Ramp's data shows AI token spending growth collapsed to 1.5% in August and September, down from 42% in July, even as companies increased their use of AI models. The slowdown threatens revenue projections for OpenAI and Anthropic as the latter approaches its IPO.
Ara Kharazian, Ramp's lead economist, attributes the stall to price competition and improved customer efficiency. The top 1% of AI-spending firms, which drove earlier hypergrowth, saw their per-employee spending dip just below the $10,000-per-month peak they hit in summer. Median firms spend $12-$15 per employee per month, while the top 10% spend $600-$1,000.
OpenAI and Anthropic both slashed API prices in September. Anthropic's pricing cuts for Fable 5.1 and Opus 5.5 reduced revenue growth by roughly $600 million, according to the company's recent filing. Both firms offered enterprise discounts to lock in customers, but those discounts are compressing revenue targets.
"The competitive nature of the marketplace" is eating into spending, Kharazian said. Meta and Nvidia offer cheap or open-source enterprise models. Companies also use routers to shift workloads to cheaper models and impose employee budgets to control costs.
Anthropic's IPO valuation fell from $2 trillion in early October to $1.2-$1.3 trillion, according to Bloomberg. OpenAI lowered its valuation to $1.25 trillion from $1.4 trillion and cut fundraising goals by $10 billion.
October data showed a reacceleration, with token spend up 8.9% from September 23 to October 23. But Kharazian said headwinds on spend growth remain strong. Investors hope new use cases like personal agents will drive future growth, though Kharazian doubts personal agents will match the impact of AI coding.
"The projection is that there will be new products being introduced that will drive further growth," Kharazian said. "But based on what we can observe today and the products available and what is commercially advanced, AI spend has started to slow down."