U.S. technology giants are ploughing unprecedented sums into artificial-intelligence infrastructure, with the four largest cloud providers spending a combined $88 billion on capital projects in the most recent quarter. Amazon led the outlays at about $32 billion, followed by Alphabet at $22 billion and Microsoft and Meta at roughly $17 billion apiece. The cash is being channelled into graphics-processing units, custom accelerators and new data-centre capacity aimed at training and running generative-AI models. The investment curve is steepening. Research firm TipRanks projects the group will devote $364 billion to AI-related spending in 2025. Amazon tops the forecast at $118.5 billion, while Microsoft is seen allocating $88.7 billion and Meta and Alphabet are also raising their budgets. Renaissance Macro estimates that, in the first half of 2025, AI capital expenditures contributed more to U.S. gross domestic product growth than household consumption, underscoring the sector’s macro-economic significance. Investors are beginning to question whether revenue can keep pace with costs. Amazon’s shares fell 8.5% on Friday—wiping out roughly $200 billion in market value—after the company reported 18% year-on-year growth at its AWS cloud unit, a figure viewed as underwhelming given the scale of planned investment. Analysts caution that sustained spending without commensurate earnings could pressure balance sheets across the industry. Chip suppliers are maneuvering to capture the next leg of demand. Futurum Research expects customised accelerators to command about 25% of the AI-processor market within five years, with Broadcom and Marvell taking the bulk of that share, while Nvidia remains the preferred provider of off-the-shelf GPUs for external workloads. At the same time, economists such as the Wall Street Journal’s Greg Ip warn that the “sky-high” cost of AI infrastructure could test capital markets if revenue projections prove overly optimistic.