根据研究机构 Omdia 的预测，随着中国科技公司在办公和工业运营中日益广泛地部署自动化工具，亚洲、澳大利亚和新西兰的人工智能代理软件市场到 2030 年将扩大 14 倍以上。
The artificial intelligence agent software market across Asia, Australia and New Zealand is set to expand more than 14-fold by 2030, according to research firm Omdia, as Chinese tech companies increasingly deploy automated tools across office and industrial operations.
Omdia 在周四于北京举行的年度活动中预测，包括中国大陆和香港在内的区域市场收入将在 2030 年达到 112 亿美元，高于今年估计的 7.82 亿美元。同期，代理式 AI 在该区域 AI 软件收入中的占比预计将从 1.7% 上升至 12.3%。
During its annual event in Beijing on Thursday, Omdia forecast that regional market revenue, which included mainland China and Hong Kong, would reach US$11.2 billion in 2030, up from an estimated US$782 million this year. Over the same period, the agentic share of the region’s AI software revenue was projected to rise to 12.3 per cent from 1.7 per cent.
AI 代理利用模型自主规划并执行多步骤任务——例如数据分析、报告撰写和软件操作——使用户能够将原本需要反复下达指令的工作委托出去。
AI agents use models to autonomously plan and carry out multi-step tasks – such as data analysis, report writing and software operations – allowing users to delegate work that would otherwise require repeated instructions.
中国国务院在其“AI+”倡议下设定目标，到 2027 年和 2030 年，新一代智能终端和智能代理的渗透率分别超过 70% 和 90%。
China’s State Council has set targets for the penetration of next-generation intelligent devices and agents to exceed 70 per cent by 2027 and 90 per cent by 2030 under its “AI Plus” initiative.
主要中国科技公司正加速部署，以抓住本国市场不断增长的采用需求。
Major Chinese tech companies are accelerating deployment to capture growing adoption in their own market.
《南华早报》所有者阿里巴巴集团控股有限公司上月发布了 AgentCore。该企业级平台旨在帮助企业构建、运行和管理 AI 代理，包括监控其性能并将其连接至现有业务系统。
Alibaba Group Holding, owner of the South China Morning Post, unveiled AgentCore last month. The enterprise platform is designed to help businesses build, run and manage AI agents, including monitoring their performance and connecting them to existing business systems.
腾讯控股也在扩展其 WorkBuddy 代理工作空间，公司表示该工作空间可直接编辑 Word 文档、Excel 电子表格和 PowerPoint 演示文稿，并能连接 Notion 和 GitHub 等服务以检索信息和执行任务。
Tencent Holdings is also expanding its WorkBuddy agent workspace, which the company said can edit Word documents, Excel spreadsheets and PowerPoint presentations directly, and connect to services including Notion and GitHub to retrieve information and carry out tasks.
惠誉评级在 8 月的一份报告中预计，包括阿里巴巴、腾讯和字节跳动在内的中国科技巨头今年的合计资本支出将较 2025 年的 650 亿美元翻倍多，达到约 1400 亿美元。
In an August report, Moody’s Ratings expected the combined capital spending of Chinese tech giants, including Alibaba, Tencent and ByteDance, to more than double to about US$140 billion this year from US$65 billion in 2025.
穆迪表示，云服务仍是AI变现的主要渠道，成本更低的开源模型正加速中国企业的采用。不过，该机构指出，激烈的价格竞争以及中国云业务规模较小，使其不如美国同行那样能从云现金流中为进一步的AI基础设施扩张提供资金。
Cloud services remain the main channel for monetising AI, Moody’s said, with lower-cost open-source models accelerating enterprise adoption in China. However, intense price competition and the smaller scale of Chinese cloud businesses left them less able than their US peers to fund further AI infrastructure expansion from cloud cash flow, it said.
对于采用AI的企业而言，最大化投资回报仍是首要任务。“使用更多token并不意味着产生更多价值，”Omdia首席分析师苏连杰表示，token是指AI模型处理的数据单位，也是向客户收费的常见依据。
Maximising return on investment remains a priority for enterprises adopting AI. “Using more tokens does not mean producing more value,” Omdia chief analyst Su Lian Jye said, referring to the units of data processed by AI models and a common basis for charging customers.
苏连杰说，企业需要评估AI输出的价值，并通过将任务与合适的模型相匹配来提高效率。他表示，中国公司也在利用开源模型吸引开发者和用户，尽管部署这些模型仍需要大量的工程专业知识。进军AI智能体领域，也在云服务商之间开辟了新的竞争战场。
Su said businesses needed to assess the value of AI’s output and improve efficiency by matching tasks with appropriate models. Chinese companies were also using open-source models to attract developers and users, although deploying them still required substantial engineering expertise, he said. The push into AI agents is also opening a new front in competition among cloud providers.
苏连杰表示，整合模型、软件和算力基础设施可能会增加客户转移成本，因为将基于智能体的应用迁移到另一家云平台，可能需要同时转移模型及其背后的优化工作，从而使迁移变得更加复杂。
Integrating models, software and computing infrastructure could make customers harder to win away, as moving an agent-based application to another cloud could require transferring both the model and the optimisations behind it, making migration more complicated, Su said.