微软在最新一个季度的资本支出低于华尔街预期，引发市场对其人工智能基础设施投资计划的关注。不过，这并不意味着微软正在收缩AI投资规模。实际情况是，部分数据中心项目建设进度落后于计划，导致原本预计在本季度发生的支出被推迟至之后的财季。
根据微软公布的数据，在截至2026年9月的季度中，公司资本支出达到452亿美元，其中包括融资租赁相关支出，低于分析师预期的517亿美元，差额约为65亿美元。尽管未达到市场预期，但与去年同期相比，微软本季度资本支出仍增长约30%，反映出人工智能基础设施建设依然需要投入巨额资金。
微软同时下调了2026年全年资本支出预测，将此前预计的1750亿美元调整至1700亿美元。不过，这一调整并不代表公司改变了大规模投资人工智能的战略方向。
微软首席财务官艾米·胡德表示，部分第三方供应商未能按时交付所需的数据中心容量，导致相关支出推迟。微软预计，这些设施将在公司财年下半年陆续投入使用。随着项目建设推进，资本支出预计将重新加速增长，而内存等硬件价格上涨也将进一步推高基础设施建设成本。
微软在本季度新增了24座数据中心，但公司并非所有基础设施都由自身直接建设和运营。为了快速扩展云计算能力，微软还与多家规模较小的云基础设施运营商合作。自2023年以来，微软已与CoreWeave、Nebius和Nscale签署总额超过800亿美元的相关合同，以获取额外的数据中心容量。
然而，第三方基础设施项目同样面临现实障碍。微软近期放弃了一项与Nscale有关的数据中心容量合作安排，原因是供电能力不足。此外，Nebius在美国新泽西州的一处设施建设进度也慢于预期。这些问题表明，在人工智能算力需求快速增长的背景下，数据中心扩张不仅取决于资金和芯片供应，还受到电力资源、基础设施建设进度等因素的制约。
类似的问题并非微软独有。亚马逊也在大量采购并储备英伟达GPU服务器机架，但部分数据中心由于缺乏足够电力，暂时无法让这些设备投入运行。亚马逊今年的资本支出预算约为2200亿美元，比微软调整后的全年预算高出约500亿美元，显示出大型科技公司之间的AI基础设施投资竞赛仍在持续。
Google母公司Alphabet同样维持着庞大的投资计划，其全年资本支出预算为1950亿至2050亿美元。不过，Alphabet在截至9月的季度中资本支出达到514亿美元，同样低于市场预期。公司预计，明年的资本支出还将显著增加。
巨额基础设施投入也正在给Alphabet的现金流带来压力。该公司本季度自由现金流为负91亿美元，延续了此前自由现金流承压的趋势。相比之下，微软目前仍保持着较强的现金创造能力，为持续投资AI基础设施提供了更多财务空间。
微软的Azure云计算业务表现，为其高额资本支出提供了重要支撑。本季度Azure收入同比增长48%，达到332亿美元。与此同时，微软自由现金流仍为正186亿美元，显示公司在扩大人工智能基础设施投入的同时，核心云业务依然能够产生可观的现金流。
值得注意的是，微软一边加大AI基础设施投资，一边继续压缩部分人力成本。此前，公司已采取包括向员工提供自愿离职补偿方案在内的措施，相关方案最多可能涉及6万名员工。这反映出微软正在重新配置资源，将更多资金和经营资源投入人工智能、云计算及相关基础设施，同时控制其他领域的成本。
总体来看，微软本季度资本支出低于预期，主要反映的是数据中心建设和第三方容量交付的时间差，而非AI投资热潮降温。随着延迟项目陆续投入使用，微软未来几个季度的支出仍可能明显增长。与此同时，电力供应、数据中心建设周期以及内存等关键硬件成本，正在成为制约全球科技巨头扩大AI算力规模的重要因素。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Microsoft ' s capital expenditure in the latest quarter was lower than expected on Wall Street, raising market interest in its AII investment plan. However, this does not mean that Microsoft is scaling down AI investments. The fact is that some data centre projects are lagging behind schedule, resulting in the postponement of expenditures that were originally expected to be incurred during the current quarter to subsequent financial periods.
