贝恩咨询发布报告指出，到2031年，AI产业必须实现6万亿美元的年收入，才能为当前规模庞大的数据中心投资提供合理的经济回报1。报告预计，新产品开发将贡献其中的4.2万亿美元，企业生产力领域贡献1-1.4万亿美元，消费服务领域贡献2000-4000亿美元1。
与此同时，支撑AI产业所需的基础设施投资持续攀升1。报告指出AI基础设施的年支出预计到2031年达到1.5万亿美元1。以Meta的Prometheus数据中心为例，2025年的600MW规模需投入240亿美元；到2027年，规模将扩至2GW、成本达800亿美元；2029年进一步扩大到5GW、投入1750亿美元；到2030年达到9GW、成本攀升至2000亿美元1。数据中心的规模和建设成本每12-16个月翻倍增长1。
贝恩咨询全球技术实践主席David Crawford表示，"AI基础设施经济学需要生产力收益之外的万亿级新收入"1。
According to Bain & Company, the artificial intelligence industry must achieve $6 trillion in annual revenue by 2031 to justify the massive investments in data centre infrastructure currently underway 1. The consulting firm's analysis breaks down the revenue sources needed to support this economic model: new product development is projected to contribute approximately $4.2 trillion, enterprise productivity between $1 trillion and $1.4 trillion, and consumer services between $200 billion and $400 billion 1.
The scale of infrastructure spending required to enable these revenue streams is substantial. Annual expenditure on AI infrastructure alone is expected to reach $1.5 trillion by 2031 1. Meta's Prometheus data centre project illustrates the accelerating costs involved, with capacity and expenses doubling every 12 to 16 months 1. The company's planned buildout demonstrates this trajectory: 600 megawatts in 2025 at a cost of $24 billion, growing to 2 gigawatts by 2027 at $80 billion, 5 gigawatts by 2029 at $175 billion, and 9 gigawatts by 2030 at $200 billion 1.
Bain's Global Technology Practice Chair David Crawford emphasized the challenge, stating that "AI infrastructure economics require trillion-dollar levels of new revenue beyond productivity gains" 1. The findings suggest that without generating sufficient revenue across these multiple sectors, the current pace of data centre expansion may become economically unsustainable.
评论
还没有评论，欢迎留下第一条。