LATTICE SEMICONDUCTOR CORP (LSCC) stock intrinsic value fair value estimate & owner earnings analysis

Lattice Semiconductor designs small, low-power field programmable gate arrays - chips customers configure themselves - and sells them, mostly through two distributors and mostly into Asia, for server and communications equipment, industrial and automotive systems, and consumer devices. Manufacturing is outsourced to foundries such as TSMC, UMC and Samsung. In July 2026 Lattice bought AMI, a maker of boot firmware and server management software, for roughly $1 billion of cash plus about 5.2 million shares, funded largely by a $925 million term loan due June 2031 that amortises quarterly. That deal turned a debt-free company into a leveraged one. What the business actually earns for owners. We measure owner earnings as operating cash flow, less working-capital swings, less the cost of stock issued to employees, less the capital spending and software/IP licence payments needed to stay in business (we count those licences, about $10-20 million a year, as capital spending because the chip-design tools are a recurring necessity). On that basis Lattice earned between a loss of $14 million and a profit of $206 million a year over 2016-2025. The cycle is the dominant fact: revenue rose from $408 million in 2020 to $737 million in 2023 as customers over-ordered, then fell 31% in 2024, and owner earnings went from $206 million to $53 million to a $14 million loss in fiscal 2025. Stock compensation is the second dominant fact: it was $116 million in 2025, 22% of revenue, against 3-5% of revenue before 2020, and it is still running near 20% of revenue in 2026. We treat it as a real cost, which is why our owner earnings sit far below the company's adjusted profit figures. The earnings level. We use the median of 2019-2025, the period since the company was restructured around its current product lines, which spans a full upswing and a full destocking. Adding back interest so we value the business before its lenders, that median is about $66 million a year. Recent quarters are much stronger: first-half 2026 revenue was up 52% with no acquisition behind it, and annualised owner earnings are roughly $80-90 million. That is the reason the first year of the base case starts at $85 million, above fiscal 2025's loss; we do not lift the long-run level on two quarters. The AMI business. Management says AMI will exit 2026 above $200 million of revenue, split roughly 60% boot firmware and 40% server manageability, with maintenance, subscription and royalty streams. Its profitability has not been disclosed in a filing yet, so we credit only what a firmware business of that size can plausibly hand owners after its own stock compensation: about $45 million a year in the base case, $60 million growing slowly in the bull case, and $25 million shrinking 5% a year in the bear case. Management's expectation of 25% AMI growth in 2027 is not credited in any case, because it is a forecast rather than something already secured. Base case: Lattice's chip business delivers $85 million next year, $72 million the year after as the current upswing normalises, then holds its $66 million median indefinitely on the strength of design wins that stay in customer products for many years; AMI contributes a flat $45 million. No market growth is credited. The business is worth about $1.87 billion; it owes roughly $880 million net (the $925 million term loan less cash beyond operating needs); the difference is about $0.99 billion, or $6.85 per share on about 145 million diluted shares. Bear case: one more destocking of the kind seen in 2024-2025 arrives in years two and three - the chip business falls to $10 million and then a $15 million loss, as fiscal 2025 already demonstrated is possible with today's stock-compensation habits - and then recovers only to $40 million a year. AMI is worth $25 million a year and erodes 5% annually as hyperscale server builders bring firmware in-house. In those trough years the combined business would not cover roughly $55-60 million of annual interest and would lean on its $200 million revolver, and refinancing in 2031 would likely cost about two points more. The business is worth about $0.93 billion against about $0.92 billion owed, leaving essentially nothing - about $0.11 per share. In this case shareholders own a business worth roughly what it owes. Bull case: the data-centre-driven upswing carries the chip business to about $130 million in year two and it then holds near $110 million, growing 3% a year on its own design-win engine, while stock compensation drifts back toward historical norms; AMI produces $60 million growing 3%. The business is worth about $3.37 billion, less $0.88 billion of debt, or about $17.15 per share. These figures are not comparable with our previous estimate, which was built on a different measure of owner earnings and before the AMI loan was drawn. How we set the margin-of-safety price. Our confidence in the earnings level is low: over ten years the business has earned anywhere from a loss to $206 million, and the single biggest swing factor - how much stock management grants itself - is a choice that has moved against owners. The worst plausible outcome is a deep destocking while carrying $900 million of floating-rate debt, with 60% or more of sales flowing through two distributors in Asia and all wafers made by outside foundries in East Asia; the structure would likely survive through the revolver, but owners would be diluted or wait years. The things we are most likely to be wrong about are AMI's real profitability, the cyclical level of the chip business, and stock compensation; if all three go modestly the wrong way, the shares are worth little more than the bear case. Lattice pays no dividend and its buybacks have only offset dilution, so owners are paid nothing during the holding period. The 20% annual target requires a price at which the base case alone, delivered over three to four years, implies roughly doubling, and the bear case does not imply a total loss - about $3 per share, which is the margin-of-safety price. At that price the whole company would be valued at roughly $430 million on top of its debt, a level the chip business's own normal earnings could support even if AMI proved worth only what was borrowed for it.