ByteDance developing custom CPU chips to support AI rollout: Report
ByteDance is developing custom CPU chips to support its AI infrastructure amid surging chip prices and supply shortages, targeting deployment in its servers and data centers for agent-based products like Coze. The company is exploring two architectures—Arm and open-source RISC-V—to hedge risks before committing to large-scale manufacturing, following a trend among tech giants like Google, Amazon, and Microsoft.
Chinese tech giant ByteDance is designing its own central processing units (CPUs) to address rising costs and supply constraints in the AI sector, according to three sources familiar with the matter. The move aligns with industry trends, as AI models increasingly rely on CPUs alongside Nvidia’s dominant graphics chips, creating shortages that have pushed global companies—including Alphabet’s Google, Amazon, and Microsoft—to develop custom silicon. ByteDance plans to integrate its proprietary CPUs into its servers and data centers to support internal operations, particularly for its upcoming rollout of agent-based AI products, such as the Coze platform. The company has engaged external partners to assist with chip design and secure manufacturing capacity, though the project remains in early stages. Sources noted ByteDance is evaluating two architectures: one based on Arm (owned by SoftBank) and another on the open-source RISC-V instruction set, a common strategy to mitigate risks before large-scale production. The push for custom chips reflects broader industry challenges, including Intel’s reported six-month delivery delays for server CPUs and AMD CEO Lisa Su’s warning about persistent supply tightness. ByteDance currently sources CPUs from Intel and AMD, but price hikes—ranging from 10% to 35% quarter-over-quarter—have accelerated its push for in-house alternatives. The company’s decision underscores the growing economic incentives for tech firms to bypass traditional suppliers and control their own hardware development. ByteDance did not respond to Reuters’ request for comment, and neither did Arm, which was approached for clarification. The initiative positions ByteDance alongside other hyperscalers investing in proprietary silicon to optimize costs and performance for AI workloads, amid a competitive landscape reshaped by Nvidia’s dominance in AI acceleration.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.