Ramp’s latest data shows AI usage growing while AI revenue growth stalls. Token spend rose only 1.5% across August and September, after a 42% month-over-month jump in July. Ramp lead economist Ara Kharazian says the slowdown is concentrated among the top 1% of venture-backed, high-growth tech companies — the most “AI-pilled” firms — whose per-employee spend has stuttered just below the $10,000 monthly mark hit at summer peaks. Median firms spend $12–$15 per employee per month and the top 10% spend $600–$1,000; their growth is not enough to make up the difference.

Kharazian attributes the spending stall to competition rather than demand. AI volumes are still up, but OpenAI and Anthropic have made AI cheaper. In September, OpenAI cut API prices in half for GPT-6 Sol and GPT-6 Luna. Anthropic made significant reductions to pricing for Fable 5.1 and Opus 5.5, a decision the company said in a filing slashed revenue growth by about $600 million. Both companies are also offering enterprise discounts meant to lock in spend, but in practice they hit revenue targets. It has become a buyer’s market.

Cost pressure also comes from cheaper alternatives and more thrifty customers. Meta and Nvidia are introducing enterprise offerings with cheap or open source models, and a new lightweight tool called Jev for choosing between predetermined outputs is increasingly among the top performers on Ramp. Companies are offloading workloads to cheaper models via routers, imposing employee budgets, and using defaults that prevent expensive models from handling long-running tasks. Notion co-founder Akshay Kothari posted internal data that matched Ramp’s insights: “tokens go brr. Costs go ahhh.”

The financial impact is already visible as Anthropic nears its IPO. Bloomberg reported Anthropic’s potential valuation sank from an estimated $2 trillion in early October to $1.2–$1.3 trillion, with a possible high “junk” rating expected from credit agencies. OpenAI also lowered its proposed valuation to $1.25 trillion from $1.4 trillion and cut its latest fundraising goals by about $10 billion. Ramp’s October data showed a re-acceleration, with token spend up 8.9% from September 23 to October 23, but Kharazian says headwinds on spend growth remain strong.

Kharazian is skeptical that personal agents will provide the next growth boost. AI coding emerged about a year and a half ago as the most commercially viable product and helped make Anthropic and OpenAI trillion-dollar companies. OpenAI, SpaceXAI, Meta, and startups like Instinct have released personal agents, but Kharazian argues they are “not as economically productive as coding.” The projection has been that new products would drive further growth, but based on what is commercially advanced today, he says AI spend has started to slow down.