Former ByteDance engineer: Domestic chips exist but are not used for training, ByteDance mainly uses H20

robot
Abstract generation in progress
ME News, April 24 (UTC+8), according to Beating monitoring, Zhang Chi mentioned ByteDance’s chip situation in an interview. ByteDance is currently mainly using NVIDIA H20 (the compliant version for the Chinese market), with the fastest chips prioritized for pre-training and post-training teams, while other teams use H20. He also mentioned that ByteDance is purchasing NVIDIA’s next-generation chips (he said something like “B300,” but the wording is uncertain), “but definitely not in mainland China.” Regarding domestic chips, Zhang Chi said that ByteDance Seed does have domestic chips, but “I don’t think any team that wants to speed up iteration is really using them,” and for training, “definitely not.” In inference deployment scenarios, some teams may consider domestic chips. When asked whether DeepSeek is using Huawei chips, he said, “That’s what the news says, but I’m not sure.” He emphasized that every major Chinese company faces challenges in chip procurement. These are Zhang Chi’s personal observations based on his experience inside ByteDance and do not represent the situation of all Chinese AI companies. (Source: BlockBeats)
View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pinned