Webscale Market Tracker, 2Q25: AI’s drunken sailors push 12 month capex to $400B, tech share hits all-time high of 60%
Description
Capex: AI Hype Sends Spending Soaring
Capex skyrocketed 77.0% YoY in 2Q25 to $121.5B, annualizing to $400.3B, up 72% from a year ago and setting another all-time high.
The AI frenzy, sparked by ChatGPT and fanned by investors, is now a dominant force. A review of the latest (2Q25) earnings calls from major US-based webscalers and other AI ecosystem players reveals a collective case of heads in the sand as AI infatuation continues. In none of these calls do tech leaders address when their AI investments will pay off. There are no signs of profitability from their early forays. Instead, there’s just an emphasis on the need to go as fast as possible to establish an early lead in this race – a race they’ve defined, hoping it will lead to new riches. Riches for them and their shareholders, not for employees or users, of course. Most companies pass the buck when justifying their capital expenditure surge, claiming they are simply responding to unprecedented demand from clients. Yet those same clients are making the same risky bets, hoping that someone will eventually land on a sustainable business model for generative AI. Does the word “bubble” appear anywhere in any of the 2Q25 earnings calls from the key data center builders? Not from the webscalers. Zero. Nobody dares talk about the elephant in the room.
Top 2Q25 capex outlays came from Amazon ($32.2B), Alphabet ($22.4B), Microsoft ($17.1B), and Meta ($16.5B). Together, that is 73% of the global total.
Notably, 61% of annualized capex was for technical infrastructure (data center compute & networking, power & cooling, fiber & transport/routing gear, etc.) (vs. 55% in 2Q24), showing a focus on retrofitting existing data centers for AI. The 61% is an all-time high on an annualized basis.
Notes: (1) This is MTN Consulting’s 31st quarterly assessment of the webscale market, part of a series we launched in 4Q17; our data and analysis spans the 1Q11-2Q25 timeframe, i.e. 50 quarters. (2) The 25 companies in our study include several recent additions: CoreWeave (added last year), and this quarter’s three new adds: Kuaishou, Nebius (Yandex spinoff), and Xiaomi.
Please Note: This product is delivered as a Excel File.
Capex skyrocketed 77.0% YoY in 2Q25 to $121.5B, annualizing to $400.3B, up 72% from a year ago and setting another all-time high.
The AI frenzy, sparked by ChatGPT and fanned by investors, is now a dominant force. A review of the latest (2Q25) earnings calls from major US-based webscalers and other AI ecosystem players reveals a collective case of heads in the sand as AI infatuation continues. In none of these calls do tech leaders address when their AI investments will pay off. There are no signs of profitability from their early forays. Instead, there’s just an emphasis on the need to go as fast as possible to establish an early lead in this race – a race they’ve defined, hoping it will lead to new riches. Riches for them and their shareholders, not for employees or users, of course. Most companies pass the buck when justifying their capital expenditure surge, claiming they are simply responding to unprecedented demand from clients. Yet those same clients are making the same risky bets, hoping that someone will eventually land on a sustainable business model for generative AI. Does the word “bubble” appear anywhere in any of the 2Q25 earnings calls from the key data center builders? Not from the webscalers. Zero. Nobody dares talk about the elephant in the room.
Top 2Q25 capex outlays came from Amazon ($32.2B), Alphabet ($22.4B), Microsoft ($17.1B), and Meta ($16.5B). Together, that is 73% of the global total.
Notably, 61% of annualized capex was for technical infrastructure (data center compute & networking, power & cooling, fiber & transport/routing gear, etc.) (vs. 55% in 2Q24), showing a focus on retrofitting existing data centers for AI. The 61% is an all-time high on an annualized basis.
Notes: (1) This is MTN Consulting’s 31st quarterly assessment of the webscale market, part of a series we launched in 4Q17; our data and analysis spans the 1Q11-2Q25 timeframe, i.e. 50 quarters. (2) The 25 companies in our study include several recent additions: CoreWeave (added last year), and this quarter’s three new adds: Kuaishou, Nebius (Yandex spinoff), and Xiaomi.
Please Note: This product is delivered as a Excel File.
Table of Contents
- Report highlights
- Outlook
- Analysis
- Key Stats
- Company Drilldown
- Company Benchmarking
- Regional Breakouts
- Raw Data
- Exchange Rates
- About
Search Inside Report
Pricing
Currency Rates
Questions or Comments?
Our team has the ability to search within reports to verify it suits your needs. We can also help maximize your budget by finding sections of reports you can purchase.