Google says AI demand exceeds supply capacity, signaling $190B capex race

Google said AI demand from enterprises and consumers is exceeding the company's current supply capacity, the clearest signal yet that the industry's infrastructure buildout is struggling to keep pace with adoption.

"Demand for AI solutions from both enterprises and consumers is strong, and it's currently outstripping our ability to supply," a Google spokesperson said.

The statement comes as Google's cloud rivals escalate spending at unprecedented levels. Microsoft said it expects to invest roughly $190 billion in capital expenditures in calendar 2026, including about $25 billion from higher component pricing, while its Azure cloud revenue grew 40% in the fiscal third quarter. Amazon's AWS and Oracle are also expanding data center footprints aggressively, with Oracle shares rallying 32% over the past three months on its own cloud acceleration story.

The supply-demand imbalance carries direct implications for the $7 trillion AI infrastructure buildout projected by Brookfield Asset Management. For investors, the question is whether hyperscalers can convert record capital spending into proportional revenue growth before margins compress further.

The Infrastructure Bottleneck

Google's capacity constraint reflects a broader industry challenge: AI data centers are approaching the limits of existing power and cooling infrastructure. Nvidia's upcoming Rubin GPU system will require around 300 kilowatts per rack, up from 150 kilowatts for the prior generation, and future chips are expected to push racks toward 1 megawatt — enough to power 750 US homes on average. Around 30 percent of the power flowing into data centers is consumed by cooling systems and electrical conversion losses, according to Nvidia.

The industry is responding with redesigns. Nvidia and power equipment makers including Flex and Vertiv are developing 800-volt direct current systems that could improve energy efficiency by 27 percent compared with current alternating current setups. Liquid cooling, already deployed for Nvidia's Blackwell chips, can increase data center energy efficiency by 15 percent, according to a study by Nvidia and Vertiv. These upgrades are expected to reach many AI factories before 2030.

The Revenue Conversion Challenge

Microsoft's AI business has reached a $37 billion annual revenue run rate, growing 123 percent year over year, and its commercial remaining performance obligation surged 99 percent to $627 billion. Microsoft 365 Copilot paid seats now exceed 20 million, with seat additions up 250 percent year over year. Nearly 90 percent of the Fortune 500 have active agents built with Microsoft's low-code and no-code tools.

Yet the spending side is equally staggering. Microsoft's capital expenditures in the fiscal fourth quarter are expected to surpass $40 billion, and the company's cloud gross margin declined to 66 percent in the fiscal third quarter, driven by AI infrastructure investments. Microsoft trades at a forward price-to-sales ratio of 8.86 times, a premium to the industry's 7.55 times, suggesting the market is pricing in continued growth but leaving little room for execution missteps.

For Google, the capacity constraint means near-term AI revenue may be capped even as demand surges. Alphabet shares have benefited from the broader AI rally, but the company faces the same margin dynamics as Microsoft: higher spending on data centers, GPUs and power infrastructure before the revenue catches up. Ciena, a key supplier of optical networking equipment for data center interconnect, reported a record backlog of approximately $7 billion and 40 percent revenue growth in India, reflecting the scale of network buildout required to support AI traffic.

The competitive stakes are high. Azure's 40 percent growth keeps Microsoft in the top tier of hyperscalers, while Google Cloud and AWS match the investment pace. Nvidia, whose servers and chips make up 70 percent of AI hyperscaler spending according to Bloomberg Intelligence, stands to benefit regardless of which cloud provider wins market share. But the supply constraints Google highlighted suggest that even the largest players cannot scale fast enough to meet current demand — a dynamic that favors early movers with existing capacity and punishes those still building out.

This article is for informational purposes only and does not constitute investment advice.