AI spending growth stalls as Anthropic approaches IPO

Companies are utilising more artificial intelligence than ever before, yet their actual expenditure on the technology has not increased. A fresh report from Ramp indicates that AI spending remained flat during late summer, even as token usage volumes climbed, suggesting that price competition and more experienced customers may be limiting revenue for major players like OpenAI and Anthropic.

According to Ramp’s findings, token expenditure rose by only 1.5 per cent across August and September. This is a significant change compared to July, when spending grew by 42 per cent from the previous month. As Anthropic prepares for its IPO, the data indicates that investors might be mistaken to assume AI companies will sustain growth at such breakneck speeds.

“Almost the entire spend on AI is from venture-backed, high-growth tech companies,” Ramp’s lead economist Ara Kharazian told TechCrunch. “That’s fine as long as AI spend continues to rise, but if it starts to slow down because of something in the market or some other factor, they will all start to slow down simultaneously, and that could lead to other cascading effects.”

Kharazian noted he is already seeing evidence that the top 1 per cent of firms—the most “AI-pilled”—have experienced a slight stutter in their spending during August and September. These companies are currently falling just below the $10,000 per-employee figure they reached during their summer peaks. He explained that a slowdown among this specific cohort is “sufficient to slow down revenue entirely, despite the fact that we’re seeing rising volumes.”

Ramp data shows growth in spending among median firms and the top 10 per cent, but it is insufficient to compensate for the difference. Median firms spend between $12 and $15 per employee each month, while the top 10 per cent spend between $600 and $1,000.

“If you look at the use cases today, it’s not necessarily clear they will get up to that $10,000 per employee per month mark that the 1% is at,” Kharazian said. He noted that while there is significant demand for AI coding agents in sectors like software and finance, “it’s not necessarily true that a doctor’s office or a car dealership will need to spend that much on AI.”

“The era of hypergrowth may be over, and now that we know there may not be as much juice to squeeze out of the top 1%,” he continued. Although October data showed a re-acceleration, with token spend increasing by 8.9 per cent from September 23 to October 23, Kharazian warned that the headwinds against spending growth remain strong.

Kharazian, who has released a series of blogs titled “Cracks in the AI Thesis,” stated that the slowdown is not a result of a demand issue. Because AI volumes are rising, he instead points to the competitive nature of the marketplace. “OpenAI and Anthropic price cuts have…just made it cheaper to use AI,” he noted.

In September, OpenAI reduced API prices by half for GPT-6 Sol and GPT-6 Luna. Responding to this price war, Anthropic also made significant reductions to pricing for Fable 5.1 and Opus 5.5—a move that the company stated in a recent filing reduced revenue growth by approximately $600 million. Both companies have also offered enterprise discounts intended to secure spending, but they are currently struggling to meet revenue targets.

Kharazian observed that this dynamic signifies a buyer’s market. While subsidising the use of expensive technology to lock in customers has historically been a Big Tech strategy, he argues these dynamics do not function in the AI sector due to excessive competition.

“Meta and Nvidia are also introducing enterprise offerings with pretty cheap models or open source models,” Kharazian said. He also highlighted a range of other open model companies from China and the United States, alongside a new firm called Jev—a lightweight tool for selecting between predetermined outputs—which is increasingly appearing among the top performers on Ramp.

“For a lot of AI use cases, you just need a simple Yes/No,” Kharazian said. “So you take your already cheap models, you shift them over to Jev, and now your already cheap stuff gets even cheaper.”

The capacity to complete workloads at a significantly lower cost is another factor driving the spending slowdown. This is partly due to discounts from AI providers, but also because firms are becoming more thrifty. “As you get really advanced in using AI, you get better at figuring out how to control it,” Kharazian stated.

“You’re more likely to use defaults that prevent usage of very expensive models for long-running tasks. You’re more likely to impose some budgets on employees. You’re more likely to offload some of the spend from standard and light models to even cheaper light models via routers,” he added. Notion co-founder Akshay Kothari recently shared internal data that mirrored Ramp’s insights, succinctly remarking: “tokens go brrr. costs go ahhh.”

“A lot of the assumptions that we had about the trajectory of AI spend are starting to break down,” Kharazian concluded.