DeepSeek Delays AI Model Launch Due to Huawei Chip Issues
Chinese AI company DeepSeek has postponed the release of its new AI model due to technical hurdles with Huawei's Ascend chips. The delay, initially slated for May, puts DeepSeek behind competitors.
DeepSeek's R2 model faced training issues with Ascend chips, prompting a team of Huawei engineers to intervene without success. AI researcher Ritwik Gupta attributes this to 'growing pains' for Huawei's chip technology.
DeepSeek has since partnered with NVIDIA to release new hardware supporting UE8M0 FP8 data types, optimizing their latest V3.1 model. Industry insiders report that Chinese chips generally lag behind Nvidia in stability, connectivity, and software. DeepSeek now uses Nvidia for training and Huawei for inference.
Initially, DeepSeek attempted to use Ascend processors for training R2 but encountered persistent problems. The switch to UE8M0 FP8 data type hints at more powerful Chinese accelerators on the horizon, as the current top Ascend 910C does not natively support FP8.
DeepSeek's R2 model launch is delayed due to technical issues with Huawei's Ascend chips. The company is now working with NVIDIA for model training and expects improved Chinese chip technology in the future.
Read also:
- Elon Musk accused by Sam Altman of exploiting X for personal gain
- China's Automotive Landscape: Toyota's Innovative Strategy in Self-Driving Vehicles
- L3Harris' RASOR Revolutionizes Military Communications with Secure Satellite Broadband
- EU Bolsters Defense Capabilities: Orbotix Secures €6.5M for AI-Driven Drones
