3 AI Infrastructure Stocks With 100% or Greater Growth Rates

CoreWeave, Nebius, and Applied Digital are all showing explosive AI-infrastructure demand, with revenue growth of 547% year over year to $228 million for Nebius, 110% to $1.6 billion for CoreWeave, and 139% to $127 million for Applied Digital. Nebius guided annual run-rate revenue to $7 billion-$9 billion by the end of 2026, while CoreWeave expects at least $28 billion of revenue recognition over the next 24 months and Applied Digital is benefiting from multiyear data-center contracts. The article is broadly bullish on AI infrastructure, though it is primarily commentary rather than a fresh company-specific catalyst.

Analysis

The real read-through is not simply “AI demand is strong,” but that the market is increasingly pricing a two-layer bottleneck: compute access at the GPU layer and physical capacity at the powered-shell layer. That tends to favor the assets with the longest contracted visibility, because in a scarcity regime the customer’s first priority is securing future capacity, not optimizing unit economics. In that setup, the neocloud names with differentiated capacity pipelines can keep monetizing even if near-term pricing power normalizes.

The second-order winner is the data-center landlord / developer model, because it gets paid on buildout optionality before the full AI workload stack is stabilized. But that also means this phase can become a financing story as much as an operating story: if capital markets tighten or if tenants delay ramp schedules, the equity value can de-rate quickly despite strong headline growth. The most important variable over the next 6-12 months is not growth rate alone, but conversion of backlog/announced demand into revenue recognized on schedule.

The contrarian risk is that consensus is extrapolating scarcity too far out the curve. These growth rates are likely peak-ish for the current cycle, and any moderation in hyperscaler capex, GPU supply normalization, or a pause in new AI model training demand would hit the highest-multiple names first. In particular, the market may be underestimating how quickly “capacity shortage” narratives can flip to “overbuild” narratives once multiple competitors complete similar data-center expansions.

Relative value looks more interesting than outright directionality: the best risk/reward is to own the names with contracted revenue plus visible capacity additions, while avoiding the ones where growth is mostly future promise. Shorting the strongest names outright is dangerous near term because the scarcity premium can persist for quarters; a better expression is to fade the most expensive name against the more operationally embedded landlord model if the spread widens further.

Key Decisions for Investors

• Long APLD vs short a basket of higher-multiple AI infrastructure proxies for 3-6 months: best if the market continues to reward visible capacity conversion over pure growth stories; cut if financing spreads widen or ramp delays surface.
• Stay long NBIS on pullbacks for 6-12 months: the full-stack positioning should capture a broader customer set than GPU-only peers, giving it the best shot at sustaining revenue quality through the next demand wave.
• Avoid chasing CRWV after strength; use it as a tactical trading vehicle only. Best entry is on a material pullback if backlog conversion remains intact, because the stock is most exposed to any perception of demand normalization.
• Pair trade: long APLD / short a basket of traditional data-center REITs if you believe AI-specific capacity scarcity will command a premium over generic colo exposure over the next 2 quarters.
• For options traders, consider selling short-dated puts on APLD or NBIS after spikes in implied volatility; the setup is fundamentally positive, but the stocks are likely to mean-revert violently on any capex or execution headline.