Anthropic's valuation doubled to $900 billion in two months, and Nvidia released an AI agent model with 9x efficiency gains—the focus of the LLM race is shifting from "who has the most parameters" to "who can do the actual work."
What this is
Three events this week collectively point to the moment of AI implementation. First, Anthropic plans to raise funds at a $900 billion valuation, surpassing OpenAI to become the world's most valuable AI company; its annualized revenue skyrocketed from $1 billion to $30 billion in 15 months. Second, Nvidia released Nemotron 3 Nano Omni, an open-source omni-modal model (capable of processing text, voice, and vision simultaneously), specifically targeting AI Agents (AI systems that autonomously execute tasks rather than just answering questions). This model incorporates GUI training, allowing AI to directly operate software and fill out forms, achieving 9x the efficiency of its peers. Third, the CPC Central Committee Politburo meeting proposed for the first time the comprehensive implementation of the "AI+" action, putting AI on par with manufacturing as core economic work.
Industry view
We note that capital, compute giants, and the state apparatus are reaching a rare consensus: AI must enter industries and do the work. Google and Amazon promised to invest $65 billion in Anthropic, betting that Claude + Agent will define the next generation of enterprise software; Nvidia stepping in to build agent models themselves shows that merely selling shovels is no longer enough—they must control the software ecosystem. However, we must pay attention to the underlying risks: Anthropic's valuation growth far outpaces its revenue growth, and the sustainability of its $30 billion revenue heavily depends on whether enterprise clients will truly pay for agents. If implementation falls short of expectations, the burst of this bubble will come at a massive cost. Meanwhile, Nvidia doing both compute and models could trigger an open-source ecosystem rebellion from its core major clients. Domestically, the "AI+" mandate also means regulation will tighten in tandem with policy dividends.
Impact on regular people
For enterprise IT: The shift goes from purchasing chatbots to purchasing agents that can operate business systems. The "AI+" policy will bring a dual driver of subsidies and compliance.
For the individual workplace: AI will transition from an "encyclopedia that answers questions" to a "digital employee that operates software." The core skill for white-collar workers will shift from writing prompts to managing digital employees.
For the consumer market: The open-sourcing of omni-modal models means voice and visual interactions will become standard. Future smart hardware will better understand its environment, leading to more natural interactions.