Quick Overview

  • Google plans to increase capital expenditures to $205 billion in fiscal year 2026 to provide computing capacity to meet demand growth. Executives said on Wednesday's second-quarter earnings call that this figure is higher than the previous maximum of $19 billioninitial estimate, and capital expenditures are expected to continue to increase significantly in 2027, Chief Financial Officer Anat Ashkenazi said on the call.
  • "We remain in a supply-constrained environment," Ashkenazi said. "Although we have significantly increased capacity over the past three years, demand still exceeds investment."
  • The company's cloud revenue grew 82% year-over-year in the second quarter, driven mainly byGoogle Cloud Platformenterprise AI product and infrastructure revenue growth. Ashkenazi said Google has also begun generating revenue from its tensor processing unit systems, which were deployed to customer data centers for the first time this quarter.

Deep Insights

Google's increased capital expenditures continue the trend of tech giants massively building AI infrastructure to meet the growing computing needs of enterprises.

According to a report released Thursday by Synergy Research Group, the overall capacity of U.S. data centers is expected todoublewithin the next three years. Meanwhile, the operational capacity of data centers owned by hyperscalers will double within the next two years, as companies including Google, Microsoft, and AWS are actively investing in construction, Synergy found.

"Power supply constraints and local concerns about data centers are certainly suppressing many new data center plans," John Dinsdale, chief analyst at Synergy Research Group, said in the report. "But it is equally clear that data center developers will continue to find ways to address these issues, and strong demand will continue to drive aggressive capacity growth."

The development of AI, such as agents,consumes large amounts of computing power, further exacerbating supply constraints as enterprises seek to deploy the technology at scale. Agentic AI in particular places additional pressure onlegacy IT systemsto support this high-computing-demand technology.

The increased use of AI also translates intomore token consumption, which is becoming a cost concern for enterprises. Tokens are often used as a measure of AI usage and have become a way for providers to price their services.

Google CEO Sundar Pichai said on Wednesday's earnings call that enterprises are using AI (and thus consuming tokens) across a variety of workflows, such as streamlining data analysis, managing customer relationships, building agents, automating processes, and improving cybersecurity.

"All this momentum is driving growth in our paid token usage," Pichai said.

Pichai said that over the past year, more than 2,000 enterprises consumed over 100 billion tokens, while nearly 500 Google Cloud customers processed over 1 trillion tokens.

While reporting capital expenditures and revenue growth this week, Google also received amassive finefrom the European Commission on Thursday for violating theDigital Markets Act. The Commission fined Google €890 million (approximately $1.01 billion) for self-preferencing its own services in Google Search and restricting businesses from steering consumers to cheaper alternatives.

"The best products should succeed because they are better, not because they are owned by the company that operates the search engine," Teresa Ribera, the Commission's executive vice president for clean, fair, and competitive transition, said in a press release.