Samsung Deploys ChatGPT Enterprise Across South Korean Workforce

Samsung Electronics is putting ChatGPT Enterprise and Codex in front of every employee in South Korea, plus everyone in its Device eXperience (DX) division worldwide. OpenAI calls it one of the largest enterprise deals in its history.

For ML engineers watching how generative AI moves from pilot to production, this is a concrete data point: a hardware giant standardizing on commercial LLM tooling across an entire national workforce, not just an innovation team.

What's being deployed, and where

The rollout covers all South Korean Samsung Electronics employees and the global DX division. Samsung says it plans to use the tools across research, manufacturing, marketing, and administration—a span that puts the same AI tooling in the hands of engineers, factory operations, and back-office staff.

That breadth is the notable part. Most enterprise AI announcements concentrate on a single function. Samsung is treating ChatGPT Enterprise and Codex as horizontal infrastructure rather than a departmental experiment.

Codex beyond developers

Codex began as a tool for writing and reviewing code. According to OpenAI, non-developers are increasingly using it to build internal tools and automated workflows—a shift that matters for how teams think about who gets access to coding agents.

A recent addition pushes that further: a record-and-replay feature. Users walk through a workflow once, and the AI repeats it autonomously afterward. For internal automation, that lowers the barrier from "write a script" to "demonstrate the task."

The usage numbers OpenAI cites give some scale. More than five million people use Codex weekly worldwide. In South Korea specifically, active users have jumped roughly 800 percent since February—the regional context that frames the Samsung deal.

The supply-chain angle

Samsung isn't just a customer here. The company already supplies OpenAI with memory chips for AI infrastructure, making this a two-way relationship: Samsung sells the hardware that runs frontier models, and now buys the software those models power.

OpenAI also names other Korean customers—LG Electronics, Krafton, Toss, and Seoul National University—suggesting the South Korean enterprise and academic market is a focus area, not a one-off.

Why the "Enterprise" tier matters

For organizations evaluating LLM tooling, the distinction between consumer ChatGPT and ChatGPT Enterprise is the deciding factor. The enterprise version is built around organizational controls and data-handling guarantees that a company the size of Samsung—working across proprietary research and manufacturing—would require before granting workforce-wide access.

A deployment at this scale implies Samsung is comfortable with the data boundaries and admin controls the enterprise tier provides. That's a useful signal for anyone making the same procurement case internally: a manufacturer with significant IP exposure cleared the security bar for company-wide rollout.

What to watch

This deal is worth tracking for a few reasons. First, it tests whether broad, cross-functional LLM access actually changes how a large workforce operates, or whether usage concentrates among a smaller set of power users—the kind of question the 800 percent regional growth figure hints at but doesn't fully answer.

Second, Codex's move into non-developer hands via record-and-replay is a pattern to watch elsewhere. If business teams at Samsung start building their own automated workflows without writing code, that reshapes assumptions about who owns internal tooling.

The takeaway for ML developers: enterprise generative AI adoption at genuine scale is happening, and the security and admin controls of the enterprise tier are clearing procurement at companies with serious IP to protect. The open question is impact—whether workforce-wide access translates into measurable productivity, or just broad availability. Samsung's rollout is a large enough sample to eventually tell us.