What's happening

On June 5, 2026, U.S. semiconductor stocks experienced a sector-wide selloff that erased approximately $1.3 trillion in combined market capitalization, with the Philadelphia Semiconductor Index declining more than 10% in a single session. The immediate trigger was Broadcom's fiscal Q2 2026 earnings report, in which the company posted adjusted earnings per share of $2.44 against analyst expectations of $2.40 — a modest beat — and reported AI semiconductor revenue of $10.8 billion, representing 143% year-over-year growth. However, Broadcom guided Q3 AI chip revenue at $16 billion, falling short of the $17.2 billion analysts had projected, and left its full-year 2026 AI chip sales forecast unchanged at $100 billion. CEO Hock Tan identified six core custom AI chip customers — including Anthropic, Google, Meta, and OpenAI — but the unchanged full-year outlook and below-consensus Q3 guidance drove broad selling pressure across the sector.

Broadcom shares fell between 12% and 15% across June 4–5, 2026, in the immediate aftermath of the earnings release. Nvidia, whose market capitalization had previously exceeded $5 trillion, declined 6% on June 5, erasing nearly $280 billion in market value and pushing the company's valuation below that threshold. Micron Technology and AMD each fell more than 9% in the same session, reflecting the extent to which the selloff extended beyond Broadcom to encompass the broader AI semiconductor ecosystem.

Why it matters for markets

The scale of the single-day valuation reset — approximately $1.3 trillion in erased market capitalization across the sector — underscores the degree to which AI-driven growth expectations had been embedded in semiconductor valuations heading into Broadcom's report. Broadcom's Q3 guidance of $16 billion in AI chip revenue, while representing substantial absolute growth, fell $1.2 billion below the $17.2 billion consensus estimate, illustrating how sensitive current valuations are to incremental guidance revisions rather than absolute performance levels. The company's decision to hold its full-year 2026 AI chip sales forecast at $100 billion, rather than raise it, compounded the market's reaction to the Q3 shortfall.

The breadth of the selloff highlights the interconnected valuation dynamics across AI semiconductor names. Nvidia, trading at a P/E of 31.5 with revenue of $253.49 billion, saw nearly $280 billion in market value erased in a single session despite having no direct earnings event on June 5. AMD, carrying a P/E of 156.5 against revenue of $37.45 billion, and Micron, with a P/E of 40.8 and revenue of $58.12 billion, similarly experienced declines exceeding 9%, reflecting how guidance disappointments at one major node in the AI supply chain can propagate rapidly through related names. The episode illustrates the concentration risk inherent in a sector where a significant portion of market capitalization is tied to AI infrastructure spending trajectories.

Broadcom's earnings also provided a window into the composition of AI chip demand. With six named custom ASIC customers — including hyperscalers and frontier AI model developers such as Anthropic, Google, Meta, and OpenAI — the company's revenue base is concentrated among a small number of large buyers. Any shift in capital expenditure plans among these customers would have an outsized effect on Broadcom's AI chip revenue, and by extension, on sector-wide sentiment.

Sectors and assets to watch

The four tickers at the center of the June 5 selloff span distinct but overlapping segments of the AI semiconductor supply chain. Broadcom (AVGO), with a market cap of approximately $1.83 trillion and revenue of $75.46 billion, is the primary custom ASIC designer for major AI model developers and hyperscalers; its guidance trajectory will continue to serve as a sector-wide reference point for AI chip demand. Nvidia (NVDA), with a market cap of approximately $4.97 trillion and revenue of $253.49 billion, remains the dominant supplier of GPU-based AI accelerators and saw its valuation fall below $5 trillion following the June 5 session. AMD (AMD), with a market cap of approximately $760.48 billion, competes in both the data center GPU and CPU markets through its Instinct accelerator and EPYC processor lines. Micron Technology (MU), with a market cap of approximately $974.37 billion and revenue of $58.12 billion, supplies high-bandwidth memory — a critical component in AI accelerator systems — making it directly exposed to fluctuations in AI infrastructure build-out rates.

Beyond these four names, the selloff has implications for the broader data center infrastructure ecosystem, including suppliers of networking components, power management semiconductors, and advanced packaging capacity. Companies positioned as enablers of AI compute density — whether through memory, interconnects, or custom silicon — share exposure to the same demand signals that drove the June 5 repricing.

What to watch next

Market participants will be focused on whether Broadcom's unchanged full-year 2026 AI chip sales forecast of $100 billion is revised upward in subsequent quarters, and whether Q3 AI chip revenue tracks toward or above the $16 billion guided figure. Nvidia's next earnings cycle will be closely scrutinized for any corroborating or diverging signals on hyperscaler AI capital expenditure, particularly given the $280 billion in market value erased from the company in a single session. Micron's high-bandwidth memory order trajectory and AMD's Instinct accelerator pipeline will also be monitored as secondary indicators of whether the June 5 selloff reflected a durable reassessment of AI infrastructure spending or a short-term reaction to a single guidance miss. Any public statements on AI capital expenditure from Broadcom's named customers — Anthropic, Google, Meta, and OpenAI — will carry additional weight in shaping the sector's near-term outlook.