Competitive Comparison
CoreWeave vs. Nebius vs. Applied Digital vs. Oracle Cloud
Four ways to bet on GPU-cloud demand outstripping supply, from two direct neocloud holdings to a speculative early-stage name and a legacy enterprise vendor reinventing itself as AI infrastructure.
| Metric | CoreWeave (CRWV) | Nebius (NBIS) | Applied Digital (APLD) | Oracle (ORCL) |
|---|---|---|---|---|
| EV/EBITDA | 26.76x | 36.44x | n/a (adj. EBITDA $44.1M fiscal Q3) | n/a (5yr EBITDA growth 15.0%) |
| PEG ratio | N/A (unprofitable) | 7.79 | N/A | 1.23 |
| Capex (2026) | $31–35B (guidance) | $20–25B (guidance) | −$1.77B (TTM) | −$55.66B (TTM) |
PEG ratio, compared
Why CoreWeave and Nebius, sized small, not Applied Digital or Oracle
CoreWeave and Nebius are both neutral GPU-cloud providers renting compute rather than picking which AI model or application wins — the purest version of this bet, which is exactly why both are unprofitable and trading on capex guidance rather than earnings. Applied Digital is the same trade at an earlier, smaller, more speculative stage; I watch it rather than hold it because its EBITDA base is still too thin to size confidently. Oracle is the opposite end of the spectrum: an established, profitable legacy enterprise vendor spending nearly $56 billion in trailing capex to reinvent itself as AI infrastructure — a real trend worth watching, but its PEG of 1.23 and diversified legacy database business make it a fundamentally different, lower-torque bet than the pure-play neoclouds I actually hold.