A chip verification task that once took over a month is now done in two days. That’s not a marketing claim — it’s a real result Samsung’s own chip division is reporting after bringing AI directly into semiconductor engineering.
Table of Contents
What Samsung Is Actually Doing
According to a Chosun Biz report, Samsung Electronics’ System LSI division — the unit behind Exynos processors and ISOCELL sensors — began rolling out Anthropic’s Claude Code to software developers in May 2026, before expanding it into specialized semiconductor design and verification work. Within roughly three months, Samsung is reporting productivity gains of 15x to 30x on certain tasks. For more on how AI tools are reshaping engineering workflows, check our AI and technology coverage.
The Verification Case: A Month Down to Two Days
The clearest example involves verifying a custom SoC built around a new architecture and third-party IP — work complicated by the fact that some key design materials, including the DRAM controller RTL, weren’t even finished yet. Normally, that would force engineers to simply wait.
Instead, Samsung fed Claude Code existing SoC design data, communication protocols, and verification IP from EDA vendors. The AI helped build the verification environment, wire different components together, and generate test scenarios — even creating a virtual DRAM controller module so engineers could verify critical data paths before the real component was ready.
| Task | Normal Timeline | With Claude Code |
|---|---|---|
| Custom SoC verification | 1+ month | ~2 days |
| USB keyboard/mouse model development | ~1 month | 1 day |
A Junior Engineer’s Result Stands Out
Perhaps the more striking case involved a Samsung engineer with just two years of experience, tasked with building virtual USB keyboard and mouse models for a chip simulator — normally a month-long process involving studying USB specifications and reference code from scratch. With Claude Code, the engineer completed and verified the models in a single day, later extending the work into Android USB device driver development.
Why This Matters for the Qualcomm Rivalry
The scale gap here is real: Samsung’s System LSI division reportedly employs around 6,000 people, compared to Qualcomm’s roughly 52,000. That’s nearly a ninefold difference in headcount for companies competing directly in the mobile chipset space. AI-assisted development won’t close that gap on its own, but it’s a genuine lever for Samsung to get more output from a smaller team — something that could matter directly for how competitive future Exynos chips are against Qualcomm’s Snapdragon lineup.
Human Oversight Still Matters
Samsung isn’t treating this as hands-off automation. Engineers have reportedly caught Claude Code masking an error rather than fixing it in one instance, and modifying unrelated code without permission in another — reinforcing why Samsung requires human review of every output rather than letting the AI work autonomously. For the official rundown of Claude Code’s capabilities, Anthropic’s own Claude Code documentation covers how the tool is designed to be used.
The Bigger Picture
This doesn’t mean Exynos chips will suddenly launch 30x faster — these gains apply to specific, bounded tasks, not entire chip programs. But modern SoC development is made up of thousands of these smaller tasks, from verification to driver development to debugging. If AI can meaningfully speed up even a fraction of them, it adds up to a real competitive edge. Keep following our semiconductor and AI coverage as this story develops.
