3 AI Stocks Crushing Nvidia This Year

CoreWeave is up more than 65% in 2026, Nebius more than 80%, and Applied Digital 37%, reflecting strong investor appetite for AI infrastructure names. Wall Street expects CoreWeave revenue growth of 143% in 2026 and 89% in 2027, while Nebius is projected to grow 523% in 2026 and 206% in 2027; Applied Digital reported 139% revenue growth in its most recent quarter. The article is bullish on all three stocks versus Nvidia, but notes they remain unprofitable and reliant on debt-funded expansion.

Analysis

The market is still in the early phase of a second-order AI infrastructure trade: not the model makers, but the picks-and-shovels that monetize scarcity in compute, power, and deployed capacity. The relative strength in CRWV and NBIS suggests investors are willing to pay for any business that can turn hyperscaler demand into contracted revenue, but the real edge is in who has pricing power versus who is merely a financing vehicle for growth. That matters because the winners in this phase are the names with the shortest path to utilization ramp and the cleanest funding access, while the losers will be the operators that need perpetual capital raises before the market re-prices their assets.

The most underappreciated risk is not AI demand fading; it is supply normalization. If GPU availability, power interconnects, or data-center build-outs loosen even modestly over the next 6-12 months, the premium multiple on incremental capacity can compress fast, especially for unprofitable operators whose equity value is being capitalized on peak scarcity rather than durable margins. NBIS looks best positioned on a relative basis because the market is paying for faster growth and a more software-like full-stack model, but that also makes it the most vulnerable to any guide-down or delay in customer deployment cadence.

APLD is the cleaner “infrastructure landlord” expression, but it is also the most levered to execution on power delivery, permitting, and tenant fill rates. If quarters continue to show rapid top-line expansion, the market may keep rewarding it as an asset-light proxy for AI demand; if not, the debt burden becomes the main story. The contrarian point is that NVDA may actually be the better risk-adjusted way to own this theme: it monetizes the entire ecosystem, has real earnings power today, and is less dependent on the capital markets staying open for its customers.

From a 1-3 month horizon, this is a momentum trade; over 6-18 months, it becomes a balance-sheet and cash-flow trade. The biggest reversal catalyst would be one or two weak quarters showing slower contract conversion, higher financing costs, or capex inflation that outpaces revenue acceleration. In that scenario, the high-beta infrastructure names could de-rate sharply even if AI demand itself remains strong.

Key Decisions for Investors

• Long NBIS vs. CRWV on a 3-6 month horizon: NBIS has higher growth optionality, but CRWV has a more obvious utilization story; the pair captures relative execution without taking full sector beta. Add on pullbacks of 8-10% and reduce if forward growth visibility slips.
• Buy NVDA as the lower-volatility AI infrastructure anchor for the next 6-12 months; it offers participation in the same capex cycle with materially less financing risk than the unprofitable pure plays. Use it as the core long and not as the tradeable momentum leg.
• Short a basket of the weakest AI-capex enablers on any post-earnings strength if revenue growth comes with rising leverage and worsening cash burn. The highest-risk setup is a name that beats on top line but needs another equity/debt raise within 2 quarters.
• For APLD, treat it as a tactical long only into evidence of tenant absorption and financing clarity over the next 1-2 quarters. If gross leverage continues rising faster than contracted revenue, fade rallies rather than chase them.
• Use call spreads instead of outright longs in CRWV/NBIS for 2-4 month exposure; the names are momentum-sensitive, but options cap downside if the market decides to de-rate unprofitable AI infrastructure before the next earnings cycle.