The AI Stock That's Been Outperforming Nvidia Since January

Nebius reported first-quarter revenue of $399 million, up 684% year over year, while AI revenue rose 841% to $390 million and its annualized run-rate reached $1.9 billion. Management raised 2026 revenue guidance to $3.0 billion-$3.4 billion and sees nearly 40% adjusted EBITDA margins, supported by $27 billion of Meta deals, a $17.4 billion Microsoft contract, and Nvidia's $2 billion strategic investment. The main offset is a higher 2026 capex plan of $20 billion-$25 billion, which increases financing and dilution risk despite $9.3 billion in cash.

Analysis

NBIS is transitioning from a pure capacity vendor to a financing-backed industrial platform, which changes the equity story materially: the long-duration contracts are no longer just demand proof, they are collateral. That should compress execution risk and lower its marginal cost of capital over the next 6-18 months, because hyperscaler counterparty quality can support project-level leverage, vendor financing, and prepayment structures. The second-order winner is NVDA, which benefits from a more diversified buyer base for high-end accelerators, while the real pressure point is smaller AI infra players that lack either scale, financing access, or anchor tenants.

The market is likely underestimating how quickly scarcity can migrate from compute to power, land, and interconnects. If NBIS actually executes on the capex ramp, the bottleneck shifts upstream into EPC contractors, electrical equipment, liquid cooling, and grid-adjacent suppliers, which can get a multi-quarter demand tailwind even if NBIS equity remains volatile. That makes this less a “one stock” story and more a capital formation cycle for the AI infrastructure stack.

The main risk is that the visible revenue is being bought with a very large capital burn ahead of cash conversion, so the equity outcome will depend on whether new capacity comes online on schedule and at target utilization. Any delay in permitting, grid hookups, or GPU delivery would hit both margins and financing optionality, and that risk is highest over the next 2-4 quarters rather than the next few days. A second-order concern is that the market may already be pricing in perfection; if management leans on equity issuance, the multiple can compress even while the business is improving.

The contrarian view is that consensus is focusing too much on reported growth and not enough on the implied scarcity value of contracted AI compute. If these deals are structured with meaningful prepayments or financing support, the equity may actually be de-risking faster than headline capex suggests. Conversely, if the contracts are mostly volume commitments without durable take-or-pay economics, the market is paying growth multiple prices for a business that still has infrastructure-style financing risk.

Key Decisions for Investors

• Long NBIS into pullbacks over the next 2-6 weeks; treat the name as a high-beta financing/security trade with upside if the market starts capitalizing contracted revenue at a lower discount rate. Risk/reward is attractive only if you size for dilution and 20-30% drawdowns.
• Pair long NVDA / short a basket of weaker AI infra names over 3-6 months; NVDA benefits from the buildout regardless of which operator wins, while smaller operators face funding and execution asymmetry.
• Initiate a tactical long in power-grid and data-center electrical equipment suppliers for 3-12 months; NBIS-style capex ramps should pull through into switchgear, transformers, and cooling, often before the market fully credits the end customer.
• Avoid chasing NBIS after sharp gap-ups; use call spreads instead of outright equity if expressing upside, because financing dilution can cap near-term multiple expansion even with strong operating prints.
• If NBIS announces incremental equity issuance or slower-than-expected capacity activation, fade the stock on the first bounce; that would likely reset the market from 'growth' to 'funding risk' and pressure the multiple quickly.