MAGNACHIP SEMICONDUCTOR Corp (MX) stock intrinsic value fair value estimate & owner earnings analysis

Magnachip designs and makes power semiconductors, mostly in its own wafer fab in Gumi, South Korea. Its products are MOSFETs and IGBTs, the switches that control electricity in TVs, phones, PCs, appliances, e-bikes, solar inverters and cars. It also makes power-management chips for TV and display-panel makers. In April 2025 the board shut down the Display chip business after failing to find a buyer on acceptable terms, which left a pure-play power company. That business had $178.9 million of sales in 2025, down 3.7%. About half was billed in Korea and most of the rest in China and Hong Kong. One customer, SAMT, took 29.7% and the ten largest customers took 74.3%. The central problem is that the remaining business does not cover its costs. Gross margin fell from 21.5% in 2024 to 17.6% in 2025. The company blames falling prices on its older products, especially in China, plus a one-time incentive paid to Chinese distributors. Selling and administrative costs ($35.1 million) plus R&D ($27.3 million) exceeded gross profit by about $31 million. The first half of 2026 was no better. Sales were $90.9 million, gross margin was about 17%, and the operating loss widened to $17.1 million from $11.9 million a year earlier, with R&D up 22%. We measure the cash an owner could actually take out: operating cash flow with working-capital swings removed, minus stock compensation and the capital spending needed to keep the plant running. We count Korean statutory severance payments as a real cost. On that basis the company lost $1.35 per share in 2025, $0.42 in 2024 and $0.95 in 2023. Part of 2025 was one-off. Of the $30.0 million spent on plant, $21.4 million went on an upgrade of the Gumi fab, largely funded by Korea Development Bank equipment loans. Another $6.5 million was severance to close the Display subsidiary. Excluding those items, the ongoing power business loses roughly $0.65 to $0.85 per share a year. The profitable years of 2018, 2019 and 2021 belonged to a different company. It still owned the Foundry Services Group and Fab 4, which were sold on September 1, 2020 for about $350.6 million, and later a Display business during a boom. Those years do not describe what is left. None of the cases we can support with evidence brings the power business to lasting positive cash earnings. That would take gross margins in the low-to-mid 30s, against the 17% to 21.5% actually achieved. So the shares' value depends on what the balance sheet is worth when the losses stop, less the cash used up before then. At June 30, 2026 the company had $87.9 million of cash, $23.8 million of receivables, $34.1 million of inventory and $92.2 million of net plant and equipment. Against that it had $98.3 million of total liabilities, including $41.5 million of bank debt. A Korea Development Bank term loan of about $27.9 million falls due on March 26, 2027. The balance sheet also carries $60.2 million of deferred tax assets, which we value at zero because they only pay off if the business makes taxable profits. Separately, a water-treatment service contract for the Gumi fab had $47.1 million of payments remaining through 2038. Base case ($0.10 per share): the business loses about $24 million in the next year and $22 million the year after. That is somewhat better than the first-half 2026 run rate. There are early signs the upgrade is helping (gross margin was 19.3% in the second quarter of 2026 against about 15.6% in the first), and overhead fell after the August 2025 voluntary resignation program. By mid-2028 the term loan will have come due and the upgrade's results will be known. We assume the board then does what it did with Display and stops the losses through a sale or closure. We assume it recovers 90% of receivables, half of inventory and 35% of the plant's book value, pays all liabilities, spends about $10 million on exit costs and settles half of the water contract. That leaves almost nothing for shareholders. Bear case (-$1.69): losses of about $30 million a year last three years, as price pressure in China continues, until cash is essentially gone. Closure then recovers less: 25% of plant value and 40% of inventory. The full water contract and about $15 million of closure costs must also be paid. In that case the company would owe roughly $73 million, or about $2 per share, more than its assets would fetch, and shareholders would get nothing. Bull case ($1.90): the upgrade restores 2024's 21.5% gross margin and losses narrow to about $15 million a year. Within 18 months the board sells the power business or the fab to a buyer that takes on the workforce and facility contracts, at half the plant's book value. The 2020 sale to Key Foundry shows that Magnachip's Korean fab operations have found buyers before. On this evidence the shares are worth little unless the losses stop soon. The margin-of-safety price applies only if the board commits to a sale or wind-down while enough cash remains; in that event it is $0.70. The estimate would be wrong if the business showed two or three quarters of gross margin clearly above 25% with growing sales. One possible source is low-voltage MOSFETs for communication devices, which lifted Korean sales 10.2% in 2025. It would also be wrong if a buyer paid a going-concern price. One figure we could not reconcile: first-half 2026 operating cash outflow was only $2.9 million, far less than the operating loss. The available filings do not explain the gap fully, so we did not treat it as the ongoing rate.