Three years of AI infrastructure investment is now flowing into everyday business software, data shows
The numbers behind the global AI infrastructure boom are large enough that they can feel abstract. But they have direct consequences for the tools that small businesses use every day, and understanding the scale helps explain why the pace of change in business software has accelerated so sharply.
Research compiled by BestBrokers, drawing on StockAnalysis data and company financial disclosures, tracked AI-related revenue across 20 publicly listed companies operating within the global AI supply chain between 2022 and 2025. The figures show a build-out that is less a software revolution and more an industrial expansion, concentrated in the physical infrastructure required to run AI at scale.
Nvidia, the dominant supplier of graphics processing units used in AI training and inference, grew its data centre and AI-related revenue from $14.6 billion in 2022 to $167.9 billion in 2025, a rise of more than $153 billion in three years. CoreWeave, a GPU-as-a-service provider, recorded the fastest growth rate in the dataset, expanding from $15.8 million to $5.13 billion over the same period, a compound annual growth rate of 586.88% according to BestBrokers’ analysis.
Paul Hoffman of BestBrokers described the pattern plainly. “What we’re seeing is less of a software revolution and more of an industrial build-out,” he said. “The AI economy runs on physical bottlenecks, and the market is rewarding whoever can remove them.”
Where the money is flowing
The investment is concentrated across several layers of the AI supply chain. At the computing layer, chip designers and manufacturers including Nvidia, AMD and Marvell Technology have seen substantial revenue growth as demand for the hardware running AI workloads has accelerated. At the memory layer, companies including SK Hynix have grown significantly as AI model training requires increasingly large volumes of high-bandwidth memory.
At the infrastructure layer, networking companies including Arista Networks and cooling and power infrastructure providers including Vertiv have grown steadily as data centres scale up to handle AI workloads. At the cloud layer, the hyperscale platforms that deliver AI capabilities to businesses, Amazon Web Services, Microsoft Azure and Google Cloud, have posted strong absolute growth, reflecting AI being layered into already large and diversified cloud businesses.
Amazon’s relevant revenue grew from $80.1 billion to $128.7 billion between 2022 and 2025 according to BestBrokers’ analysis. Microsoft’s grew from $81.8 billion to $120.4 billion. Alphabet’s grew from $26.3 billion to $58.7 billion. These are the same platforms delivering AI-powered tools to small businesses through products like Microsoft 365, Google Workspace and AWS-backed services.
The infrastructure investment at the top of this supply chain is what makes the AI features appearing in everyday business software possible. The Copilot assistant arriving in Microsoft Word and Excel, the Gemini features in Google Docs and Sheets, the AI tools embedded in accounting platforms, CRM systems and marketing software, all depend on the compute, memory and cloud infrastructure being funded at the scale reflected in these figures.
For small business owners, this means the tools available in 2026 are meaningfully more capable than those available in 2023, and the trajectory of improvement is continuing. It also means the major platforms have significant commercial incentive to keep embedding AI features into products small businesses already pay for, because the infrastructure investment requires ongoing revenue to justify it.
What SME owners should take from this
The practical implication is not that small business owners need to track AI infrastructure stocks. It is that the tools reaching them through existing software subscriptions will continue to improve, that the major platforms are competing aggressively to deliver AI capability at the business software layer, and that the window for getting ahead of this shift is now rather than later.
The AI build-out is not a future event. It has already happened. The products it produces are already in the software small businesses use every day. The question is whether those businesses are using them well enough to benefit from the investment that made them possible.
Data sourced from BestBrokers analysis of StockAnalysis financial disclosures. All figures reflect AI-related or closest disclosed segment revenue for publicly listed companies.
Keep up to date with our stories on LinkedIn, Twitter, Facebook and Instagram.
