Hyperscaler data center capex jumped 57% in 2025 as AI deployments accelerated
Key Highlights
- Data center capital expenditures increased by 57% in 2025, reaching over $420 billion among the top hyperscalers.
- The four largest US cloud providers collectively spent over $340 billion, with Amazon leading at $125 billion.
- AI investments are shifting focus from training to inferencing, especially as models become more complex and token-hungry.
- The data center capex is projected to surpass $1 trillion in 2026, driven by high-end accelerators and infrastructure needs.
- Dell and Supermicro dominate the AI server market, with white-box vendors capturing a significant share of shipments amid rising server prices.
Hyperscaler data center 2025 capex
The largest hyperscalers spent over $420 billion in data center capital spending in 2025.
Here’s a breakdown of the key players' spending patterns:
· Amazon: $125 billion, up from $83 billion in 2024
· Microsoft: $80 billion, up from $55 billion in 2024
· Google: $75 billion, up from $52.5 billion in 2024
· Meta: $65 billion, an increase from $39 billion in 2024
· StarGate: $100 billion in 2024
Data center spending by hyperscalers remains robust.
Driven by AI deployments and general infrastructure investments, Dell’Oro said that worldwide data center capital expenditures increased 57 percent in 2025.
“The Top 4 US cloud service providers—Amazon, Google, Meta, and Microsoft—increased data center capex by 76 percent in 2025,” said Baron Fung, Senior Research Director at Dell’Oro Group.
Fung noted that “Oracle more than tripled its data center capex as it builds capacity for the Stargate project.”
The research firm projects hyperscalers and AI model developers to sustain strong capex momentum in 2026.
“We forecast full-year 2026 data center capex to surpass $1 trillion,” Fung said. “The trillion-dollar threshold marks a major milestone, supported by well over 10 million high-end accelerators as the primary capex driver, along with related infrastructure.”
Inference becomes a driver
Most AI-related investments were focused on supporting hyperscalers’ internal frontier models. These models allow them to vertically integrate AI capabilities, reduce dependence on third-party labs, optimize their massive compute infrastructure, and secure a competitive edge in AI service provision.
Dell’Oro noted that while most AI investments are dedicated to training workloads, inferencing will likely become a larger capex driver in the future, especially as token-hungry reasoning models become more widespread.
“This heightened level of investment raises the potential for overcapacity in AI infrastructure, although hyperscalers are taking proactive measures to mitigate risks and optimize costs,” Fung said.
Dell and Supermicro remain dominant
From a vendor perspective, Dell led all OEMs in AI-optimized server revenue in 2025, followed by Supermicro, driven by strong shipments of NVIDIA Blackwell.
Meanwhile, White-box vendors captured the majority of server shipments, supported by hyperscale AI deployments for Blackwell and custom systems, as well as a surge in demand for general-purpose servers for compute and storage workloads.
Dell’Oro has forecast that average selling prices for general-purpose servers will rise by high double digits in 2026, with escalating DRAM and storage prices as the primary growth driver.
For related articles, visit the Data Center Topic Center.
For more information on high-speed transmission systems and suppliers, visit the Lightwave Buyer’s Guide.
To stay abreast of fiber network deployments, subscribe to Lightwave’s Service Providers and Datacom/Data Center newsletters.
About the Author
Sean BuckleySean Buckley
Sean is the Editor-in-Chief of Lightwave. He establishes and executes Lightwave's editorial strategy across its website, email newsletters, events, and other information products. Before coming to Lightwave in July 2023, he served as editor of Broadband Communities. Earlier, he served as senior editor of FierceTelecom and the editor of the former Telecommunications Magazine. He got his start in the optical industry in 1998 when he became the editor of enterprise and optical networks at Information Gatekeepers. He has a BA in English from the University of Massachusetts, Boston, and lives in Dracut, Mass, with his wife, two sons, and his cat, Dove.
