Huawei said in a new technology report that autonomous software agents are expected to dominate global AI token traffic within the next decade, accounting for in excess of 90% of that load by 2035. The Chinese technology and telecoms firm also reiterated an earlier projection that as many as 900 billion AI agents could be active globally by 2035, a population roughly 100 times the projected number of people.
The company's study, titled "Intelligent World 2035: Turning Vision into Action," describes a future in which agents continuously sense their environment, reason about it, make decisions and invoke external tools. That pattern of persistent, automated processing would push annual consumption of AI tokens - the units models process when taking in and producing information - sharply higher, Huawei said.
Rather than calling for a slowdown in frontier development, the report assumes broad and rapid growth in agent capabilities and deployment. It lists security and privacy protection for autonomous agents among 10 core technologies that Huawei regards as necessary to support that expansion.
The report's stance diverges from calls by some leading AI researchers and U.S. technology executives who have urged slowing the pace of frontier development. Those requests followed incidents involving rogue agents and raised warnings about future systems that could autonomously operate physical hardware and acquire resources, creating scenarios that might be hard for humans to control. Huawei's report does not advocate pausing development, instead starting from the premise that agent use will widen and emphasizing technical safeguards.
Huawei framed the transfer of AI capabilities into the physical world as essential for achieving artificial general intelligence (AGI), a term the report uses to refer to systems able to perform a broad range of tasks at or beyond human levels. The company argues that embedding agents in real-world systems will be a key pathway toward that outcome.
Practical steps toward managing agent risks are already part of Huawei's commercial activity. The company said it launched an autonomous security-operations platform in April that uses sensing and enforcement agents to detect and respond to cyber threats. Huawei has also published proposals for "AI firewalls" intended to shield enterprise AI deployments.
Guo Ping, chairman of Huawei's supervisory board, told employees in remarks released on Tuesday that the company views AI as its "biggest opportunity." He said Huawei aims for its computing infrastructure to play a role in China similar to that of major U.S. chipmakers in their markets, highlighting the firm's focus on scaling compute capabilities alongside agent deployment.
The report emphasizes two intertwined challenges. First, token consumption will expand dramatically if agents operate continuously, which implies much larger requirements for data-center computing and related infrastructure. Second, widespread agent use raises new security and privacy needs, prompting Huawei to include protections for autonomous agents among a short list of essential technologies.
Huawei's outlook reflects a broader policy and industry divide: in some parts of the U.S. technology sector there are active debates about coordinated limits on the speed of progress in frontier capabilities, while China has generally prioritized rapid adoption of AI across its economy coupled with efforts to develop technical standards to manage emerging risks.
The report lays out an expectation of rapid change rather than a plea for restraint, and connects that expectation to concrete product and standards work the company is pursuing. It does not, however, prescribe policy timelines or regulatory measures beyond the technical protections and platforms Huawei describes.