According to data released by Microsoft, corporate capital expenditure amounted to $45.2 billion by September 2026, including finance lease-related expenditure, which was lower than the analyst ' s projection of $51.7 billion, leaving a difference of approximately $6.5 billion. Although market expectations were not met, Microsoft ' s capital expenditure increased by about 30 per cent this quarter compared to the same period last year, reflecting the continued need for significant investment in artificial intelligence infrastructure.
At the same time, Microsoft revised its capital expenditure projections downwards throughout 2026, adjusting the previously projected $175.0 billion to $170 billion. However, this adjustment does not represent a change in the strategic direction of large-scale investments in artificial intelligence.
According to Microsoft Chief Finance Officer Amy Hood, some third-party vendors were not able to deliver the data centre capacity required on time, resulting in delays in related expenditures. Microsoft expects these facilities to be operational in the second half of the company ' s financial year. As project construction advances, capital expenditures are expected to accelerate again, while higher prices for hardware, such as memory, will further increase infrastructure costs.
Microsoft added 24 new data centres this quarter, but not all of its infrastructure is directly built and operated by itself. To rapidly expand cloud computing capacity, Microsoft also works with a number of smaller cloud infrastructure operators. Since 2023, Microsoft has signed contracts totalling over $80 billion with CoreWeave, Nebius and Nscale to acquire additional data centre capacity.
However, third-party infrastructure projects also face real obstacles. Microsoft recently abandoned a data centre capacity cooperation arrangement related to Nscale because of insufficient power supply. In addition, the construction of a facility in New Jersey, United States, was slower than expected. These problems suggest that in a context of rapidly growing demand for artificial intelligence, the expansion of data centres depends not only on the availability of funds and chips, but also on electricity resources and the pace of infrastructure development.
Similar issues are not unique to Microsoft. The Amazon is also procuring and storing GPU server racks in large quantities, but some data centres are temporarily unable to operationalize the equipment owing to lack of sufficient electricity. The Amazon ' s capital expenditure budget this year is about $220 billion, about $50 billion more than Microsoft ' s adjusted full budget, indicating that the AI infrastructure investment competition between large technology firms continues.
The Google parent company, Alphabet, also maintains a large investment programme with an annual capital expenditure budget of between 1950 and $20.5 billion. However, Alphabet ' s capital expenditure reached $51.4 billion in the quarter to September, again below market expectations. The company expects a significant increase in capital expenditure next year.
Large infrastructure inputs are also putting pressure on Alphabet's cash flow. The company ' s free cash flow this quarter was negative $9.1 billion, continuing the previous trend of free cash flow pressure. In contrast, Microsoft still maintains a strong cash-generating capacity, providing more financial space for sustained investment in AI infrastructure.
Microsoft ' s Azure cloud computing performance provides an important support for its high capital expenditure. Azure's income increased by 48 per cent over the same period to $33.2 billion. At the same time, Microsoft's free cash flow remains positive at $18.6 billion, showing that while companies expand their inputs to artificially intelligent infrastructure, core cloud operations can still generate significant cash flows.
It is noteworthy that Microsoft continues to reduce part of its human costs while increasing AI infrastructure investment. Prior to this, the company had taken measures, including the provision of a voluntary separation compensation programme for staff, which could involve up to 60,000 employees. This reflects the fact that Microsoft is reallocating resources to invest more financial and operational resources in artificial intelligence, cloud computing and related infrastructure, while controlling costs in other areas.
Overall, Microsoft ' s quarterly capital expenditure was lower than expected, mainly reflecting the time lag between data centre construction and third-party capacity delivery rather than the heat and temperature of investment in AI. As projects become operational with delays, Microsoft's expenditures may continue to increase significantly in the coming quarters. At the same time, key hardware costs, such as electricity supply, the data centre construction cycle and memory, are becoming important constraints on the global technological giants ' ability to expand AI ' s computing capacity.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.