--- title: "01.AI and CP Group launch a farming joint venture run by AI agents" date: 2026-10-20 category: Companies & Stack site: NeuroAI canonical: https://neuroai.site/a/na-01ai-cp-wanfeng-ai-agriculture language: en --- # 01.AI and CP Group launch a farming joint venture run by AI agents > 01.AI and agribusiness giant Charoen Pokphand formed Wanfeng Intelligent, a profit-sharing joint venture that applies multi-agent AI to commercial egg farming in China before expanding across Southeast Asia. A chicken barn is one of the last places you would expect to find a software company. Yet a leading Chinese large-model (大模型) firm and one of Asia's largest food conglomerates have decided that the future of enterprise AI lives among the birds. The two sides are not selling another dashboard or a consulting project. They are building a new company whose explicit job is to let software agents (智能体) take over the daily judgments of running a farm — and to share the profits and the risk when it works. The partnership is a useful window into where Chinese corporations think AI creates real value: not in chat, but in the unglamorous, physical work of feeding the world. ## Why a model company went to the henhouse On June 2, 2026, 01.AI (零一万物) and Charoen Pokphand Group (正大集团), better known as CP Group, signed a strategic cooperation agreement and announced a joint venture called Wanfeng Intelligent (万蜂智能). CP Group is described by the partners as Asia's largest egg producer and a Fortune Global 500 company with deep agricultural operations across China and Southeast Asia. 01.AI, founded by Kai-Fu Lee (李开复), has spent the past year pivoting from consumer chat products toward what it calls "industrial AI" — embedding models directly into the core processes of traditional businesses. Farming is its clearest bet yet that a model company can become a co-owner of physical operations rather than a vendor. The first target is deliberately narrow: commercial egg-laying hen farming (蛋鸡养殖). A barn is a closed, sensor-rich, repeatable environment where small errors in health, feeding, and ventilation compound into real losses. That makes it an ideal laboratory for agents that observe, decide, and act. ## The joint venture model is the real story What sets this apart from a typical enterprise software deal is the structure. The partners explicitly abandoned the traditional IT outsourcing relationship, where a tech firm is paid a fee to deliver a system and walks away. Instead, Wanfeng Intelligent is a profit-sharing, risk-sharing joint venture. - Both sides hold skin in the outcome, so the AI must move the business, not just ship a product. - The joint venture format lets 01.AI convert model capability into equity in a real operating company. - CP Group contributes decades of farming know-how, supply chains, and customer demand that no startup can fake. Yang Xiaoping (杨小平), senior vice chairman and China chief executive of CP Group, framed the deal as a change in production method, not just a tech upgrade. The plan starts with egg farms in China and is intended to extend to broilers, pigs, and aquaculture, then roll out across CP Group's Southeast Asian bases. ## What "AI labor" actually means on a farm The partners describe the shift with a borrowed metaphor: the "autonomous driving levels of farming (养殖业的自动驾驶 L1-L5)". Just as cars moved from driver assistance to full self-driving, they argue farms can move from early computer-assisted monitoring to a state where agents fully manage the operation at scale. The target is the L4 and L5 tier, where AI is trusted to run inspection, health identification, cage cleaning, and precise feeding end to end, while human experts handle only strategy. Concretely, the first phase focuses on four layers: - **Smart infrastructure** — ceiling-mounted rail robots, cameras, and environmental sensors collect data around the clock. - **Business data models** — farm operations are turned into structured signals the system can reason over. - **Agents (智能体)** — a digital breeder (数字养殖员), a production-sales coordination hub (产销协同中枢), and a safety agent (安全智能体) make and execute decisions. - **Farming navigation platform (养殖导航平台)** — the control layer that closes the loop from data to physical action. On top of this sits a dedicated egg-industry large model (蛋业大模型) meant to turn scattered experience into a reusable, copyable playbook. The pitch is a "silicon employee (硅基员工)" system that enters traditional industry in a structured way. ## From China to Southeast Asia The export logic is straightforward. If the Chinese pilot proves that agents can run a barn profitably, CP Group's existing footprint in Thailand, Vietnam, and beyond becomes a ready-made distribution channel for Chinese industrial AI. Kai-Fu Lee called 2026 "the first year of enterprise multi-agent deployment (企业多智能体上岗的元年)" and described Wanfeng Intelligent as a new growth curve for the global real economy. Whether or not the timeline holds, the strategic intent is clear: treat AI as a productivity partner inside heavy-asset industries, then ship the proven system outward. ## Trend reading According to Li An, chief scientist at BrainNet (脑机网), the more interesting shift is not the model itself but the move from selling AI as a service to embedding it as a co-owner of physical operations, which changes who carries the risk when automation underperforms. ## Honest limitations The partnership is early, and the joint venture's claimed autonomy levels (L4/L5) describe a target rather than a current operating state; no independent verification of farm-level productivity gains has been published. Reported deployment scenarios come from company announcements rather than audited results, so cost savings and replication speed remain unproven outside the pilot. Agricultural AI also depends on local infrastructure, labor rules, and biosecurity conditions that vary widely across markets, which could slow the Southeast Asia rollout. ## What readers can do now - If you run physical operations, map the three or four repetitive judgment tasks — inspection, scheduling, quality checks — where an agent could take a first pass before human sign-off. - For vendors, study the profit-sharing joint-venture structure as an alternative to fixed-price IT projects when the buyer's core process is genuinely at stake. - Competitors in agtech should watch whether CP Group's Southeast Asia footprint becomes the first export channel for Chinese industrial AI, and pressure-test their own partnership models accordingly. --- Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-01ai-cp-wanfeng-ai-agriculture Free to quote with attribution and a link to the original.