Wall Street analysts expect S&P 500 semiconductor and semi-equipment companies to grow second-quarter earnings 133% year over year, according to CNBC-TV18 — and that group alone is forecast to contribute roughly 44% of the entire index's profit growth. One sector, nearly half the earnings engine. That concentration is the clearest statement yet about where AI capex is landing.
Numbers this lopsided cut both ways. They confirm that inference and training demand is real revenue, not a slide-deck promise. They also mean the index has become a leveraged bet on a handful of chip names — and the recent volatility in those stocks is the market pricing that fragility in real time.
The Economics
Start with what 133% growth actually measures. A year ago, semiconductor earnings were recovering off a cyclical trough — memory oversupply, soft PC demand, thin margins. The comparison base was low. AI accelerator demand has since rebuilt the entire cost structure around high-value silicon.
The relevant metric is revenue-per-wafer. A wafer that once yielded commodity logic chips now yields AI accelerators selling at multiples of the price. Same wafer start, radically different revenue-per-unit. That is why the earnings line moves faster than the shipment line — the mix shifted toward the most expensive product the fab can produce.
The catch is that a low base flatters the growth rate. A 133% jump off a depressed 2025 quarter is not the same as 133% off a strong one. The dollar contribution — that 44% share of index profit growth — is the number that actually matters, because it strips out the base effect and tells you how much the market now depends on this one sector.
Who Benefits
The accelerator designers capture the fattest margins — the companies selling AI GPUs and custom silicon where pricing power is highest. Below them sit the foundries turning wafer starts into finished chips, and the semi-equipment makers selling the lithography and deposition tools that build the fabs. Each layer of the stack participates, but the margin is not evenly distributed.
The equipment names are the quieter beneficiary. AI capex forces new fab construction, and fabs need tools before they produce a single chip. That pulls equipment revenue forward of chip revenue — the capex tells you where the puck is going. When a hyperscaler commits to a build, the tool order lands first.
If semiconductors deliver the forecast growth, a single sector accounts for nearly $0.44 of every dollar of S&P 500 profit expansion this quarter — the tightest earnings concentration the index has carried into an AI cycle.
The vulnerability is symmetric. Concentration that powerful means any guidance miss from a top accelerator name reprices the whole complex. That is the mechanics behind the volatility CNBC-TV18 flagged — investors reassessing how much of the index's growth rests on how few names.
The Second-Order Effect
Cheaper, more abundant compute pushes AI out of the data center and toward the edge. The signal is in the accessory layer. Belkin's new Modular Magnetic Charging Grip for the Nintendo Switch 2 — a detachable 10,000mAh power bank delivering up to 30W — is a small artifact of a larger pattern: consumer devices building the power and connectivity scaffolding that edge inference will eventually run on.
The demand side is broadening geographically too. HSBC argued this weekend that emerging markets are positioned to outperform in the second half of 2026, citing attractive valuations and AI-led growth. That matters for the semiconductor thesis because EM adoption expands the addressable market beyond the US hyperscalers currently driving the capex. A wider customer base makes the earnings concentration less fragile over time — the stack propagates outward.
The uncomfortable use case sits at the other end. OCCRP reported that Russia's FSB had been developing an AI system to merge public media monitoring with restricted databases. It is a reminder that cheaper inference lowers the cost of surveillance as readily as it lowers the cost of a chatbot query. The same silicon serves both.
The Leading Indicator
Watch the semiconductor share of S&P 500 profit growth over the next quarter. If it holds near the forecast 44%, the AI trade remains an earnings story and the multiples are defensible. If that share climbs while the number of contributing names shrinks, the index is not diversifying into AI — it is narrowing into it, and the volatility gets worse before it gets better.