Domestic Semiconductors

The question

Are US-domiciled analog, mixed-signal, and trailing-node semiconductor franchises priced for a normal cyclical recovery, or for a permanently impaired industrial and automotive cycle -- and what is the additional optionality from domestic-content share gains that consensus does not credit?

Base rates

The reference class is analog and mixed-signal cycles: 2001-2002, 2008-2009, 2012-2013, 2015-2016, and 2019-2020. In each case, trough-to-mid-cycle windows produced 25-45% annualized returns for the cohort, with gross margins expanding 300-600 basis points as utilization moved from the low 60s into the high 70s. The base rate for an analog basket beating the S&P across a three-year window anchored at a trough is roughly the 70th-80th percentile of semiconductor sub-baskets, materially better than the cycle odds for leading-edge logic or memory.

The imputed probability embedded in current multiples is that 2027-2028 industrial end-market demand remains soft and that gross margins recover only partially. That is a reasonable prior if the current slowdown is a secular demand break. It is likely too pessimistic if it is a standard inventory correction compounded by a channel destock, which the disclosed lead-time and distributor-inventory data suggests.

Why the consensus view is wrong (or incomplete)

The sell-side treats the industrial and automotive analog cycle as an undifferentiated cyclical recovery trade. The causal mechanism it under-credits is content growth. A mid-cycle industrial or automotive platform refresh in 2026-2028 ships with materially more analog, power, and sensor content than the platform it replaces, driven by electrification, edge compute, and connectivity requirements. That is a structural tailwind layered on top of a cyclical recovery, and it shows up in bill-of-materials disclosures but not in top-down revenue models.

The second missed mechanism is the reshoring of trailing-node capacity. Domestic foundry and packaging capacity is expanding at a pace that changes the unit-economics of supply-chain-sensitive end-markets -- aerospace, defense, medical, automotive-safety -- where customer qualification favors US-domiciled manufacturing. Customers pay for supply-chain certainty, and that premium accrues to domestic producers.

The market is pricing a cyclical recovery. The real story is cyclical recovery layered with content growth and a supply-chain premium. That asymmetry is the inefficiency.

Position construction

The book organizes into three sub-books.

The first is core analog and mixed-signal compounders: ADI at 20% and TXN at 20%, both with decades of capital-return discipline, gross-margin durability through prior cycles, and direct exposure to industrial and automotive content growth. MCHP (~9%) adds microcontroller breadth, and MPWR (~12%) captures power-management share gains into data-center and automotive.

The second sub-book is domestic-manufacturing optionality, anchored by INTC at 20% as the clearest asymmetric bet on foundry-services traction and domestic leading-edge capacity, plus GFS (~4.8%) for trailing-node foundry exposure and AMKR (~2%) for advanced-packaging capacity with a growing Arizona footprint. ON (~5.4%) adds power and silicon-carbide exposure that spans industrial, automotive, and renewables.

The third sub-book is smaller specialty and recovery-leverage positions: LSCC (~2.4%) for programmable-logic in industrial edge, CRUS (~1.6%) for audio and mixed-signal, SLAB (~1.1%) for low-power wireless, SMTC (~1.5%) for data-center and industrial signal integrity, and DIOD (~0.6%) for discrete and small-signal breadth. These are sized to provide cohort exposure without concentrating in the most idiosyncratic names.

The three 20% anchor positions reflect where conviction on durable cash generation is highest, with INTC carrying the widest dispersion of outcomes and sized for that role explicitly.

Asymmetric payoff

If the industrial and automotive inventory correction normalizes by mid-2027 and gross margins recover 300-400 basis points off trough, the weighted book returns roughly 25-40% per year. If the slowdown extends and multiples compress, the book returns roughly -10 to -20%, with dividend and buyback support cushioning the drawdown at the top three holdings. If content-growth disclosures from auto and industrial OEMs accelerate and domestic-manufacturing qualification wins flow through faster than expected, the right tail is 50-70% with meaningful multiple expansion at INTC and the foundry-exposed names.

Assigning a 55% probability to the base case, a 25% probability to the bear case, and a 20% probability to the bull case, expected value is roughly +16 to +24% annualized against an SPY base rate near +8%. The payoff profile is asymmetric because the downside is cushioned by cash generation and capital returns at the largest holdings, while the upside is uncapped on content growth.

Three things that would change our mind

1. Distributor inventory days at ADI, TXN, or MCHP failing to decline for two consecutive quarters, indicating that the channel destock is stalled rather than completing.
2. Gross-margin guidance at any of the top three holdings moving down by more than 200 basis points on a forward basis, signaling a structural rather than cyclical margin reset.
3. Evidence that domestic-manufacturing qualification wins are flowing to non-US producers at meaningful scale, invalidating the supply-chain-premium component of the thesis.

What we are explicitly NOT betting on

We are not buying memory, leading-edge logic merchant, or pure AI-accelerator exposure; those theses belong in a different portfolio and have different reference-class base rates. We are not betting on a specific tariff or export-control outcome; the domestic-manufacturing thesis holds across a wide range of policy configurations because the capex has already been committed. And we are not buying a single-cycle call; the book is designed to deliver positive expected value across a range of normalization paths, with the asymmetry coming from the combination of cycle recovery and content growth rather than from either alone.