Qualcomm is in talks with ByteDance over a custom chip design project centered on a VPU, or video processing unit, with mass production targeted for the end of 2026. Reuters, citing four people familiar with the matter on June 24, reported that ByteDance could become an early customer for Qualcomm’s chip design services business if the discussions lead to an agreement. After the report, Qualcomm shares climbed more than 4% in after-hours trading.
Discussions center on video workloads and AI vision tasks
According to the report, the proposed chip would be designed for video encoding, image enhancement, and AI visual processing. That focus matches ByteDance’s video-heavy operations, including TikTok and the Doubao large model. The talks are still ongoing, and no final outcome has been settled.
People familiar with the matter also said ByteDance is evaluating other potential partners at the same time. That leaves room for parallel discussions with additional chip design companies, meaning Qualcomm is not yet locked in as the sole option.
After-hours move followed a weak session for AI chip stocks
The market reacted quickly. Qualcomm shares jumped more than 4% in after-hours trading after the news surfaced. The move came one session after a broad selloff in U.S. AI chip names on June 23, when Intel fell more than 8%, AMD lost over 6%, and both Qualcomm and TSMC dropped more than 5%. The ByteDance report helped lift Qualcomm against that backdrop.
ByteDance extends its chip push beyond general compute
The discussions fit into ByteDance’s wider AI push. On June 23, ByteDance CEO Liang Rubo said at the Volcano Engine Force conference that the company would narrow its business scope and focus fully on AI, while making Volcano Engine PaaS part of its core business foundation. ByteDance has already developed several CPU chips for AI computing. A custom VPU would point to a broader chip strategy, moving from general compute toward processors built for video-centric workloads.
For ByteDance, that would align with demand from large-scale video processing. For Qualcomm, a deal would give its chip design services effort a major early customer and offer a real test of whether its capabilities can extend beyond mobile chips into a wider range of AI hardware applications.