On June 18 Korea's Patent Court ruled against HPSP Co., Ltd. (KOSDAQ:403870). It found that a competitor's high-pressure annealing equipment does not fall within the scope of HPSP's patent, and the rival said the ruling removed the legal uncertainty that had been holding back its own product.
HPSP's business is one process step. It makes high-pressure hydrogen annealing tools, which repair defects in the ultrathin layers inside leading-edge logic and memory chips. Until now it has effectively been the only supplier. That is why the company earns the margins it earns.
The first half of 2026 brought revenue of ₩78.1bn, down 11.5% from ₩88.2bn a year earlier. Operating income was ₩44.5bn against ₩47.4bn. The operating margin went up, to 57.0% from 53.7%.
A shrinking revenue line and a widening margin, in the same six months the monopoly lost a court case. Those three facts have to be reconciled.
Start with how unusual the profitability is.
FY2022: revenue ₩159.3bn, operating income ₩85.2bn. Margin 53.5%. FY2023: revenue ₩179.1bn, operating income ₩95.2bn. Margin 53.2%. FY2024: revenue ₩181.4bn, operating income ₩93.9bn. Margin 51.8%. FY2025: revenue ₩173.0bn, operating income ₩89.9bn. Margin 52.0%.
Gross margin ran 73.0% in FY2025. Cost of revenue was ₩46.7bn on ₩173.0bn of sales.
Semiconductor equipment companies do not earn these numbers. Even the best of them run operating margins in the thirties. A number above fifty, held for four consecutive years, is not manufacturing skill. It is pricing power from having no alternative supplier, and customers who cannot switch without requalifying a process step that took years to approve.
Which is exactly what a court ruling can change, slowly.
The distinction matters and it is easy to get wrong.
The Patent Court did not strike down HPSP's patent. It examined the rival's locking mechanism and found it structurally different. HPSP's patent claims a rotating fastening ring. The competitor's design rotates the outer door itself, so it needs no such ring. Different structure, no infringement.
So HPSP keeps its patent. It just no longer blocks this particular competing design.
That is a slower and smaller problem than losing the patent outright, but it is a real one. The barrier here was never only legal. It was also the qualification cycle: a chipmaker takes years to approve a new tool for a production step, and nobody starts that clock while litigation is running. The ruling starts the clock.
Nothing about second-half 2026 revenue depends on it. Something about 2028 pricing might.
Here is the part of this company that gets less attention than the margin.
Revenue was ₩179.1bn in FY2023, ₩181.4bn in FY2024, and ₩173.0bn in FY2025. Three years, no growth, then a small decline. The first half of 2026 is down 11.5%.
For a company with a monopoly on a process step inside the fastest-growing part of semiconductors, that is a strange record. It says the number of tools sold is governed by customers' capacity decisions rather than by HPSP's ability to sell. When a foundry adds a fab, HPSP ships. When it pauses, HPSP does not.
Quarterly revenue swings enormously as a result. ₩36.9bn, ₩51.3bn, ₩32.0bn, ₩52.7bn, ₩31.9bn, ₩46.1bn across the last six quarters. There is no smoothing mechanism in a business that sells a handful of large tools.
One accounting note before anyone builds a trailing multiple. The first quarter of 2026 carried a negative income tax charge of ₩5.7bn, a credit rather than an expense, which lifted that quarter's net income to ₩24.9bn on ₩19.1bn of pretax income. Trailing earnings are flattered by it.
Inventories were ₩16.0bn at the end of December. They were ₩23.5bn in March and ₩41.9bn in June. That is a rise of 163% in six months, on a balance sheet where inventory had been shrinking for two years.
A capital equipment maker builds inventory when it has orders it has not yet shipped. This is the clearest signal in the accounts that the second half is meant to be much larger than the first.
The sell side agrees. Samsung Securities has forecast 2026 revenue of ₩243.4bn, up 40.7%, with operating profit of ₩132.2bn. Against a first half of ₩78.1bn, that requires a second half above ₩165bn. The first half of a year has never been that small a share of the total.
I would treat the forecast as a hypothesis and the inventory as the evidence for it. Both get tested in the third-quarter report.
There is a governance story running alongside all this.
HPSP's controlling shareholder is a vehicle set up by the private equity firm Crescendo Equity Partners. It held 39.28%. In January 2026 it sold 10.01% in a block trade. In February it sold another 9.05%, at ₩41,600 to ₩42,800 a share, raising about ₩325.3bn.
Then in June it changed course. Crescendo dropped the sale of the management stake and moved to direct control, saying it would build value over the medium to long term instead. The company has since called an extraordinary shareholder meeting for October 22 to elect the chief executive as an inside director.
For minority holders that is mostly good news. A sponsor that keeps selling into the market caps the price. A sponsor that stops has decided the asset is worth more later. It also means the block-trade overhang, which weighed on the stock through the first quarter, is largely spent.
It does not remove the exit question. Private equity firms sell eventually.
Valuation is the first risk and it is not subtle. The market capitalisation was ₩4.1 trillion at Thursday's close. Trailing net income is roughly ₩86bn. Total equity was ₩313.5bn in June.
Those numbers put the shares near fifty times trailing earnings and above ten times book. That is a monopoly multiple, and the monopoly just got a court ruling against it.
The share price already reflects how contested this is. The 52-week range runs from ₩26,100 to ₩83,500. The stock closed at ₩50,200 on September 3, roughly in the middle. Anyone buying here is taking a position in a fight, not a compounder.
Customer concentration is unquantifiable from the outside. HPSP sells to the largest foundry and memory makers, and the filings do not break out who. Two or three decisions determine the year.
One last technical point. HPSP files separate, parent-only financial statements. There are no consolidated accounts because there is nothing material to consolidate. That is normal in Korea for single-entity companies, and it means what you see is genuinely all of it.
Third-quarter revenue against that ₩41.9bn of inventory. If the tools ship, revenue should jump sharply and the full-year forecast becomes reachable. If revenue prints near ₩35bn again, the inventory is a delayed order rather than an imminent one.
Second, watch for the competitor announcing a customer qualification. That, not the court ruling itself, is the event that ends the monopoly.
Third, watch gross margin. It is 73%. The first sign of real competition will show up there before it shows up in revenue.
kstock reads DART every morning and writes up what moved — the contract, the buyback, the number that does not add up. The daily post and a Saturday roundup, by email.