Baidu sets AI stack strategy with Ernie 5.0

Baidu launched Ernie 5.0 and announced in-house AI chips due in 2026 and 2027 at its Beijing showcase, outlining plans across models, chips and apps.
Baidu introduced Ernie 5.0 and previewed in-house AI chips scheduled for 2026 and 2027 at its annual tech showcase in Beijing on Thursday, to keep pace in China’s AI race.
Founder and CEO Robin Li described Ernie 5.0 as “natively omni-modal,” built to handle instructions across text, audio and visuals. He pledged rapid updates across the model lineup, telling attendees, “Intelligence itself is the greatest application, and the speed of technological iteration is the only moat.”
On stage, Baidu showed benchmarks comparing Ernie 5.0 with DeepSeek, Google’s Gemini and OpenAI’s GPT-5 on language, audio and visual tasks. Ernie 5.0 did not top every category, and the company presented the results as evidence it is keeping pace with leading systems.
Baidu has developed two chips for AI training and inference. Commercial availability is planned for 2026 and 2027, and technical specifications were not disclosed. Executives positioned the hardware within a stack that spans chips, models and applications.
New software tools accompanied the launch. The company introduced an agent-based product for industrial uses such as traffic light control, and will sell existing offerings overseas, including a digital-avatar service for streamers and a no-code development tool.
Li also addressed industry economics, arguing that most of the value should accrue higher up the stack. Without naming firms, he called the current setup “unhealthy, unsustainable” if chip suppliers capture outsized gains. “No matter how much money chipmakers make, the models built on top of the chips should generate 10 times the value, and the applications developed on top of the models should create 100 times the value,” he concluded.
Baidu led China’s early AI push but has faced pressure from ByteDance in AI-native apps and Alibaba in open-source models. After favoring proprietary systems, it open-sourced parts of Ernie following DeepSeek’s rise. The company also reversed course on AI video generation after rivals drew revenue, and a weak ad market has weighed on results, with an 8% revenue decline expected for the September quarter.
The material on GNcrypto is intended solely for informational use and must not be regarded as financial advice. We make every effort to keep the content accurate and current, but we cannot warrant its precision, completeness, or reliability. GNcrypto does not take responsibility for any mistakes, omissions, or financial losses resulting from reliance on this information. Any actions you take based on this content are done at your own risk. Always conduct independent research and seek guidance from a qualified specialist. For further details, please review our Terms, Privacy Policy, and Disclaimers.







