South Korean chipmaker DeepX is working with Intel to integrate its AI processors into the OpenVINO ecosystem, CEO Kim Lok-won told Yonhap. The company is also discussing cooperation with AMD and developing software to manage fleets of AI-enabled robots, a push that could take its chips beyond individual devices and into factory systems.
Watch Desk analysis
What happened
Kim said DeepX and Intel are optimising models and software development kits so the company’s neural processing units can work more smoothly with Intel systems. He expects an initial result early next year. The work includes integrating DeepX’s NPUs into OpenVINO, Intel’s open-source toolkit for AI inference and deployment.
DeepX is in early discussions with AMD after Kim met CEO Lisa Su in South Korea. Kim said AMD’s technical team had followed up, but the companies are preparing for further talks, not announcing a signed partnership.
Kim also said DeepX plans to complete a chip design tape-out using Samsung’s 2-nanometre process this month, with first chips expected around March or April. Later this year, the company plans to unveil DX-Newton, software designed to manage robot operations and distribute AI models across fleets. These plans and timelines are Kim’s account in an interview published by Yonhap on 18 October.
Why it matters
DeepX is trying to make AI processing work across a mix of hardware and software, from edge devices to robots. OpenVINO integration could make its chips easier to use within Intel’s existing ecosystem; DX-Newton points to the less glamorous, very necessary problem of updating models across many robots without installing them one by one.
The partnerships are at different stages. Intel cooperation is under way, while AMD discussions remain early. That distinction matters: an ecosystem opportunity is not yet a shipment or a customer deployment.
Our read
The interesting part is not simply that another chip startup wants to sell NPUs. DeepX is pursuing the software links and fleet-management tools that could make specialised processors useful in real products. OpenVINO integration and the planned robot software are the concrete tests to watch, rather than the company’s “first mover” ambition, which is naturally easier to announce than to prove.
What to watch
- Whether DeepX and Intel deliver the expected initial OpenVINO integration early next year.
- Whether AMD discussions become a defined technical partnership.
- Whether the Samsung 2-nanometre chips arrive on the stated timetable.
- What DX-Newton can do when DeepX unveils it, and whether customers adopt it for robot fleets.
Discussion spark: For edge AI to take off, what matters more: chips that use less power, or software that makes them straightforward to deploy and manage?
Sources and evidence
- Source update (17 October 2026, 22:00 UTC)
- Source update (17 October 2026, 22:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.