China’s most influential search and AI company is no longer talking about artificial intelligence as a niche technology. Baidu co-founder Robin Li is sketching a future in which AI saturates factories, offices, classrooms and homes, turning what he calls “new productive forces” into the backbone of the country’s growth model. His message is blunt: China is not just building bigger models, it is preparing to flood everyday life with AI-powered applications and services.
That ambition is backed by a rare alignment of corporate strategy and state planning. Beijing’s national AI agenda, Baidu’s latest Ernie models and Li’s own long‑running evangelism for automation all point in the same direction, toward a society where most digital interactions are mediated by intelligent systems. The question is not whether this diffusion will happen, but how quickly and on whose terms.
From “AI race” to AI everywhere
Robin Li has been trying to shift the conversation away from an arms race narrative and toward a competition over usefulness. In a recent interview he contrasted talk of an AI “race” with what he described as two diverging paths, explaining that, On the Chinese side the focus is on making models more efficient rather than simply larger. That framing matters, because it underpins his argument that the real contest will be won by whoever can embed AI most deeply into real‑world workflows, not by whoever tops a benchmark chart.
Li has been consistent about this pivot to practicality. He has said publicly, “Don’t focus on models; focus on applications,” describing how Baidu spent the past two or three years turning its research into products that people actually use. In his telling, that shift has already reshaped the company’s priorities, with teams pushed to ship tools for consumers and enterprises rather than chase abstract technical milestones, a stance he reinforced when he recalled that he publicly said those words and then reorganized Baidu around them.
Baidu’s Ernie 5.0 and the hardware behind the vision
The most visible symbol of this strategy is Baidu’s latest large model, Ernie 5.0, which sits at the core of its consumer chatbot and enterprise stack. Earlier this month the Chinese tech giant rolled out a new version of its assistant, Ernie Bot, known domestically as Wenxin Yiyan, describing it as a native system that uses full‑modality unified modeling to handle text, images and other inputs in a single architecture. The company framed the upgrade as proof that a Chinese champion can still push the frontier of generative AI despite export controls on advanced chips.
Under the hood, Baidu has touted Ernie 5.0 as a system with “2.4 Trillion” parameters, while emphasizing that only a fraction of those parameters are activated during inference to keep costs manageable. In its own description, the company said Baidu Unveils ERNIE 5.0 with 2.4 Trillion Parameters, Ushering in a New Era of Multimodal AI, a phrase that captures both the scale of the model and the company’s desire to present it as a platform for everything from search to autonomous driving. The technical feat is not just about bragging rights, it is the engine that Baidu hopes will power thousands of downstream applications.
“New productive forces” and the state’s AI Plus blueprint
Li has started to describe AI as the core of China’s next growth engine, aligning his rhetoric with Beijing’s push to upgrade the country’s industrial base. In a recent speech he cast AI as the driver of China’s “new productive forces,” arguing that intelligent software and data‑driven services will do for the digital economy what heavy industry once did for manufacturing. That argument landed at a moment when Baidu founder Robin Li could point to concrete momentum, including a surge in AI‑related revenue that has helped convince policymakers that this is more than a branding exercise.
Official policy is moving in the same direction. Under an “AI Plus” implementation guideline issued last year, authorities set explicit penetration targets for artificial intelligence across key sectors, aiming for 60 per cent adoption by 2027 and 90 per cent by 2030 in areas such as manufacturing, finance and public services. The document, which described the Plus initiative as a way to weave AI into everything from logistics to education, effectively gives companies like Baidu a mandate to build tools that can be slotted into government and enterprise systems at scale.
Explosive growth in applications, from chatbots to factories
Li’s bet is that the real payoff will come from a wave of AI applications that sit on top of these models and policies. He has predicted that there will be an “explosive” increase in AI apps, telling investors that Baidu’s own ecosystem would see a surge in services built on its platforms during 2025. In that forecast, Baidu’s BIDU CEO framed the coming year as a turning point when developers would finally move from experimentation to products that millions of users rely on every day.
That vision dovetails with broader expectations in the enterprise world. Analysts at one major technology firm have argued that 2025 might be the year companies move from pilots to large‑scale adoption of AI agents, and they have warned that executives will need to balance speed with risk as they roll out these systems. The idea that “2025 might be the year we go from experiments to large‑scale adoption” of such agents, as one analysis put it, aligns neatly with Li’s own timeline for when Baidu’s application‑first strategy should start to show up in usage metrics and revenue.
Long‑term vision: low‑carbon growth and 3‑D intelligent agents
Li’s ambitions for AI stretch far beyond chatbots and office software. He has argued that artificial intelligence is a “transformative force” that can reshape human development over the next 40 years, including by helping societies transition to low‑carbon economies. In a speech at a major industry conference he said Baidu’s chief priority was to harness technological advances so that humanity could gradually restore itself to a low‑carbon state, a claim that framed Share of AI in climate and sustainability debates as central rather than peripheral.
He has also painted a picture of AI systems that can move through and manipulate the physical world, not just parse text on a screen. Looking ahead, Li has described a future in which intelligent agents learn in three dimensions and interact with environments in sectors such as health care, entertainment and architecture, effectively becoming co‑workers and creative partners. In one profile he was quoted as saying that Looking ahead, he envisions AI that not only interprets but also interacts with the 3‑D world, a vision that ties together Baidu’s work in autonomous driving, robotics and virtual reality.
More from Morning Overview