Open in interactive viewer → charts, metric popovers & call review

KT Wants to Be Korea's Token Factory, But Its Phone Bill Is Still Shrinking

A 1-gigawatt data-center pipeline and a 500-model inference platform are the new story; a 3.9% decline in wireless service revenue is the old one, and it still pays the bills.
KT · Earnings Call · 2026-11-10

The quarter the pivot was supposed to pay for

KT's third quarter of 2026 is a study in two companies occupying one balance sheet. Operating revenue fell 1.8% year-on-year to KRW 6.9975 trillion, operating profit fell 6.1% to KRW 505.3 billion, and net income dropped 17.5% to KRW 367.3 billion. Management's framing is that last year's property development gain (the NCP one-off) flatters the comparison, and on that basis operating profit actually rose 3.5%. Fair enough — but the underlying line is still going the wrong way: wireless service revenue fell 3.9% to KRW 1.6668 trillion on subscriber churn during the cancellation-fee waiver window and the customer appreciation program, media revenue slipped 2.4%, and B2B services dipped 0.7%.

The CFO's own summary captures the bet: “we were able to affirm the recovery trajectory of our core telecom business, while expanding B2B order wins anchoring on the data centers as we fortified the foundation for mid- to longer-term turnaround.” — Hye-Byung Min, Chief Financial Officer (CFO) · 2026-11-10 Recovery plus foundation. Nothing yet that looks like a growth engine on the P&L.

And KT is not pretending otherwise. The Head of the AX Business Division, Park Sang-won, gave an unusually blunt read on the home market:

the telecom market itself is actually quite saturated, including the mobile, the internet and the TV market. So all of the 3 telcos are facing certain challenges in growing their businesses.

Sang-Won Park, Head of AX Business Division · 2026-11-10
That is the admission that makes the rest of the call matter. A mature, saturated domestic telco cannot grow its way into a re-rating; it has to sell something else.

Where the story actually moved: order intake, not revenue

The genuinely new material this quarter is the order pipeline. B2B order intake cumulatively through Q3 rose more than 20% year-on-year, spread across six customer segments — manufacturing, public, healthcare, distribution, national defense and finance — with wins described as Woori Bank's Agentic AICC and an intelligent national defense platform. That is a material change in tone from six months ago, when the same question got a defensive answer about why KT's B2B growth trailed peers: “in terms of the reason why our B2B growth rate is lower compared to the industry peers” — Min Jang, CFO · 2026-02-10 was answered with a combined KT-plus-KT-Cloud growth figure of 6%. Now the company leads with order intake growth of 20%+, and Head of Enterprise Kim Bong-gyun promises the number will rise again by year-end.

The catch is the one investors should hold onto: order intake is not revenue. Order intake entering the door today shows up as AX revenue later — the company reported AX revenue up 9.8% in Q3 and 11.9% cumulatively, against a stated ambition of more than doubling by 2028 off the 2025 base. A pipeline growing at roughly twice the rate of the revenue it is supposed to feed means the conversion gap is widening, not closing.

The data-center plan is the biggest number on the call and the least bankable. Head of Enterprise Kim laid out the arithmetic: 152 megawatts of capacity today, a target of adding 1 gigawatt by 2031, with a gigawatt-scale pipeline where “the capacity pipeline that we have already secured, it amounts to 911 megawatts across 20 sites. As a backup, we have 500 megawatts across 10 sites.” — Bong-Gyun Kim, Head of KT Enterprise · 2026-11-10 Of that ~900MW, only 200MW has actually been decided, with a further 60MW to be approved before year-end. The 80/20 leased-to-owned mix and the explicit reference to the company's debt ratio as the gating factor for financing tell you management knows this is a capital-allocation story, not a signed-revenue story.

The Token Factory is the one genuinely proprietary idea

Buried behind the megawatts is something more interesting and much harder to copy: Token Factory, launched at the end of October. Park Sang-won described a dual model — locally hosted GPUs running inference on KT's own models, plus routing to third-party models through a partnership with OpenRouter — with more than 500 models served under one roof. “The key strength that we have is the fact that we can generate the token ourselves as well as intermediate. We are the only company who would have that competitive edge.” — Sang-Won Park, Head of AX Business Division · 2026-11-10 The real moat, though, is invoicing: a telco with precise billing infrastructure can meter inference the way it meters minutes. That is a credible, company-unique angle on the global personal AI agents and agentic operating system build-out, and it is the piece of this call that peers cannot simply announce.

The analyst reaction was telling — the first question did not ask how many megawatts, it asked how to value token flow at all, wondering whether to focus on raw inference volume or on bundled telecom-and-cloud revenue. When the buy side is still negotiating the KPI, the business is early.

Meanwhile the unifying platform language, AX business built on Agentic On and reusable standardized models for verticals like healthcare, is no longer distinctive — agentic offerings and AI data centers are now close to boilerplate across every reporter in this window, from semiconductor equipment to utilities. KT's differentiation has to come from the token layer and the sovereign-AI angle, not from saying "agentic" louder.

What vanished, what lingers, and the tape in the way

The most quietly important change is what stopped being said. The data breach and the customer trust remediation that dominated the Q1 2026 and Q4 2025 calls — the 14-day cancellation-fee waiver, the SIM replacement, the KRW 1 trillion five-year security spend — has fallen out of the current keyword set almost entirely, displaced by AX and Token Factory. That is management choosing to move the narrative forward, and the fact that complaints fell 60% in Q3 versus Q1 gives them license. What lingers is the bill: the customer appreciation program is still a top-ranked theme and still the stated reason operating profit fell.

Shareholder return got the most evasive answer of the day. Pressed on whether the KRW 2,400 minimum DPS holds and what Q4 looks like, the CFO offered only that “the final decision will be made by the BOD based upon the annual net profit that we book” — Hye-Byung Min, Chief Financial Officer (CFO) · 2026-11-10 — a softer answer than the February call, when the outgoing CFO told investors plainly that “our performance or earnings in 2026 is going to be better compared to 2025.” — Min Jang, CFO · 2026-02-10 Operating profit is down 6.1%. The commitment to KRW 1 trillion of cumulative buyback by 2028 and a 9-10% ROE target by 2028 remains, but the earnings bridge to it just got narrower — and note the CFO seat itself has turned over since February.

Finally, the tape is not cooperating. Over the last 30 and 90 days the listed complex KT is now pitching itself into — data-center capacity, critical IT load, high-bandwidth memory, co-packaged optics — sits among the market's heaviest decliners, with dozens of names selling off together. KT is asking investors to pay for a 2031 gigawatt vision precisely as the market is marking down the companies that already own those gigawatts. That asymmetry, not the KRW 505 billion of operating profit, is the thing to watch.