To The Moon

1. Has NBIS become “too expensive,” à la CoreWeave — or is there still room to run?

Positives for Nebius

Massive Microsoft deal: NBIS just locked in a $17.4 bn five-year GPU infrastructure agreement, with a potential $2 bn extension through 2031. 

De-risking their buildout: That deal gives Nebius strong visibility into revenue, and Microsoft cash-flow or committed payments can underwrite capital expenditures and debt financing. Analysts see this as significantly reducing execution risk. 

Strong “neocloud” positioning: Nebius is carving out a niche as a vertically integrated AI cloud provider using Nvidia-powered GPU infrastructure, which is exactly what hyperscalers are scrambling for. 

Room for further contracts: If this Microsoft deal is Phase 1, investors and analysts seem to believe Nebius is now credibly positioned to expand contracts with other large tech or frontier AI labs. 

Risks and caution flags

Valuation jump: Even with a transformational deal, the stock reaction (50-70 %) means Nebius is now priced for very high growth and success. Any delay, technology setback, or competition could hit sentiment hard. 

Execution risk persists: Just because Microsoft has signed doesn’t mean build-out, deployment, and profitable scaling are trivial. Data centers, GPU procurement cycles, supply chain risks, and operational costs remain significant — even if contractually supported.

Competitive pressure: CoreWeave is already a major player with large contracts and is backed by Nvidia. Big providers (AWS, Google Cloud, Microsoft’s own Azure) continue to expand their AI cloud capabilities. Margins and pricing power may erode.

Heavy CAPEX and leverage: Even with prepayments or committed contracts, the need to build fast and scale could lead to leverage, which becomes problematic if demand or utilization weakens.

Cyclicality and “AI hype”: Markets have a habit of swinging wildly in this space. Stretch valuations based on growth narratives can reverse rapidly if investor sentiment turns—or if macroeconomic pressures tighten financing availability.

Verdict

Nebius has made a credible leap from speculative infrastructure play to a contract-backed AI neocloud with real scale potential. That Microsoft deal is a game-changer. But the stock is now priced with the expectation that Nebius executes nearly flawlessly, scales rapidly, wins more big clients, and avoids falling behind technologically or financially.

If you believe those things will happen, there is more upside — especially if Nebius can compound long-term growth and add multiple large contracts. But if you’re skeptical of perfect execution or cautious about falling into AI “growth trap” valuations, the current price demands very high conviction.

In short: not necessarily “too expensive,” but expensive enough that the margin for error is small.

---

2. Is “following Nvidia” a good play for picking growth stocks?

This is a tempting heuristic — Nvidia is the dominant supplier of AI-GPUs, and many downstream beneficiaries (cloud providers, AI infrastructure firms, software vendors, etc.) ride on that tidal wave. But there are a few caveats to keep in mind:

Pros of the Nvidia-follow strategy

1. Tidal AI tailwinds: Nvidia GPU supply shortages, rapid AI model scaling, and increasing compute intensity means companies that provide GPU compute (or build services around it) have a strong structural tailwind.

2. Ecosystem leverage: Companies that partner with Nvidia or optimize for Nvidia hardware can benefit from spill-over demand, early access to new hardware, and sticky developer ecosystems.

3. Market sentiment tends to reward AI-adjacent names: During bull phases, those “AI cloud / neocloud / GPU compute” names can outperform broader markets, especially if Nvidia is booming.

But there are downsides

1. Not all Nvidia-based plays are created equal: Just because a company uses Nvidia GPUs doesn’t mean it has defensible differentiation, good unit economics, or a path to profitability. Execution, capital structure, market position, pricing power, and customer concentration still matter a lot.

2. Valuation risk is real: Many Nvidia-leveraged companies trade at very high multiples on forward growth assumptions. If growth disappoints, or if Nvidia’s next generation hardware makes prior infrastructure obsolete too quickly, valuations can compress severely.

3. Competition and vertical integration: Hyperscalers, chip designers, and legacy cloud providers can decide to internalize parts of the infrastructure stack. If companies are overly dependent on “selling GPU compute” without differentiation, they may get squeezed.

4. Overbetting on a single tech trend: AI and GPU compute are booming now, but markets rotate. If Nvidia suddenly loses dominance (e.g., through competition, regulation, or technical disruption), or if AI compute demand plateaus, “Nvidia-follow” stocks could get hit hard.

5. Sentiment swings amplify risk: AI/ML is a narrative-heavy space. Investor sentiment can swing wildly — sometimes more on headlines or hype than fundamentals.

Recommendation

A smarter way to “follow Nvidia” might be:

Focus on companies with diversified moats or services around GPU compute, not just raw hardware rental.

Look for contracted/committed revenue (like Nebius’ Microsoft deal) or recurring revenue models that help de-risk execution.

Pay attention to unit economics, gross margins, customer concentration, debt levels, and CAPEX intensity — not just headline growth or market buzz.

Be disciplined about valuation: assume that not all Nvidia-adjacent firms will survive or win. Build margin-of-error into your investment thesis, or use position sizing accordingly.