Huawei Forecasts AI Agents Will Drive Over 90% of Token Traffic by 2035

Huawei expects autonomous agents to generate more than 90% of global AI token traffic by 2035, a shift that would demand enormous increases in computing capacity and new safeguards to keep independent software secure and controllable.

AI tokens are the units that models process when receiving and generating information. Their annual consumption is projected to rise sharply over the next decade as agents continuously perceive their surroundings, reason, make decisions and call external tools, the Chinese technology and telecoms company said on Wednesday.

The forecast appears in Huawei's "Intelligent World 2035: Turning Vision into Action" report, which builds on an earlier estimate that 900 billion AI agents could be active globally by 2035 — roughly 100 times the projected human population. The report lists security and privacy protection for autonomous agents among ten technologies needed to support this expansion.

Huawei's outlook contrasts with a debate in parts of the United States technology industry, where leading researchers and executives have called for slowing frontier development after rogue agent incidents and warnings that future systems able to autonomously operate hardware and acquire resources could become difficult for humans to control. The Huawei report makes no such call to slow development, instead assuming that agent capabilities and deployment will expand dramatically.

The difference reflects a wider divide over AI pacing. While sections of the U.S. technology sector debate coordinated limits on how quickly frontier capabilities improve, China has generally sought to accelerate AI adoption across its economy while building technical standards around emerging risks.

Huawei said moving AI into the physical world would be essential to achieving artificial general intelligence, or AGI — systems generally able to perform a broad range of tasks at or beyond human levels.

The company is also deploying agents as part of its security response. In April it launched an autonomous security-operations platform that uses sensing and enforcement agents to detect and respond to cyber threats, and it has published proposals for AI firewalls aimed at protecting enterprise AI systems.

Guo Ping, chairman of Huawei's supervisory board, told employees in remarks released on Tuesday that he viewed AI as the company's "biggest opportunity" and wanted its computing infrastructure to become China's equivalent of U.S. chipmaking giant Nvidia.