What’s Faster Than “Hyper”?

Over the past couple of years, it has seemed difficult to find business headlines that do not reference terms like “agentic AI,” “AI-driven workflow automation,” and “multi-hundred-billion-dollar hardware deal announced.” As the torrent of new information regarding AI model capabilities and investment announcements starts to blend together, it can be helpful to take a step back to view recent figures in the context of prior innovation waves. Such a review highlights the unprecedented speed and scale of AI’s proliferation.

There is a term, “hyperscalers,” that has been used to describe Google, Amazon, and Microsoft. This term was coined to describe how quickly these companies built cloud infrastructure and revenue over the past two decades. There may now need to be a new word invented to describe the pace at which AI firms like Anthropic are accelerating. The latter company first reached a rate of annual revenue of over $1 billion in January of 2025, and the latest estimates this month (a little over one year later) place its revenue run rate above $30 billion. For context, achieving $1 billion in annualized revenue in 3 years is around 2x more quickly than the hyperscalers, and $30 billion in 4-5 years is approximately 3-5x faster. That implies roughly 1,400% growth over the past 15 months. As the hyperscalers reached a few billion in cloud revenue, they were growing at around 100% per year, with a steadily declining trend from there as they continued to expand. The only periods in which they managed to grow at a comparable rate to Anthropic were when they were much smaller.

Anthropic’s revenue acceleration seems to reflect the nature of the product. Large language models are being deployed across enterprises in roles that have immediate impact: software development, customer support, internal research, and workflow automation. Unlike earlier software waves that required long integration cycles, these systems are often layered directly onto existing processes, shortening the path from pilot to scaled usage.

In the cloud computing technology cycle, data centers, networking capacity, and enterprise adoption developed in sequence, with revenue following long after the investment in infrastructure. In contrast, AI demand and infrastructure buildout are currently occurring much closer to simultaneously. Companies like Anthropic are scaling usage while also rapidly expanding the computer capacity required to support it. AI infrastructure spending across the industry is rising at a rate that also exceeds prior technology cycles, both in absolute dollars and in speed of deployment. Partnerships across semiconductor manufacturers, cloud providers, and AI developers are forming quickly to meet demand, with capacity often committed before it is fully built.

Rapid enterprise adoption and revenue growth, paired with heavy infrastructure investment, are driving the expansion of Anthropic and the whole AI industry at unprecedented speed and scale. It will be interesting to see which side ends up being shorter than expected over the next few years, customer demand or the scale of the infrastructure needed to meet it.

Corey Erdoes