OpenClaw Craze Is Driving Next Phase of AI Development, Insiders Say
Listen to the full version
The meteoric rise of open-source artificial intelligence (AI) agent OpenClaw is forcing a reset across the industry as developers strive to build viable monetization models and new infrastructure for agents designed to autonomously execute complex tasks, industry executives said at a recent forum.
Users now expect large language models to execute actual work rather than simply engaging in conversations, Zhang Peng, chief executive officer of Zhipu AI, said during a panel discussion at the Zhongguancun Forum on Friday. This requires large models to independently plan long-term tasks, continuously debug during task execution and process massive multimodal data.
Unlock exclusive discounts with a Caixin group subscription — ideal for teams and organizations.
Subscribe to both Caixin Global and The Wall Street Journal — for the price of one.
- RELATED
- DIGEST HUB
Digest Hub
Back
- OpenClaw's rise is driving major changes in AI, shifting focus from conversation to complex task execution, causing soaring demand for computing power.
- Models like Zhipu AI's GLM-5-Turbo can handle up to 72-hour workflows, but costs have risen 20-50% due to high computational needs.
- Technical and security challenges remain, requiring redesigned infrastructure and cohesive software-hardware ecosystems to fully enable autonomous AI agents.
- Zhipu AI
- Zhipu AI's CEO, Zhang Peng, emphasizes the industry's shift from conversational AI to models capable of executing complex tasks. To address this, Zhipu AI launched the GLM-5-Turbo model, enhancing agent capabilities for uninterrupted workflows. The company increased prices for this model, explaining that complex tasks consume significantly more computing power, and higher prices ensure sustainable development.
- Infinigence AI
- Infinigence AI (行深智能) is an AI company co-founded by Xia Lixue. They are experiencing a rapid increase in token consumption (doubling every two weeks since late January) due to the shift towards complex task execution by AI agents. To address the resulting computational resource shortage, Infinigence AI integrates various homegrown chips and computing clusters, aiming to build an efficient "token factory." They foresee a future where infrastructure is specifically designed for AI agents, with AI managing its own computational resources.
- Xiaomi Corp.
- Luo Fuli, head of **Xiaomi Corp.'s** MiMo large model team, highlights how OpenClaw activates the upper limits of pre-trained models by placing them in long-term task execution environments. This allows models to continuously debug and optimize, shifting the competitive focus toward processing long contexts at high speeds and low costs as tasks become more complex. Luo predicts this self-evolution could significantly accelerate scientific research milestones.
- PODCAST
- MOST POPULAR
Sign in with Google
Sign in with Facebook
Sign in with 财新