Macquarie finished selling LG CNS Co., Ltd. (KRX:064400) in January 2026. It moved its last 8.3% in a block trade, closing out a position that had been the single most discussed thing about this stock since it listed.
That overhang was supposed to be the problem. It is gone. The shares closed at ₩72,500 on September 3, against a 52-week high of ₩143,700. The IPO in February 2025 priced at ₩61,900.
So the stock has round-tripped almost the whole way back to its offer price, with no seller left to blame. Something else is going on, and it is visible in the income statement.
FY2024 revenue was ₩5,982.6bn. FY2025 revenue was ₩6,129.5bn. That is growth of 2.5%.
The first half of 2026 did better, at ₩2,835.8bn against ₩2,671.5bn, up 6.2%. But operating income over the same half went from ₩219.7bn to ₩222.1bn. That is growth of 1.1%.
The second quarter was worse than that. Revenue rose 4.2% to ₩1,520.8bn while operating income fell 9.2% to ₩127.9bn. Operating margin came in at 8.4%, against 9.6% a year earlier.
The company's explanation is investment. It says it is spending ahead of demand on what it calls Physical AI and Agentic AI, and it expects visible growth in the second half. That may well be true. It is also what every systems integrator says when margins compress.
Here is the number that matters most, and it does not get quoted enough.
LG CNS reported ₩1,671.4bn of revenue from AI and cloud in the first half. That is roughly 59% of total revenue. It grew 5.1% year on year.
Think about what that means. The part of this company that carries the AI label is already the majority of it, and it is compounding at about 5%. There is no small, fast segment waiting to take over the mix. The mix has already turned, and the growth rate is what it is.
That is the core problem with the ₩143,700 print. At that price the market was paying for a business whose largest line was accelerating. Its largest line is growing at roughly the rate of Korean nominal GDP.
The market capitalisation was ₩7 trillion at Thursday's close. FY2025 net income was ₩439.2bn, so the multiple is around sixteen times last year's earnings. Total equity was ₩3,056.0bn in June, which puts the shares near 2.3 times book. Those are not distressed numbers. They are ordinary numbers for an ordinary growth rate.
US investors coming to Korean IT services need to understand the calendar before they read a quarterly report.
Fourth-quarter operating income was ₩212.0bn in 2025. Full-year operating income was ₩551.8bn. So 38% of the year arrived in three months. In 2024 the same figures were ₩200.1bn and ₩512.9bn, or 39%.
The reason is how Korean enterprise projects are contracted and accepted. Systems work is signed against a delivery milestone, and clients accept and pay at year-end, when budgets close. Revenue recognition follows acceptance.
The practical consequence is that the first three quarters of any year look weak by design, and a first-half miss tells you far less than it would at a US software company. It also means the fourth quarter carries enormous single-period risk. One large project slipping across December 31 moves the annual result.
Cash and equivalents were ₩1,140.99bn at the end of December. They were ₩565.0bn at June 30.
Some of that is normal. Trade receivables fell from ₩1,664.5bn to ₩1,073.7bn over the same six months, which is the year-end billing cycle unwinding. Financing took ₩386.8bn out, mostly dividends. FY2025 dividends paid were ₩218.7bn.
But investing activities absorbed ₩521.7bn in the half, and purchases of property and equipment were only ₩13.7bn of it. So roughly half a trillion won moved into something other than plant. Intangible assets did not rise. They fell slightly, from ₩52.8bn to ₩49.1bn.
That last detail is worth pausing on. Before the IPO, the company said it would use proceeds partly to buy a foreign IT firm. No goodwill has appeared on the balance sheet since. Whatever the investing outflow bought, it was not an acquisition of any size.
Operating cash flow was negative ₩88.1bn in the second quarter. It was also negative in the second quarter of 2025, so this is seasonal rather than new. Still, a company that generates its cash in a single quarter is a company you judge in January.
The bull case is not empty.
LG CNS won an order to build an AI data center in Indonesia, which it describes as the first such overseas order by a Korean company, with completion expected in the second half. If that becomes a repeatable export business rather than a one-off, the growth rate changes. Building data centers abroad is a different economic activity from integrating systems for LG affiliates.
The margin decline is also consistent with front-loaded investment. Systems integrators book costs on large implementations before the revenue catches up. If the second half delivers what management expects, the 9.2% profit decline in the second quarter will read as a timing artefact.
And the shareholder register is now clean. LG Corp holds just under half. There is no private equity seller waiting for the next window. Every block trade that could suppress the price has already happened.
Against that, the dependence on the LG group is the risk nobody quantifies publicly. Korean law restricts how much work large business groups can direct to affiliates without competitive tender, and the regulator watches these relationships. A captive customer base is a floor under revenue and a ceiling on the multiple, at the same time.
There is also a straightforward competitive point. Samsung SDS does the same things for a bigger group, and both companies are now telling investors the same AI and cloud story. Neither is growing quickly.
The fourth quarter is the number. Look at operating income against the ₩212.0bn of Q4 2025. If the second-half investment story is real, that comparison is where it shows up, and nowhere earlier.
Second, watch the AI and cloud revenue growth rate. It was 5.1% in the first half. If it moves into double digits, the mix argument becomes an argument about the multiple. If it stays near 5%, the multiple is already about right.
Third, watch for an acquisition. The company said it wanted to buy abroad and has not yet. Roughly ₩500bn of investing outflow in one half without a matching intangible tells you the money is moving. Where it lands will say more about the next three years than any quarterly margin.
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.