Kakao's AI spinoff is a bet — and the market isn't buying it yet

The split that should have worked — and didn’t

Kakao announced the most consequential corporate restructuring in Korean tech history on a Tuesday morning. The plan was straightforward on paper: split the conglomerate into two entities — KakaoAI, an operating company focused on AI, messaging, advertising, and commerce; and KakaoX, an investment holding that would shepherd KakaoBank, KakaoPay, Kakao Entertainment, SM Entertainment, Piccoma, and Kakao Mobility.

The logic was also straightforward. Kakao’s combined group companies carry an estimated potential value of 34.2 trillion KRW (roughly $24 billion). Yet Kakao itself trades at just 16.8 trillion KRW in market capitalization. The gap — nearly 50 percent — exists because the market applies a conglomerate discount: different businesses under one roof, competing for capital, muddled valuation. Separate them, the theory goes, and each unit trades at its own premium.

Kakao’s stock dropped 7.49 percent that day. Intraday, it fell as much as 13.18 percent.

The market was not impressed.

The NAVER parallel nobody mentions

Kakao is echoing a playbook NAVER first laid out in Korea. When NAVER spun off its search and cloud businesses into a separate structure, it forced analysts to re-rate the company on pure AI and platform merits rather than burying them under slower-growing segments. The question now is whether Kakao’s version of this move will produce the same re-rating — or whether the substance underneath is thinner than the architecture suggests.

Kim Do-young, nominated to lead KakaoX, made the case bluntly. At a current P/E of roughly 21.9x, Kakao is priced as a mature B2C IT company with dominant market shares — not as an AI company. Analysts who model Kakao as a pure AI player, he said, see room for 30x or even 40x. That is a massive gap, and it is precisely the kind of gap that makes restructuring exciting — and dangerous.

The danger is that restructuring alone does not change a company’s revenue curve. It changes how revenue is reported.

The math problem at the center of KakaoAI

Here is where the announcement runs into its first wall of arithmetic.

KakaoAI’s guidance targets 6 trillion KRW in revenue by 2030, up from 2.7 trillion KRW today. Of that 6 trillion, only 1 trillion is supposed to come from AI itself. The remaining 5 trillion comes from advertising and commerce — the old businesses, just relocated to a new company.

In other words, the “AI revolution” at KakaoAI is expected to generate roughly one-sixth of the total revenue target. The rest is the status quo with a new name on the door.

This is not necessarily a dealbreaker. Many AI transitions start this way — advertising funds the AI buildout until the product side catches up. But it does mean the re-rating thesis depends entirely on whether the market believes Kakao can credibly deliver on the AI portion, or whether it will continue pricing Kakao as an ad-and-commerce business dressed in AI clothing.

There is also a structural tension baked into Kakao’s plan. CEO Jeong Sin-hyeon stated that AI applied to real user needs could drive 20 million daily active users on the platform. But he also set a target of increasing KakaoTalk’s average time spent by 50 percent. Those two goals pull in opposite directions. AI that actually solves problems efficiently tends to reduce time spent — you get your answer and you leave. AI that keeps you browsing generates more ad impressions but delivers less utility.

Kakao appears to be betting on the second outcome. That is not a dishonest strategy. It is an advertising strategy.

Who wins, who loses

If the split works as intended, three groups benefit. Institutional investors who have been unable to price Kakao’s components separately finally get clean exposure to either an AI-focused operator or a diversified investment hold. Minority shareholders in the current structure capture the conglomerate discount unwind. And Kakao’s AI team gains autonomy it never had before — no more begging the finance committee for budget when the rest of the group is pulling capital toward banking and payments.

The losers are harder to name precisely but easier to feel. KakaoTalk’s users may find the experience optimized for ad delivery rather than conversation efficiency. Kakao’s employees across the split entities face the friction of dual reporting lines, renewed internal negotiations over resource allocation, and the quiet anxiety that comes with any major restructuring. And analysts who preferred the simplicity of one consolidated income statement now have to track two.

What happens next

The timeline matters less than the milestones. KakaoX targets 10 trillion KRW in revenue by 2030, up from 5.6 trillion KRW last year. KakaoAI targets 6 trillion, up from 2.7 trillion. Both are aggressive growth rates that require more than organizational chart changes — they require products that actually move the needle.

The critical signal to watch is not the stock price in the weeks following the split. It is whether KakaoAI can demonstrate that its AI tools generate measurable revenue beyond the existing advertising and commerce base within the next 12 to 18 months. If the 1 trillion KRW AI revenue target slips, the re-rating narrative collapses and the market reverts to treating Kakao as a conglomerate once again — only now with two balance sheets instead of one.

Kakao’s restructuring is directionally correct. The conglomerate discount is real, and splitting the business is the right structural response. But the harder question — whether AI can genuinely become a standalone profit center rather than a branding exercise — remains unanswered. The market, for now, is waiting to see.