciation and amortization was $259.7M versus adjusted EBITDA of $236.2M, producing a GAAP loss of $190.4M, while capex of $5.7B in the quarter raises concerns about future profit recognition. Shares jumped ~34% to ~$259 after results, but the stock trades at ~20x guided FY revenue of ~$3.0B–$3.4B and the company raised ~$2.8B by selling 12.7M shares through June 30.

Analysis

Near term, the stock is still a squeeze candidate, but the quarter shifts the debate from "is demand real?" to "what is that demand worth after the hardware bill?" The beneficiaries are the upstream picks-and-shovels with cleaner economics — especially NVDA and, more broadly, power/cooling and networking vendors — while the vulnerable names are AI cloud platforms that need repeated capital recycling and lease-heavy buildouts to keep growth compounding. If NBIS keeps using customer prepayments to fund incremental capacity, the market may keep paying up, but that is a financing flywheel, not proof of durable earnings power.

The real bear case is a timing mismatch: revenue can compound for quarters while economic depreciation silently catches up. Extending server lives flattens GAAP pain today, but if true replacement cycles are closer to 2-3 years, reported margins should peak before cash economics do; that is usually when multiples compress hardest, not when growth is strongest. The next 1-3 month catalyst is less about another good print and more about whether management has to lean on more equity, longer lives, or higher capex to preserve the growth trajectory.

Contrarian view: the consensus may be overfocused on the accounting debate and underestimating how long the market can ignore it while utilization stays high and capacity is pre-sold. That argues against chasing an outright short here; the cleaner trade is relative value against names with less balance-sheet intensity. Falsifiers are simple: if NBIS sustains 40%+ EBITDA margins while capex intensity falls and prepayments remain high into the next quarter, the short thesis weakens materially; if not, the stock is still trading on hope plus dilution capacity, not normalized earnings power.

Key Decisions for Investors

• Do not chase an outright NBIS short after the squeeze; if entering bearish exposure, use 3-6 month NBIS put spreads on a failed push through the post-earnings high, with a stop if the stock holds above that level for a week.
• Relative-value: long NVDA / short NBIS for 1-3 months. Thesis is cleaner economics vs. capital-intensive growth; this should work if the market starts distinguishing chip demand from cloud accounting. Exit if NBIS guides up revenue without a commensurate capex slowdown.
• Set a watchlist alert for NBIS equity issuance, lease disclosures, or any change in useful-life assumptions over the next 1-2 quarters; those are the highest-conviction bearish catalysts for a re-rate.
• If you want AI exposure, prefer upstream infrastructure over AI-cloud operators: overweight NVDA on any NBIS-driven sector pullback, since the market may temporarily misprice the whole stack when the real issue is depreciation, not demand.