ALPHA & OMEGA SEMICONDUCTOR Ltd (AOSL) stock intrinsic value fair value estimate & owner earnings analysis

Alpha and Omega Semiconductor (AOS) designs and makes power semiconductors - power MOSFETs (power discrete, $438M of FY2026 revenue) and power management chips (power IC, $234M) - used in PCs and graphics cards, servers, phones, chargers, appliances and industrial power supplies. It runs its own wafer fab in Hillsboro, Oregon and packaging and test plants in Shanghai, and sells mostly through two large Asian distributors that together took 72% of FY2026 revenue (53% and 19%). Nearly all product ships to Hong Kong for Asian electronics assemblers. It also owns 18.9% of a 12-inch wafer fab joint venture in Chongqing, China, which supplies it with wafers. What the business earns. We measure owner earnings as operating cash flow, with working-capital swings removed, less stock compensation and the capital spending needed to keep the business running. Revenue has been flat at roughly $660-700M since FY2023, but the business has not earned money on that basis. Gross margin has fallen from 28.9% in FY2023 to 22.3% in FY2026 (higher material costs, lower unit volume), while research spending rose to $104M. Operating losses were $3.8M, $28.4M and $43.2M in FY2024-FY2026, and the FY2024 and FY2025 figures were flattered by $21.2M and $13.8M of one-time silicon-carbide license fees that ended in February 2025. Excluding that license income and about $13M of taxes paid on the JV stake sale, owner earnings were roughly -$12M (FY2024), -$14M (FY2025) and -$34M (FY2026). Over the last decade the company earned good money only in the FY2021-FY2022 chip shortage; outside that window, owner earnings have been around breakeven or negative. The FY2025 net loss of $97M was mostly a non-cash $76.8M write-down of the JV stake, which we add back. Before FY2022 the company consolidated the Chongqing fab and its heavy construction spending, so we measure the current business from FY2023, cross-checked against the full cycle back to FY2015. Balance sheet. At June 30, 2026 AOS held $181M of cash against about $5M of debt and finance leases, roughly $5.87 per share net. In July 2025 it sold 20.3% of the Chongqing JV for $150M, all of which has now been received; on that price its remaining 18.9% is worth about $140M before tax (about $130M after tax, based on the roughly 7% tax paid on the last sale). The JV is still loss-making and pays no dividends, so that value only reaches shareholders through another sale. Because we count the cash directly, we leave its interest income out of the earnings stream. Base case (about $7.00 per share). Owner earnings before interest start where FY2026 ended, about -$37M, since nothing in the filings points to a quick turn (FY2026 product revenue was flat and margins slipped). Losses then narrow to -$28M, -$18M and -$8M, and the business runs at about breakeven afterwards, which is its demonstrated average through past cycles. We credit no growth: none is contracted or backed by a repeatable record. The remaining JV stake is valued at the July 2025 transaction price after tax and assumed sold within about two years, as AOS has sold down the JV several times since 2021. Value = net cash $176M + JV about $116M in today's money - about $81M of future losses = about $210M, or $7.02 per share. Bear case (about $1.70). Losses stay at the FY2026 rate of about $37M a year for four years, run at $20M a year for three more before being cut out, and the JV stake realises only AOS's share of the JV's book equity (about $55M) about three years out. That burns most of the cash, leaving about $51M of value. Supporting evidence: losses have grown each year since FY2023, gross margin keeps falling, and 72% of sales depend on two distributors. Bull case (about $18.25). Losses shrink to about $20M next year and end in year two. The business then earns about $18M a year indefinitely, which is below the roughly 5% margin on sales the standalone business earned in a normal year such as FY2017 and is supported by a franchise that has operated for more than two decades. The JV stake sells within a year at the 2025 price. We do not count a repeat of the 2021-2022 chip-shortage profits, because that was a market windfall, not something this business controls. What the shares appear to be worth. Nearly all the value is cash and the JV stake, not the operating business, and the key question is how much cash the losses consume before they stop. The estimate is too low if the business goes back to its pre-2023 margins, and too high if losses persist or the JV stake cannot be sold near the 2025 price. We could not reconcile the full breakdown of the $17.5M of FY2026 cash taxes; we treated about $13M of it as tied to the JV sale, based on the tax amounts the company disclosed for that transaction.