Tech leaders are investing in artificial intelligence infrastructure more than ever before, with expenditure reaching record heights. The battle to control AI computing power has introduced a surge of transactions worth hundreds of billions to redefine the way firms are tackling cloud services and processing capability.
How tech giants battle for AI computing supremacy
Worldwide cloud infrastructure expenditures totaled $99 billion in Q2 2025, a staggering 25% year-over-year growth led largely by artificial intelligence requirements. Amazon’s top 30% remains, followed by Microsoft Azure at 20% and Google Cloud at 13% of the fast-growing market. The move up from the average 19% growth rate of 2023 shows AI workloads are irreversibly changing infrastructure requirements.
So impressive about this growth is the degree to which niche players such as CoreWeave, Oracle, and Databricks are recording some of the highest industry growth rates, competing head-to-head against entrenched hyperscalers. In their earliest days, just a few years ago, CoreWeave today has more than $1 billion in quarterly cloud revenue, just outside the top dozen providers in the world.
Large infrastructure deals reshape competitive landscape
- OpenAI-Broadcom tie-up: Homebrew AI chips debuting in 2026
- Meta-CoreWeave deal:ย $14 billion computing power supply contract
- Oracle-OpenAI agreement:ย $300 billion five-year cloud computing deal
- NVIDIA-Intel investment agreement: $5 billion equity providing 4% firm stake
Why Neocloud providers boom with niche services
A new specialty neocloud provider niche is quickly revolutionizing the competitive dynamics by providing exclusively GPU-as-a-Service and AI-tuned infrastructure. They reported over $5 billion in Q2 revenues, a 205% year-over-year growth, indicating how legacy hyperscalers have zero hope of competing with AI demand today. They are distinguished by catering to industries that need absurd compute density and high-capacity data centers tailor-made for artificial intelligence workload.
The neocloud segment will bring 23billion in 2025, and by 2030, the revenue will be 180 billion per year at a compounded rate of 69% per annum. The big neoclouds are CoreWeave, Crusoe, Lambda, Nebius, and OpenAI, with consumer-facing services and infrastructure investments by OpenAI positioning it as the largest of this fast-expanding segment.
“It is a very fast-growing market. As aggressively growing neoclouds come up, they will create fast-growing others” – John Dinsdale, Synergy Research
Providers possess knowledge of high-performance computing requirements
Most of the neocloud firms are repurposing crypto-mining data centers or starting life as high-performance computing specialists, which enables them to grow quickly in power-dense buildings. They’re building far more power-dense buildings and locating them beyond costly metro markets with power constraints, where they achieve competitive advantages over traditional suppliers.
What record investments portend for infrastructure development
The infrastructure race has driven record deal-making in the technology industry as firms ink multi-billion-dollar deals for processing power. Oracle inked one of the largest cloud deals ever with OpenAI, with the ChatGPT developer buying 300 billion of computing capacity for five years.Metatooka4914.3 billion of Scale AI, sending its CEO to a top position in the tech giant’s artificial intelligence plans.
These massive investments are a clear reflection of the reality that AI workloads that use over 100 kW per rack require sophisticated cooling and floor load capabilities that force data centers to greatly improve their infrastructure. The Stargate datacenter project is a joint effort by SoftBank, OpenAI, and Oracle to build data centers where companies are investing up to $500 billion to fund the development of artificial intelligence infrastructure.
The AI infrastructure bubble is not just more spending – it’s a profound transformation in the way technology companies think about computing power and data processing. Artificial intelligence is becoming the business centerpiece, and the firms that are investing most in top-shelf infrastructure today will probably control the digital economy tomorrow. They’re building far more power-dense buildings and locating them beyond costly metro markets with power constraints, where they achieve competitive advantages over traditional suppliers.