2 Artificial Intelligence (AI) Infrastructure Stocks Outgrowing Nvidia

Nebius reported Q1 revenue growth of 684% year over year, while CoreWeave's Q1 revenue rose 112% to $2.1 billion and its backlog reached nearly $100 billion, with more than a third expected to convert to revenue over the next two years. Wall Street expects Nebius revenue to rise 551% this year and 224% next year, and CoreWeave revenue to grow 147% in 2026 and 97% in 2027. The article is bullish on both names, but highlights meaningful funding and dilution risk because neither company is profitable.

Analysis

The market is treating NBIS and CRWV as pure “GPU landlords,” but the more important second-order effect is that both are becoming financing conduits for Nvidia demand. That matters because the more backlog converts, the more capital intensity shifts upstream into suppliers, power, and networking, while equity holders absorb dilution and balance-sheet risk downstream. In the near term, the winners are clearly Nvidia’s ecosystem vendors; the eventual losers are lower-quality neoclouds that try to scale without locked-in capital or cheap power.

The key question is not growth, but durability of spread economics after the initial capacity build-out. If utilization normalizes even modestly below peak, the market will compress revenue multiples fast because these businesses have little operating leverage protection once depreciation, interest, and replacement cycles are layered in. That creates a sharp asymmetry: the next 6-12 months can look exceptional on top-line momentum, while the 18-36 month outcome depends on whether contracted demand outruns capacity additions enough to preserve margins.

Consensus is likely underpricing how much of this demand is already “pre-financed” by customers anxious to secure supply, which can keep numbers elevated longer than bears expect. But consensus may also be overestimating the scarcity premium: if hyperscalers, colo providers, or hyperscale-adjacent vendors accelerate capex, the market can quickly decide these are not software-like compounders but highly levered infrastructure trades. The real tell will be whether backlog translates into cash flow before dilution compounds.

NVDA is the cleaner expression of the theme because it captures the infrastructure build regardless of which neocloud wins share. NBIS is the higher-beta operating leverage story, but CRWV has the more visible backlog conversion path; both are vulnerable if equity issuance becomes the default funding tool and per-share economics lag headline growth. The trade here is less about “AI demand” and more about who can convert demand into capacity without destroying incremental returns on capital.

Key Decisions for Investors

• Long NVDA vs. long/short basket NBIS + CRWV on a 3-6 month horizon: keep exposure to AI capex while avoiding dilution and depreciation risk embedded in neocloud equities.
• If seeking higher beta, buy NBIS only on pullbacks and size it as a tactical 4-8 week trade, not a core hold; upside is larger if capacity comes online on schedule, but dilution risk makes the risk/reward fragile.
• Prefer CRWV over NBIS for a backlog-to-revenue catalyst over the next 2-4 quarters, but hedge with out-of-the-money puts into financing windows or large capex announcements.
• Watch for any equity issuance or convert deal from either name; if announced, trim 25-50% immediately because the market is likely to punish per-share economics before revenue upside is fully recognized.
• For a cleaner thematic trade, pair long NVDA with short a broad basket of AI infrastructure names that rely on external funding, since the market is likely to continue paying up for the toll collector rather than the highly leveraged capacity owner.