Better Datacenter Stock: CoreWeave (CRWV) or Nebius (NBIS)?

CoreWeave and Nebius both posted explosive AI infrastructure growth, with CoreWeave revenue up 112% year over year to $2.1 billion in Q1 2026 and Nebius revenue up 684% to $399 million. The article argues Nebius is the better buy because of faster expected revenue growth, lower leverage at a 2.1 debt-to-equity ratio versus CoreWeave's 10.7, and a path to GAAP profitability in 2028. Despite strong operating momentum, both remain loss-making and the piece is primarily comparative analyst commentary rather than a new corporate event.

Analysis

The market is treating both names as “AI picks-and-shovels,” but the economic quality of that exposure is diverging. NBIS is better positioned for the next leg of AI spend because inference and managed services tend to be stickier, less capex-light customer relationships than pure GPU rental, which should compress churn and improve pricing power as workloads move from training bursts to persistent production use. That makes NBIS the cleaner beneficiary of enterprise AI adoption even if headline revenue growth starts to decelerate.

CRWV’s setup is more fragile than the multiple suggests: a backlog-heavy story financed by balance-sheet expansion can look great until GPU supply, power delivery, or customer concentration creates timing slippage. The key second-order risk is that large hyperscale buyers increasingly optimize across vendors, so incremental volume may come with lower economic margin than the backlog implies. In other words, growth can remain fast while equity value creation stalls if each dollar of revenue requires disproportionate debt and depreciation.

The contrarian view is that NBIS may still be underappreciated despite trading rich on sales, because the market is likely underpricing the transition from “capacity provider” to “workflow owner.” If its full-stack model captures more software attach and utilization smoothing, operating leverage could inflect faster than consensus expects over the next 6-12 months. By contrast, CRWV’s valuation discount may be a trap if leverage keeps rising faster than gross profit, especially in a risk-off tape where investors re-rate unprofitable infrastructure names on funding risk rather than growth.

Near term, the trade is not about revenue beats; it is about which balance sheet can fund the next 18-24 months of AI buildout without forcing dilutive capital raises or expensive debt. NBIS looks like the superior long-duration compounding asset, while CRWV is a higher-beta financing trade that works only if execution and capital markets stay open.

Key Decisions for Investors

• Long NBIS / short CRWV as a relative-value pair for the next 3-6 months; thesis is superior leverage profile, better mix, and lower financing risk. Risk: CRWV backlog converts faster than expected and NBIS multiple compresses on any growth slowdown.
• Buy NBIS on pullbacks rather than chasing strength; target 8-12 week entries after broad AI weakness. Risk/reward improves if the stock de-rates toward high-teens sales multiples while forward EBITDA inflects.
• Avoid new long CRWV equity exposure until debt growth decelerates for at least one quarter; this is a capital structure trade, not a pure growth trade. If owning it, prefer defined-risk upside via call spreads instead of stock.
• Hedge AI infrastructure exposure with a short basket of the more levered datacenter names if funding conditions tighten; CRWV should be the first name to underperform in a liquidity shock.
• Use NVDA as the cleaner upstream hedge for both names rather than owning either operator outright; chip demand is monetized earlier in the cycle and carries less balance-sheet risk.