The Nasdaq hit an all-time high in May 2026. The round-trip happened in five sessions. That kind of volatility at record levels isn’t unusual for this market — it’s been the pattern throughout 2025 and into 2026 — but it’s worth understanding what’s driving it, because the structural setup has gotten more complicated.
The underlying earnings story is genuinely strong. Of 440 S&P 500 companies that have reported Q1 results, 83% beat estimates. AI-related infrastructure spending is real and accelerating — hyperscaler capital expenditure commitments remain among the highest in history, and the demand signal from the semiconductor complex has been consistent for several quarters. Broadcom (AVGO), Analog Devices (ADI), and others in the ecosystem are posting results and guiding higher in ways that don’t look like a cycle peak.
The complication is concentration. NVIDIA (NVDA) has become the single largest contributor to U.S. earnings growth in 2026. When one company functions as both the largest index weight and the primary engine of earnings expansion, the index’s behavior starts to look more like a single-stock risk than a diversified market. Any quarterly miss on revenue, gross margins, or forward guidance — in an environment where expectations have been set exceptionally high by prior beats — would likely cascade across the entire semiconductor complex, not stay contained.
That’s not a reason to exit the AI trade. It’s a reason to think carefully about how exposure is constructed. Investors fully concentrated in the mega-cap AI names are taking on more idiosyncratic risk than the broad index label implies. The more interesting positioning, arguably, is in the second-order infrastructure layer — power management, custom silicon, data center cooling, grid buildout — where the same capex tailwind applies but valuations are less extended, and the earnings risk is spread across a wider set of companies. That part of the trade has lagged the headline names and may have more room to run with less binary risk attached.
