In its "Intelligent World 2035" report published Wednesday, Huawei forecasts that autonomous artificial intelligence agents will account for more than 90% of global AI token consumption by 2035. The company says this shift will demand substantial expansion of computing capacity and new security and data-protection technologies.
Huawei defines "AI tokens" as the units of information processed and generated by models. The report builds on an earlier Huawei projection that as many as 900 billion active AI agents could be operating worldwide by 2035 — roughly one hundred times the expected human population at that time.
According to the company, the surge in token use is driven by agents that continuously sense their environment, draw inferences, make decisions and invoke external devices, producing sustained high volumes of token traffic. To address that future, Huawei lists ten key technologies it considers essential for securing and protecting autonomous agents, with particular emphasis on privacy and specialized security measures.
Huawei’s outlook highlights a contrast in global approaches. While some AI researchers and executives in the United States have recently urged slowing or more coordinated limits on the most advanced capabilities following incidents involving uncontrolled agents, China — and Huawei in particular — is prioritizing accelerated deployment across the economy alongside the development of technical standards to manage risks.
The company has already taken concrete steps on agent-based security: in April it launched an autonomous security operations platform that uses sensing and actuator agents to detect and mitigate cyber threats. Huawei has also published proposals for firewalls designed to protect enterprise AI systems.
Kuo Ping, chairman of the Huawei Supervisory Board, described artificial intelligence in an internal statement as the company’s "biggest opportunity," and said Huawei aims to become China’s Nvidia through its computing infrastructure.
Why it matters: if Huawei’s projections materialize, the AI ecosystem would undergo structural change — a dramatic rise in token demand would require new hardware capacity, energy and data-protection solutions, and make safe governance of autonomous systems a central policy and technical challenge.
(This article was prepared with the assistance of an AI tool; a journalist edited and verified the final content.)
