The Nasdaq Is on Fire. Here Are the 2 Best Artificial Intelligence (AI) Growth Stocks That Still Look Cheap.

The article argues that Nvidia still looks inexpensive at a 41x P/E despite fiscal 2026 revenue of $216 billion (+65%) and net income of $120 billion (+65%), while CoreWeave trades at under 10x sales after 2025 revenue jumped to $5.1 billion (+168%). CoreWeave’s valuation is held back by $21.4 billion of debt versus $3.3 billion of book value and a $1.22 billion loss, but a $66.8 billion backlog underscores strong AI demand. Overall tone is constructive on AI stocks, but the piece is primarily valuation commentary rather than new hard catalysts.

Analysis

The market is implicitly rewarding scarcity of credible AI capacity, but it is pricing the balance sheets very differently across the stack. NVDA remains the cleanest way to own AI spend because it is effectively monetizing everyone else’s capex binge, while CRWV is a leveraged call option on that same spend with financing risk layered on top. That creates a classic second-order setup: if AI demand stays hot, the vendor selling picks-and-shovels keeps compounding, but the infrastructure operator must keep refinancing to stay in the game.

The key distinction is duration of catalyst. NVDA can keep surprising over multiple quarters because earnings power is already visible and cash generation is self-funding, which should attract incremental capital from investors who want AI exposure without binary solvency risk. CRWV, by contrast, is a months-to-years story: the stock can re-rate sharply only if backlog converts into profitable utilization faster than debt service and depreciation consume operating gains. Until then, the equity behaves more like a volatility vehicle tied to capital markets access than to simple revenue growth.

The consensus is likely underestimating how much the market will compress the “AI winners” trade into two buckets: quality compounders versus capital-intensive enablers. That should widen the valuation spread between NVDA and other AI beneficiaries with weaker free-cash-flow visibility, while also creating periodic squeeze risk in CRWV if short interest is crowded and execution improves even modestly. The biggest hidden risk is not demand failure; it is a funding pause, where rising rates or tighter credit conditions force a reset in the economics of AI infrastructure expansion.

From a positioning perspective, this is a better relative-value setup than a directional one. Use NVDA as the lower-risk expression of continued AI capex, and treat CRWV as a tactical trade only if you have conviction that utilization inflects before debt maturities become a narrative headwind. The asymmetry is attractive in CRWV, but only if you can survive the path dependency.

Key Decisions for Investors

• Long NVDA / short CRWV as a 3-6 month relative-value pair: own the self-funding AI compounder against the balance-sheet levered AI infrastructure story; thesis works if AI capex stays strong but credit conditions remain tighter than equity enthusiasm.
• Buy NVDA call spreads out 6-9 months to express continued AI spend without paying for runaway multiple expansion; risk/reward favors moderate upside rather than another parabolic rerating.
• Avoid initiating fresh long CRWV ahead of the next financing or earnings window unless you have conviction on utilization ramp; the equity is highly sensitive to any sign that backlog is converting slower than debt amortization.
• If already long CRWV, pair it with a short in a more stable AI beneficiary to neutralize market beta and isolate execution alpha; otherwise size as a high-volatility satellite position, not core exposure.
• For investors seeking AI exposure with lower drawdown risk, shift marginal capital toward NVDA rather than neocloud names; the better risk-adjusted return likely comes from the company capturing the ecosystem toll, not the one funding the ecosystem buildout.