WAIC重磅发布！云天励飞亮出芯片、万卡集群、软件栈完整算力版图
Quick Answer
Yuntian Lifei unveiled its AI inference infrastructure roadmap at WAIC 2026, introducing three high-performance chips—DeepVerse100P, DeepVerse100D, and DeepVerse100L—targeting diverse inference scenarios.
Quick Take
The company aims for a long-term goal of 'one cent per hundred billion tokens' to optimize AI computation costs, marking a shift towards a comprehensive system-level approach in AI inference.
Key Points
- Introduced three chips optimized for different inference stages: DeepVerse100P, D, and L.
- Aims for 'one cent per hundred billion tokens' to reduce token generation costs.
- Focuses on system-level challenges in large-scale heterogeneous clusters.
- Developed the IFWA software stack for model adaptation and optimization.
- Initiated the '1001 Plan' to foster industry-wide collaboration on AI standards.
DeepSignal Analysis
What happened
At WAIC 2026, Yuntian Lifei introduced its AI inference infrastructure roadmap, featuring three chips: DeepVerse100P, DeepVerse100D, and DeepVerse100L. These chips are designed for various inference scenarios and aim to reduce AI computation costs to 'one cent per hundred billion tokens.' This marks a strategic shift towards a comprehensive system-level approach in AI inference.
Key evidence
- Yuntian Lifei unveiled three high-performance chips—DeepVerse100P, DeepVerse100D, and DeepVerse100L—targeting different stages of AI inference.
- The company aims to achieve a long-term goal of 'one cent per hundred billion tokens' to optimize AI computation costs, reflecting a broader industry consensus.
- The AI inference infrastructure includes a software stack called IFWA, which supports model development and optimization, indicating a move towards integrated solutions.
Why it matters
Yuntian Lifei's developments highlight a significant shift in the AI industry from isolated chip performance to a holistic system approach. This transition is crucial as AI applications become more complex and demand efficient resource utilization. The focus on reducing token generation costs aligns with industry trends towards optimizing AI operations, which could influence competitive dynamics among AI infrastructure providers.
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from WebSearch (Tavily)
See more →Stop just chatting with AI. Learn to build production-ready software in ...
The 2026 Bootcamp offers hands-on training in building production-ready software using Generative AI, applications, and AI agents, emphasizing practical skills over casual interaction with AI. Participants will learn to develop applications like Cursor AI, preparing them for real-world challenges in AI development.
