A growing number of AI-exposed mega-cap tech stocks have crossed into official bear market territory — defined as a 20%+ decline from recent highs — as the market reassesses the timeline and monetization potential of large AI infrastructure spending. The selloff has been indiscriminate, sweeping up both direct AI infrastructure plays and companies that merely rode the AI enthusiasm wave.

NVIDIA remains the clearest fundamental anchor in the group: FY2026 revenue hit $215.9B, up 65.5% year-over-year, with a 71.1% gross margin and $4.90 in diluted EPS — numbers that are difficult to square with a bear market designation unless the market is pricing in a peak-cycle risk. META, similarly, posted $201B in revenue (+22.2% YoY) and a 30.1% net margin, making its selloff look more sentiment-driven than fundamental. TSLA is the outlier — revenue actually declined 2.9% YoY to $94.8B, with a thin 4.1% net margin and $1.08 EPS, suggesting it was an AI-hype beneficiary rather than an AI-fundamental story.

The bull case for this basket hinges on NVDA's numbers being real: if data center capex from hyperscalers holds, the AI infrastructure cycle hasn't peaked and the selloff is a sentiment dislocation. META's advertising machine continues to compound even as it layered in AI costs, providing a more durable earnings base than pure-play AI names.

The bear case is a valuation compression story: even after the drawdown, multiples on NVDA and META remain elevated relative to historical norms, and any sign that hyperscaler capex is being trimmed — or that AI ROI is disappointing — could extend the derating further. TSLA adds idiosyncratic risk; its inclusion in 'AI plays' was always more narrative than fundamental, and its declining revenue and thin margins leave little cushion.

What to watch: hyperscaler earnings commentary on AI capex budgets, NVDA's next gross margin print, and whether breadth of the selloff narrows to the weaker names (TSLA-like) or continues to engulf the fundamental leaders (NVDA, META).