2026年全球绿色AI数据中心市场将达676亿美元

Technology - The competition in AI large models will continue into 2026, with companies like OpenAI, Google, and DeepMind releasing larger or more efficient models [2] - AI models are making progress in spatial understanding, aiming to develop capabilities in semantics, physics, geometry, and dynamic interactions [2] - By 2026, 40% of enterprise applications are expected to embed task-oriented AI agents, a significant increase from less than 5% in 2025 [2] Industry - The combination of digital twins and AI agents is reshaping product design processes, marking a strategic opportunity for "smart manufacturing" [4] - By 2026, 40% of manufacturers with production scheduling systems are predicted to upgrade to AI-driven scheduling for autonomous resource management [4] - The shift from traditional manufacturing to "smart manufacturing" is expected to enhance market perception, product innovation, and international competitiveness for Chinese enterprises [4] Energy - The energy demand from global data centers is projected to double by 2030, reaching approximately 945 terawatt-hours, with AI being a major driver of this increase [5] - The global green AI data center market is expected to reach $67.6 billion by 2026 and grow to about $123 billion by 2035 [5] - China is focusing on optimizing supply capabilities and green low-carbon initiatives to support sustainable AI development [5] Governance - 2026 is anticipated to be a pivotal year for the acceleration of global AI governance measures, shifting focus from ideological debates to compliance and cross-border collaboration [7] - The EU's AI Act, set to take effect in August 2026, will be the first comprehensive regulation of AI globally [7] - China's AI governance path is becoming clearer, with a focus on legal frameworks and ethical guidelines to promote healthy AI development [7] Overall Outlook - The future of global AI development will hinge not only on the strength of models but also on the integration of safety, compliance, energy consumption, and industry implementation into a cohesive system [8]