Anthropic Opens Seoul Office as Samsung, LG and NAVER Deploy Claude - Even as a Korean Telecom Clouds the Day

Anthropic Opens Seoul Office as Samsung, LG and NAVER Deploy Claude — Even as a Korean Telecom Clouds the Day

Anthropic planted its flag in Seoul this week with a roster of marquee Korean customers that reads like a who's-who of the country's industrial economy — NAVER, Samsung, LG, Nexon, Hanwha and more. But the celebration carried an asterisk: less than a week earlier, Washington had ordered the company to cut off foreign access to its most advanced models, an episode triggered in part by a Korean telecom's brush with one of those systems.

The office, which opened June 17 in Yeouido, is Anthropic's third in Asia-Pacific after Tokyo and Bengaluru. It is led by Representative Director KiYoung Choi, a 30-year industry veteran who has held senior roles at Microsoft, Adobe, Autodesk, Google Cloud and most recently Snowflake Korea. The company is organizing the operation around four functions — sales, technology, policy and operations — with an explicit goal of winning enterprise accounts in what is now Asia's fourth-largest economy.

A deep customer roster, announced all at once

What distinguished the launch was not the ribbon-cutting but the breadth of deployments unveiled alongside it. NAVER, Korea's dominant search and platform company, has rolled out Claude Code across its entire engineering organization. Samsung SDS has deployed Claude Cowork and Claude Code inside Samsung Electronics. LG CNS is distributing Claude across the LG Group. Nexon, the gaming giant, is using Claude Code for live-service game development.

Beyond the conglomerates, Hanwha Solutions is bringing Claude to its global workforce through AWS Bedrock with in-region data controls, and Channel Corp is using Claude to power its Channel Talk customer platform, which serves more than 230,000 businesses. Anthropic also signed a memorandum of understanding with Korea's Ministry of Science and ICT and committed to providing Claude free of charge to as many as 60 researchers affiliated with the National AI Research Lab consortium, which spans KAIST, Korea University, Yonsei and POSTECH.

The momentum is not purely top-down. Anthropic says weekly active usage of Claude Code in Korea grew sixfold over four months — a developer-led surge that mirrors the tool's trajectory in other markets. Separately, Korean alternative-data trackers reported earlier this spring that Claude had overtaken ChatGPT in the country's paid generative-AI market for the first time, a milestone Anthropic is clearly eager to convert into durable enterprise revenue.

"South Korea is a country with great ambitions for AI that fits the company naturally," international managing director Chris Ciauri told reporters at the launch, citing shared philosophical ground with Korea's AI Framework Act, one of the few comprehensive national AI laws in force anywhere. Choi, for his part, said the AI transition was the most consequential of the multiple technology shifts he had witnessed across three decades in the industry, and pledged to "grow together" with Korean enterprises, institutions and research labs.

The telecom that clouded the day

The launch's awkward backdrop was an export-control order that landed on June 12, when the Trump administration directed Anthropic to suspend foreign nationals' access to its frontier Fable 5 and Mythos 5 models, citing national security. The trigger, according to multiple reports, was Washington's discovery that a South Korean telecommunications company suspected of ties to China had gained early access to Mythos 5 through Project Glasswing, Anthropic's cybersecurity consortium.

SK Telecom — one of Anthropic's closest Korean partners, having invested $100 million in the company in 2023 — was among roughly 150 organizations granted Glasswing access to Mythos before that access was reportedly revoked at Washington's request. The fallout rippled wider than one carrier: work to onboard the Korea Internet & Security Agency (KISA), Samsung Electronics and SK Hynix to the second phase of Glasswing was also halted.

Ciauri sought to defuse the issue without dwelling on it, declining to elaborate on Washington's reasoning and pointing reporters to the company's official blog for its formal position. He framed the underlying concern — a narrow jailbreak vulnerability — as the kind that "could commonly appear in models released over the past six months," and struck a confident note on the timeline.

"We are very confident that in the coming days, the models will become available again," Ciauri said, predicting the restriction would be lifted in short order and that the affected systems would return for Glasswing's second stage.

The APAC enterprise land grab

The Seoul opening is the clearest signal yet of Anthropic's strategy in Asia-Pacific: chase enterprise and developer adoption through deep integrations and cloud partnerships rather than consumer reach. Claude is already distributed through AWS Bedrock and Google Cloud's Vertex AI, with the company signaling plans to add Microsoft Azure and to court system-integration partners like LG CNS that can package Claude into large corporate deployments. Anthropic is also weighing domestic infrastructure and data-residency options to satisfy Korea's regulatory environment — a notable concession to the in-country data demands that often decide enterprise contracts in the region.

That posture sets up a direct contest with OpenAI, which has pursued its own aggressive expansion across Korea and Japan. The difference in emphasis is telling. Where OpenAI's brand strength rests heavily on ChatGPT's consumer ubiquity, Anthropic is leaning into Claude Code, agentic tooling and back-office deployments — betting that the stickiest, highest-margin AI revenue will come from engineering organizations and regulated enterprises rather than from chat subscriptions. The reported flip of Korea's paid market in Claude's favor, if it holds, suggests the wager has early traction.

It also arrives as Anthropic eyes a possible IPO, lending the international footprint added strategic weight: a diversified, enterprise-heavy revenue base across multiple geographies is exactly the story a public-markets debut would want to tell.

What to watch

The immediate test is whether Ciauri's "coming days" prediction holds. If Fable 5 and Mythos 5 access is restored quickly and Glasswing's second phase proceeds, the export-control episode becomes a footnote. If it drags, it raises uncomfortable questions for every Korean customer about the reliability of a frontier vendor whose top models can be switched off by a foreign government overnight.

Beyond that, watch three things: whether Anthropic follows through on domestic data residency, which would unlock the most security-conscious accounts; whether the government MOU with the science ministry produces concrete policy or procurement wins; and whether the sixfold Claude Code growth curve flattens or compounds. Korea has quietly become a proving ground for which of the U.S. AI labs can actually convert enthusiasm into enterprise contracts — and this week, Anthropic moved first.