Microsoft and Meta made one thing clear in their earnings calls this week. They are spending billions more to expand their AI infrastructure, including building data centers to support it. 

On July 29, both companies reported second-quarter earnings, and AI dominated the conversation. A key takeaway from both calls was a sharp emphasis on capital spending focused on AI and data centers as they hurry to add more computing capacity.

  • Meta, owner of Facebook, raised its 2026 capital expenditure guidance from $115 billion to $135 billion up to a new range of $125 billion to $145 billion. That compares with about $75 billion in capex spending last year. On the earnings call, CFO Susan Li and CEO Mark Zuckerberg said much of that investment is going toward AI infrastructure. The company spent $31.1 billion on capex in the second quarter, driven by investments in servers, data centers, and network infrastructure. Meta executives said it is investing aggressively to meet growing infrastructure needs while continuing to improve efficiency.
  • Microsoft also signaled that investment in AI infrastructure is a top priority. Capex reached $41 billion during the quarter, with management noting that higher component pricing contributed to the increase. Roughly two-thirds of that spending went toward short-lived assets, primarily central processing units (CPUs) and graphics processing units (GPUs), highlighting the company's investment in the computing power needed to support AI demand and Azure cloud services. The company has said it will spend about $190 billion for calendar year 2026 on capex.

Energy is one of the highest costs behind AI because the models and the data centers that run them require enormous amounts of electricity. At RBN, we spend a lot of time evaluating how these power-hungry facilities will secure reliable supply, whether from natural gas, nuclear, renewables or a combination. See: Go Speed Race Go. Neither Microsoft nor Meta broke out energy costs, but powering AI is becoming a major challenge and expense, with implications for both the tech and energy industries.