What Jensen Huang Said That Nobody Noticed

Nvidia reported $81.6 billion in revenue, up 85% year over year, and guided Q2 revenue to $91 billion, while CEO Jensen Huang said demand has "gone parabolic" as agentic AI expands beyond hyperscalers. Luke Lango upgraded CoreWeave (CRWV) from Hold to Buy, citing Jensen's endorsement, Anthropic-related demand, and a price below his $150 buy-up-to level. The article also highlights bearish insider-selling signals in Palantir, a bullish rotation into QUBT, and Mitek Systems (MITK) as a cybersecurity/fraud play on the rise of AI agents.

Analysis

The setup here is a classic capital-rotation regime, not an isolated stock-picking story. The important signal is that AI spend is broadening from a handful of model-heavy hyperscalers into the infrastructure and pick-and-shovel layer, which should extend the cycle for the beneficiaries of power, compute, networking, and identity/fraud tooling. That broadening tends to support second-derivative winners longer than the headline AI names because the revenue base becomes more diversified and less dependent on a single buyer cohort.

PLTR’s risk is less about fundamentals slowing tomorrow and more about multiple compression once insider behavior becomes a durable narrative. When a premium-rated software name loses trust with incremental buyers, it can underperform the index for months even if operating results remain strong, because the market starts discounting governance and opportunity cost rather than growth alone. The downside here is asymmetric: if the AI infrastructure trade stays hot while PLTR merely treads water, relative underperformance can widen quickly.

CRWV and NVDA sit on the clearest second-order opportunity: if enterprise/sovereign AI demand is really outgrowing hyperscale, the market may be underpricing the duration of the buildout and the pricing power of the infrastructure enablers. The contrarian risk is that this becomes a crowded trade fast; any capex pause, customer concentration concern, or financing squeeze in the neocloud ecosystem would hit high-beta names first. MITK is more interesting as a stealth beneficiary: agentic AI increases the attack surface, and fraud-prevention vendors can get a secular tailwind without needing the full AI hype premium.

QUBT and RGTI remain the most speculative expression of the rotation. The move can continue mechanically if flows chase scarcity and narrative torque, but these names are vulnerable to abrupt reversals if investors demand proof of commercialization rather than optionality. In other words: the trade works as long as capital is rotating from crowded winners into early-stage infrastructure; it fails if the market switches from story to cash-flow discipline.

Key Decisions for Investors

• Add CRWV on pullbacks for a 1-3 month momentum trade; best risk/reward is a tactical long below recent resistance with a stop 8-10% below entry, targeting continuation if neocloud demand stays broadening.
• Stay short or underweight PLTR versus a basket of AI infrastructure names for 2-6 months; pair trade long NVDA/CRWV against short PLTR to isolate the rotation away from high-multiple software into infrastructure.
• Initiate a starter long in MITK for a 3-12 month thesis on AI-driven fraud and identity verification; risk/reward improves if the stock remains below broad AI-complex valuation multiples while revenue tailwinds accumulate.
• Trade QUBT/RGTI only as event-driven momentum longs, not core positions; use small size and tight stops because upside can be explosive, but drawdowns can erase gains in days if flow momentum fades.
• Use NVDA strength to own the infrastructure basket rather than the single name; if the data-center buildout is broadening, the cleaner expression is long NVDA plus selected second-tier beneficiaries, with a 6-12 month horizon.