Huawei said on Wednesday that autonomous agents - software entities that monitor environments, reason, decide and use external tools - will generate over 90% of global AI token traffic by 2035. The company warned this shift will drive steep increases in computing demand and necessitate novel security systems to govern independently operating software.
The projection appears in Huawei's "Intelligent World 2035: Turning Vision into Action" report. The document anticipates a sharp rise in annual consumption of AI tokens, which Huawei describes as the units processed by AI models when they receive and generate information. According to the report, this growth will be driven by agents operating continuously, using sensing and reasoning to take actions and interact with external tools.
Huawei does not call for slowing AI development in the report. Instead, it assumes that agent capabilities and deployments will expand substantially over the next decade and lists security and privacy protection for autonomous agents as one of 10 technologies necessary to support that expansion.
The recent forecast builds on an earlier Huawei projection that as many as 900 billion AI agents could be active globally by 2035 - a figure the company says is about 100 times the projected human population. In its analysis, Huawei also argues that embedding AI into the physical world will be a necessary step toward achieving artificial general intelligence, or AGI - systems capable of performing a broad range of tasks at or beyond human levels.
Alongside the forecast, Huawei highlighted concrete product moves aimed at the emerging landscape. In April the company introduced an autonomous security-operations platform that uses sensing and enforcement agents to detect and respond to cyber threats. Huawei has also published proposals for AI firewalls intended to protect enterprise AI systems.
Guo Ping, chairman of Huawei's supervisory board, told employees in comments released on Tuesday that the company views AI as its "biggest opportunity" and that Huawei aspires for its computing infrastructure to become China's equivalent of the U.S. chipmaker Nvidia.
What this means - If Huawei's projections play out, most AI workload traffic will be machine-generated and continuous rather than episodic user interactions. That implies large, sustained compute demand and a need for security architectures that can manage software agents operating with autonomy.
What Huawei is doing - The company is both forecasting the trend and investing in tools and proposals aimed at securing agent-driven deployments, from an autonomous security-operations platform to suggested enterprise AI firewalls.