Something good has happened to Hyundai Mobis Co., Ltd. (KRX:012330) over the last two years and it does not show up in the revenue line at all.
Gross margin was 11.41% in FY2023, the worst year in the eleven on file. It reached 14.09% in FY2024, 14.45% in FY2025, and 14.74% in the second quarter of 2026. Three and a third points of gross margin recovered in two and a half years, at a company with over ₩60tn of revenue.
Over the same period revenue went from ₩59.25tn to ₩57.24tn to ₩61.12tn. In the second quarter of 2026 it grew 2.4% year on year.
Rising margin on flat volume is usually a good sign. Two other lines complicate it.
Start with what makes the improvement credible. FY2023 was the bottom of a genuinely bad stretch: operating margin of 3.87%, down from 8.15% in FY2015 and 6.20% in FY2019. Mobis had been running a business where revenue grew 24.5% in FY2022 and 14.2% in FY2023 while operating income barely moved. It was buying volume with margin, which is what a captive supplier does when its customers are ramping production and demanding price.
That reversed. Operating margin went to 5.37% in FY2024 and 5.49% in FY2025, and Q2 2026 came in at 5.97%. The gain is entirely at the gross line: SG&A as a share of revenue actually rose over the same stretch, from 7.54% in FY2023 to 8.95% in FY2025. So the improvement is not cost cutting. Something about what Mobis sells, or what it charges for it, changed.
The likeliest explanation is mix, and specifically the after-sales business. Mobis supplies replacement parts for every Hyundai and Kia on the road anywhere in the world, and that business earns far better margins than module assembly because the customer has no realistic alternative and the volumes track the installed base rather than current production. The global Hyundai and Kia car parc has been growing for two decades and keeps growing even in years when new vehicle sales don't.
If after-sales is growing while modules flatten, you get exactly this pattern: margin up, revenue roughly flat. The half-year report's segment note is where that would be confirmed, and the summary financials don't split it, so treat the attribution as reasoning rather than fact.
Inventories were ₩7.50tn at June 30 against ₩6.61tn a year earlier, up 13.5%. Trade receivables were ₩12.24tn against ₩10.83tn, up 13.0%.
Revenue in the quarter grew 2.4%.
Working capital growing at five times the rate of sales is the sort of divergence that deserves an explanation, and it does not fit comfortably with the after-sales story. Spare parts distribution does carry inventory, but it turns predictably against a stable installed base. It does not require a 13% build in a year when sales grew 2%.
What that pattern does fit is module inventory positioned ahead of production that has not arrived. Mobis builds chassis, cockpit and front-end modules to a schedule set by Hyundai and Kia, and if those customers have trimmed output while Mobis kept building to the earlier plan, the difference sits on the balance sheet. Hyundai's Korean plants were also disrupted by strike action in July and August, which stops assembly lines but not necessarily the supplier feeding them.
The receivables build has a simpler possible explanation, which is that roughly nine-tenths of module sales go to affiliates and intra-group payment terms are a matter of group treasury policy rather than credit risk. Money owed by Hyundai Motor is not money at risk. It is money not yet received.
The cash flow statement registers it. Operating income rose 8.0% in the first half, to ₩1.78tn from ₩1.65tn. Cash from operating activities fell 12.7%, to ₩2.33tn from ₩2.67tn.
A company converting less cash on more profit is a company funding working capital, and the two balance sheet lines say where. Mobis can absorb it easily: cash stood at ₩5.45tn at June 30 and total liabilities of ₩23.77tn sit against ₩51.68tn of equity. This is a solvency non-event.
It matters for a narrower reason. Mobis committed ₩500bn to buybacks between April and July and paid ₩446.3bn of dividends in the first half. Capital returns of that size are funded from operating cash, and operating cash is the line that just went down.
Two counters are worth taking seriously.
The first is that a single half-year of working capital movement proves nothing. Automotive suppliers build inventory ahead of model launches and new programme starts routinely, and Mobis has been adding electrification content, which carries higher unit values and therefore more inventory value for the same physical volume. A 13% build in won terms could be a smaller build in units.
The second is that the margin recovery is the more durable fact. Three points of gross margin sustained across ten consecutive quarters is not a working capital artefact, and it has held through a period when Hyundai and Kia have both been squeezed by US tariffs and have been pushing hard on supplier pricing. A captive supplier that improved its margin while its captive customers were losing theirs has real pricing power, or a mix that provides it.
Both can be true. The margin can be structurally better and the current inventory position can still be a quarter or two of pain.
The inventory line at September 30, reported in the Q3 filing in late October. If it flattens or falls while revenue holds, the build was timing and it has cleared. If it grows again against another quarter of low single-digit revenue growth, Mobis is producing ahead of demand and a write-down or a production cut follows.
The second thing is the segment disclosure in the FY2026 annual report next March, and specifically the after-sales division's revenue and operating profit as a share of the total. That converts the central inference in this piece into a number. If after-sales is a growing share of profit while modules stagnate, Mobis is quietly becoming a better and more defensible business than its revenue line suggests, and the 0.785 times book the market pays for it is the wrong price for the wrong reason.
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.