What drives 140T Token calls?

Enterprise AI Agents, 70-80% cost reduction in domestic models, government AI+ strategy.

China's global share?

35-40% of global volume (US 40-45%). China focuses on inference, US on training.

~40% of inference workload via Huawei Ascend 910B and Cambricon MLU370.

China's AI Daily Token Calls Exceed 140 Trillion, Evolving from Chatbots to Decision-Making Agents

China's AI daily Token calls surpassed 140 trillion, up 300%+ YoY. Driven by enterprise AI Agent deployment, 70-80% cost reduction in domestic models, and government AI+ strategy. Paradigm shift from conversational AI to decision-making agents.

China's AI Daily Token Calls Exceed 140 Trillion: The Paradigm Shift from Chatbots to Decision-Making Agents

In March 2026, data jointly released by China's Ministry of Industry and Information Technology (MIIT) and the China Academy of Information and Communications Technology (CAICT) revealed that daily Token invocations across China's AI systems have surpassed 140 trillion, representing a more than 300% increase year-over-year from 2025. This figure not only sets a global record but, more importantly, reflects a fundamental paradigm shift in China's AI industry — from consumer-grade conversational AI applications toward industrial-grade decision-making agent systems.

The Industrial Landscape Behind 140 Trillion Tokens

Three primary drivers fuel this explosive growth in Token consumption.

First is the large-scale deployment of enterprise AI Agents. China's leading internet companies and traditional manufacturers are accelerating AI Agent deployment for customer service automation, supply chain decision optimization, and financial risk management. Unlike simple conversational AI, Agent systems may involve dozens or even hundreds of LLM calls during a single task execution (planning, reasoning, tool invocation, verification), multiplying Token consumption.

Second is the declining cost of domestic large models. DeepSeek V4, Qwen 3, Kimi 2, and other domestic models have reduced inference costs by 70-80% compared to a year ago while maintaining performance. This cost reduction directly stimulates API call volume growth, enabling SMEs previously unable to afford AI deployment to begin using AI services at scale.

Third is the government's AI+ strategic initiative. The AI+ industrial policy confirmed during the Two Sessions explicitly requires accelerated AI deployment in finance, healthcare, education, government services, and other key sectors. Government procurement of AI services has risen from under 5% in 2025 to approximately 15%.

From Conversation to Decision: Three Dimensions of Paradigm Shift

Autonomous Decision-Making: New-generation AI Agents have surpassed simple Q&A and content generation. In finance, agents autonomously execute trading strategies, conduct credit assessments, and perform risk pricing. In manufacturing, agents coordinate production scheduling across hundreds of machines, responding in real-time to order changes and supply chain disruptions.

Multi-Agent Collaboration: Complex business scenarios increasingly favor multi-agent systems. A typical enterprise AI workflow may involve planning agents, execution agents, verification agents, and monitoring agents working in concert, enabling end-to-end business process handling.

Real-Time Perception and Action: AI Agents no longer passively respond to user requests but proactively perceive environmental changes and take action. In smart city scenarios, agents continuously monitor traffic flow, weather changes, and energy consumption, automatically adjusting traffic signals, public transit scheduling, and power grid load distribution.

Infrastructure Challenges

The 140 trillion daily Token calls present unprecedented infrastructure demands. China possesses the world's second-largest AI compute cluster (after the US), but GPU supply remains affected by export controls. Domestic chips like Huawei's Ascend 910B and Cambricon's MLU370 handle approximately 40% of inference workloads but still lag NVIDIA in training efficiency.

China is aggressively pursuing inference optimization. DeepSeek's MoE architecture and Alibaba Tongyi's speculative decoding technology have increased inference throughput 3-5x on equivalent compute. The national "East Data West Computing" project continues expanding, with western data centers handling increasing non-real-time inference tasks.

Global Competition Landscape

China's 140 trillion daily Token calls represent an estimated 35-40% of global volume (US approximately 40-45%). The application structure differs significantly: US consumption concentrates more on large model training and R&D, while China's skews toward application-layer inference services.

This difference reflects divergent AI development strategies. The US maintains leadership in foundational model innovation, while China holds advantages in application deployment speed and scale. As the AI Agent era arrives, application-layer scaling capability may become the more critical competitive dimension.

Risks and Outlook

Rapid Token growth brings data security and privacy challenges. China is finalizing AI data security regulations, with specialized AI Agent data management measures expected in the second half of 2026.

Furthermore, this development signifies a broader transformation in the global technology landscape. As governments and enterprises worldwide continue to increase their investments in AI technologies, the accompanying legal and regulatory frameworks are also evolving rapidly. Regulators face the ongoing challenge of balancing innovation promotion with user protection.

From an industry perspective, these changes are driving increased cross-sector collaboration and integration. The convergence of traditional industries with AI technologies is creating new business models and value chains. Simultaneously, talent development and technology standardization have become critical factors for industry advancement.

Looking ahead, we can anticipate continued rapid development in this field, bringing forth more innovative applications and solutions. Companies must maintain acute market awareness and adjust their strategic directions promptly to adapt to emerging technological trends. User experience enhancement and data security assurance will become core competitive elements.

The implications extend beyond immediate technical considerations to encompass broader socioeconomic impacts. Organizations must prepare for significant shifts in operational paradigms while ensuring sustainable and responsible implementation of these technologies.