Nebius Cloud 3.6: Software Polish Doesn't Solve the Capacity Riddle

What happened

Nebius announced Cloud 3.6, featuring an AI DevOps agent (Echo) and deeper SkyPilot integration, aiming to reduce operational friction and attract enterprise customers with enhanced governance and security. While this software update burnishes Nebius's platform story, the DeepValue master report underscores that the stock's valuation already capitalizes on aggressive growth assumptions. The critical investment variable remains Nebius's ability to convert its massive contracted demand – $21.3B in RPO – into energized capacity fast enough to support a $7B-$9B ARR target by end-2026. The report warns that the Q3 2026 connected-power disclosure is the true proof point; if it fails to show a credible step-up, the thesis breaks. This product announcement is a welcome incremental improvement but does nothing to de-risk the core execution challenge.

Implication

Cloud 3.6 may help Nebius retain enterprise customers and improve unit economics over time, but it does not alter the fundamental dependency on energizing thousands of megawatts of power on schedule. The stock at ~$220 already prices in robust growth; any slippage in capacity ramp – especially if Q3 2026 connected power misses 350 MW – could trigger a severe re-rating. Investors should remain on the sidelines until physical execution is evidenced. The software layer is a potential differentiator in 2027+ but offers no near-term margin of safety.

Thesis delta

The product upgrade is a positive but incremental development; it does not shift the thesis that Nebius's near-term value hinges on converting contracted demand into live power, not on software features. The market's focus should remain squarely on Q3 2026 connected-power milestones, which are the only metrics that can validate the premium valuation.