For years, artificial intelligence has lived within a narrative of almost inevitable growth. However, new data centers, multi-billion-dollar funding rounds, and ever-higher valuations presuppose something very concrete: that companies and users continue to spend enough to sustain all this infrastructure. Data from Ramp for August introduces a small element of doubt.
According to the index built by the corporate spend management platform across roughly 70,000 businesses, 56% of customers paid for at least one AI product in August. That share grew by just 0.4 percentage points compared to the previous month, a slowdown compared to the pace observed at other times of the year.
One month does not make a trend
August is a peculiar month, especially for European and American businesses. Ramp had also previously observed a nearly flat phase between August and October of the prior year, followed by renewed acceleration. It would therefore be premature to speak of saturation.
The data remains noteworthy because AI has entered a phase where usage growth must translate into recurring revenue. Major labs and hyperscalers are pouring enormous sums into GPUs, energy, and compute capacity on the conviction that end demand will continue to expand.
From experiment to budget
Many companies have spent the past two years experimenting with dozens of tools. The next phase is more selective: deciding which products genuinely boost productivity and which are simply expensive duplicates. This can reduce the number of subscriptions without necessarily signaling a decline in AI usage.
The most important question for the industry is therefore not whether every company “uses AI,” but how much economic value it can extract and how much it is willing to pay on an ongoing basis. Ramp's data offers one of the few ways to observe this transition through actual spending rather than surveys.
