I read a lot of AI coverage, and it can be difficult to cut through the noise. Noise isn’t useful if you are a commercial leader trying to make well informed decisions about AI now. Ramp is a US financial operations platform, and an ‘AI first’ business. They recently published data that cuts through the noise. They sit on real business spend data from tens of thousands of companies. Not survey responses. Not forecasts. Actual purchasing behaviour.
What the data tells us
Over half of US businesses on Ramp’s platform now pay for at least one AI product or service. However, paid AI adoption actually dipped in September 2025, the second decline that year. It fell across small and medium-sized businesses, across almost every industry. Three data points matter more than the adoption dip.
First, retention. In 2022, barely half of companies that subscribed to an AI product were still paying for it by the end of the year. By the end of 2024, that figure was above 80%. AI tools are becoming embedded in how businesses operate. Companies are not trialling and walking away they are committing to use.
Second, contract sizes. The average AI contract value was $39K in 2023. It hit $143K in 2024, $530K in 2025, and Ramp projects it will reach $1M in 2026. That is not bubble behaviour. That is deepening investment by businesses that have found value and are scaling it. Our expense management platform has a set of AI modules, and that has increased our average deal size by almost double in 6 months. Our customers buy them, use them, and love them.
Third, where the money is going. The largest contracts are not with the model companies OpenAI, Anthropic, Google. They are going to AI-powered products built for specific enterprise problems: customer service agents, observability platforms, workflow tools. In other words, applied AI. Solutions that do a defined job inside a business.
The divergence
This is where it gets interesting for anyone running a business or a commercial function. The Ramp data shows spend concentrating. Larger businesses increased AI adoption through the September dip while smaller ones pulled back. That pattern maps onto what PwC found in their 2026 AI Performance study: roughly 20% of companies are capturing three-quarters of AI’s economic gains. Those companies are two to three times more likely to be using AI to pursue growth and reinvent their business model, not just to cut costs. McKinsey’s data says the same thing differently. Their AI high performers, the companies reporting the broadest range of benefits, are more than three times as likely to be using AI for transformative change. And the common thread is not bigger budgets. It is that they set growth and innovation as objectives, not just efficiency.
What this means practically
The useful question is not “should we adopt AI”, because that is already settled. 88% of organisations already use it in at least one function. The question is whether your AI investment is a cost line or a growth driver. Most businesses are using AI to do existing things slightly faster. A smaller group is using it to change what they do, how they sell, how they serve clients, how they make decisions, what they offer. The Ramp retention and contract data suggests that second group is not experimenting. They are committing at scale, and the gap between them and everyone else is widening every quarter.
For businesses in our space, payroll, HR, expense management, this is directly relevant. The companies that treat AI as a feature to bolt on will compete with companies that have rebuilt their products and operations around it.
We see this in our own market. Our expense management platform was built AI-first by our CTO, James Rowell. That architectural decision is now a compounding advantage in product capability, in speed of iteration, and in what we can offer clients. It is the way that we are building all products moving forward.
The same principle applies internally. I wrote recently about creating a Go To Market Engineer role inside my commercial team, someone building AI agents that sit on our CRM and pipeline data. That decision came from the same logic as the Ramp data: the return on AI investment comes from embedding it into how you work, not from subscribing to another tool.
Where this goes
The data is clear on direction. AI budgets are increasing 86% of companies plan to spend more in 2026. Contract sizes are growing. Retention is climbing. The businesses making these investments are outperforming the ones that are not. The Ramp data, the PwC study, the McKinsey findings all point to the same conclusion: this is not a technology trend that rewards waiting. I am interested in how other commercial leaders in our space are thinking about this. If you are working through similar decisions, where to invest, what to build, what to buy, I’d welcome the conversation.
