The AI small caps covered in this series — Gaonchips, VUNO, OpenEdges, Konan — share one trait: they all lose money. In the 2026 KOSDAQ, “AI” and “profitable” rarely fit in one sentence. Which is what makes this company interesting. Polaris Office (KOSDAQ 041020): 2025 consolidated revenue of 324.2 billion won, operating profit of 9.1 billion, net income of 13.5 billion. Profitable — on a clean balance sheet with a 28% debt ratio.
Yet the stock closed July 24 at 2,845 won — down 7.8% in a single day, near its 52-week low of 2,520 and less than half the high of 6,150. Market cap: 141.4 billion won. Why does a profitable company draw the same chart as the loss-makers? Because, I think, its signboard and its body are different things. Taking that structure apart is the point of this piece.
The signboard: Korea’s first office AI
Polaris Office’s core business is document-editing software — an office suite compatible with MS Office and Hancom, served across PC, mobile, and web, with a large global accumulated user base. In September 2023 it launched Polaris Office AI, Korea’s first office suite with an embedded generative-AI model, and it performs national R&D projects in lightweight AI models, on-device AI, and intelligent document processing. A document app is a natural seat for an AI assistant, so the adjacency logic here is far sturdier than Nable-style signboard-swapping.
The core numbers point the right way too: in H1 2025 the software division grew 11% year-on-year across both B2C and B2B, helped by a push into government (B2G) customers.
The body: where 324.2 billion won of revenue comes from
Look at the table first.
| Metric | Value | Note |
|---|---|---|
| 2025 consolidated revenue | 324.2bn won | H1 alone: 154.2bn (+28%) |
| 2025 operating profit | 9.1bn won | Operating margin 2.8% |
| 2025 net income | 13.5bn won | |
| Share price (Jul 24) | 2,845 won | 52w range 2,520–6,150 |
| Market cap | ~141.4bn won | |
| P/B · debt ratio | ~1.58x · 28% | Current ratio 367% |
For a software company’s income statement, a 2.8% operating margin is a strange number — office subscription margins should run far higher. The secret is the consolidation scope. The Polaris Office group spans Polaris Sewon, Polaris Uno, Polaris AI, and Polaris AI Pharma around the listed parent, and manufacturing affiliates — auto parts among them — form the body of consolidated revenue. In Dealsite’s phrase, this is “topline growth built on solid affiliate businesses”. The fact that H1 revenue grew 28% while operating profit fell 7% summarizes the structure: the low-margin body drives the topline while the quality of profit moves separately.
In other words: buy this as a pure AI-software stock and the 324.2 billion won revenue line deceives you; buy it as an industrial holding company and the office-AI franchise comes as a free option. The same 141.4 billion won market cap reads completely differently depending on which frame you choose.
Then why is it cut in half?
The 52-week high of 6,150 won was the price when the AI theme was valuing this stock’s signboard; today’s 2,845 is the price after theme flows rotated into AI infrastructure and the market re-anchored on the body’s margin. Another variation on the round trip we saw at Nable and Gaonchips. With one decisive difference — this company still makes money where the round trip ends. A loss-maker’s new low is a survival countdown; a profit-maker’s new low at least has time on the company’s side.
Reasons to take it seriously. Real profits (9.1bn operating, 13.5bn net); balance-sheet slack (28% debt ratio, 367% current ratio); the lowest asset multiple in this series at P/B 1.58x; a growing core business (software +11%) with B2G expansion; and office AI / on-device AI as substantive growth options. It is a configuration where the “failure scenario” is genuinely hard to draw.
Reasons for skepticism. The body of the profit isn’t AI — even if the AI story reignites, whether earnings growth follows is a separate question. The 2.8%-margin consolidated structure inherits the cyclical and currency exposure of manufacturing affiliates; H1’s profit decline (−7%) may already be that signal. And a many-affiliate governance structure makes it harder for minority shareholders to trace where profits accrue.
Checkpoints that actually change the picture. (1) Software-segment revenue and profit separately confirmed in segment notes, with the mix shifting toward it; (2) paid-conversion or subscription metrics for office AI appearing as numbers in filings or IR materials; (3) a trend improvement in consolidated operating margin (from 2.8%); (4) on-device AI national R&D converting into commercial revenue. In my personal framework, Polaris Office differs from the rest of this series: it sits on the watch list waiting not for survival but for re-rating — which also makes it the cheapest name on the list to wait on.
The lesson in one line: in the 2026 KOSDAQ, the AI theme torched loss-makers’ stocks first and abandoned them first. The profitable one rose less and was abandoned just the same. But a company that keeps compounding profit where the theme left it, and a company burning cash there, are arithmetically different one year later. If that difference needs a name — it is the direction of time.
Nothing in this article is financial or investment advice. All figures come from public data as of July 24, 2026, may contain errors, and must be verified against primary disclosures before any decision. The author holds no position in any security discussed.
Sources
- Bosoop — Polaris Office (041020) price and financials snapshot
- Hankyung — Polaris Office H1 consolidated revenue 154.2bn won, +28% YoY (2025.8.18)
- Money Today — Polaris Office H1 revenue 154.2bn: “the B2G push worked” (2025.8.18)
- Dealsite — Polaris Office: topline growth on solid affiliate businesses
- THE VC — Polaris Office company profile
- E-Today — Overlooked KOSDAQ looks to ‘AI data centers’ for its comeback