Microsoft lowered its annual capex projection to $170 billion, while Alphabet held its forecast flat following third-party delivery delays that squeezed its quarterly cash flow. Outside GPU and cloud suppliers faced setbacks that will postpone some planned capacity rollouts into the first half of 2027. This deceleration rattled specialized neocloud vendors, who are struggling to secure expansion debt amid cautious lenders and higher interest rates. Even with these hiccups, overall hyperscaler spending is projected to exceed $1 trillion next year as investments accelerate.
Key points
Infrastructure bottlenecks are temporarily restraining the rapid pace of artificial intel…: Infrastructure bottlenecks are temporarily restraining the rapid pace of artificial intelligence hardware deployments across the technology sector.
Specialized cloud suppliers face severe capital constraints as lenders pull back due to n…: Specialized cloud suppliers face severe capital constraints as lenders pull back due to non-investment-grade credit ratings and high interest rates.
Long-term tech infrastructure investments remain historically massive, with combined hype…: Long-term tech infrastructure investments remain historically massive, with combined hyperscaler spending projected to exceed $1 trillion next year.
Sources checked by our analysis pipeline: 1
The past few quarters of hyperscaler earnings have been marked by ever-increasing capital expenditure forecasts as tech giants race to build out artificial intelligence infrastructure to meet what they describe as insatiable demand.
At least for the moment, Microsoft and Alphabet are telling a slightly different story.
In their earnings reports Wednesday, both companies shared capex figures that came up shy of analysts' estimates. Microsoft reduced its capex forecast for the year to $170 billion from $175 billion, and Alphabet held its guidance at $195 billion to $205 billion.
The numbers suggest a slowdown is nowhere in sight, but they do represent another potential point of concern for the AI build-out. Shares of neoclouds CoreWeave, Nebius and IREN all dropped after the reports, as did Oracle, which is one of the biggest spenders on AI infrastructure.
Microsoft CFO Amy Hood said on the software company's earnings call that third-party providers of graphics processing units had delays that will push out some available capacity until the first half of 2027. She didn't provide names, but Microsoft relies on CoreWeave, IREN and…
During a conversation with John Collison, OpenAI leadership examined practical capabilities alongside pressing infrastructure and safety challenges facing the industry. The discussion touched on technical risks such as containing advanced systems while highlighting AI applications in cybersecurity and data operations. These insights illustrate the technical hurdles organizations encounter as models become increasingly capable and autonomous.
In a comprehensive discussion on the evolving artificial intelligence landscape, an industry veteran reflected on two decades of operational experience building scalable technology systems. The conversation centered on technical principles like diffusing intelligence, ensuring systemic safety, and deploying engineers directly into customer environments. These operational insights provide a critical window into how enterprise software leaders evaluate modern AI deployment and geopolitical tension.
The capital injection will support the company's planned relocation to the United States as it scales its engineering tools. Led by CEO Tomide Adesanmi, the team is targeting hardware workflows by deploying software to automate complex electronics and circuit engineering. This geographic move positions the firm closer to major American technology markets and potential hardware partners.