Companies are using more AI than ever before, yet their spending has plateaued.
A new report from Ramp indicates that AI spending flatlined during the late summer even as token volumes increased, suggesting that price competition and sophisticated customers are limiting revenue growth for companies like OpenAI and Anthropic. According to Ramp's data, token spend only grew by 1.5% across August and September, a stark contrast to the 42% growth recorded in July.
As Anthropic gears up for its IPO, these findings imply that investors might be overestimating the sustained breakneck growth of AI companies.
Almost the entire spend on AI is from venture-backed, high-growth tech companies, Ramp's lead economist Ara Kharazian told TechCrunch. He noted that this reliance creates a risk of simultaneous slowdowns if market conditions shift.
Kharazian observed that the top 1% of firms—those most heavily invested in AI—have slightly reduced their spending in August and September, falling just short of their previous summer peaks of $10,000 per employee. He argues that even a minor slowdown in this elite cohort can significantly impact overall revenue, despite rising usage volumes.
Ramp's data shows growth among median and top 10% spenders, though it is insufficient to offset the decline at the top. Median firms spend $12-$15 per employee per month, while the top 10% spend between $600-$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 stated. While demand for AI coding agents in sectors like software and finance remains strong, he doubts that industries such as doctor's offices or car dealerships will require such high expenditure.
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 up 8.9% from late September to late October, Kharazian emphasized that strong headwinds against spending growth persist.
Regarding the cause of the slowdown, Kharazian clarified that it is not a demand issue since volumes are rising. Instead, he attributes it to the competitive nature of the marketplace.
OpenAI and Anthropic price cuts have…just made it cheaper to use AI, he explained. In September, OpenAI cut API prices for GPT-6 Sol and GPT-6 Luna by half, while Anthropic reduced pricing for Fable 5.1 and Opus 5.5. This pricing war reportedly slashed Anthropic's revenue growth by about $600 million, and while discounts aim to lock-in spend, they are ultimately affecting revenue targets.
This dynamic has turned the market into a buyer's market. Kharazian compared this to ride-hailing subsidies, noting that the AI sector lacks the monopoly power to sustain such pricing strategies.
Meta and Nvidia are also introducing enterprise offerings with pretty cheap models or open source models, Kharazian added. He highlighted the rise of numerous open model companies from China and the US, as well as a new tool called Jev, which is increasingly among the top performers on Ramp.
For a lot of AI use cases, you just need a simple Yes/No, Kharazian said. Companies are leveraging these cheaper models and tools like Jev to reduce costs, making their existing technology even more affordable.
The ability to execute workloads at a lower cost is another factor driving the spending slowdown. This is due to both the discounts from AI providers and companies becoming more efficient in their AI usage.
As you get really advanced in using AI, you get better at figuring out how to control it, Kharazian noted. Advanced users are more likely to use budget controls, prevent long-running tasks on expensive models, and route standard workloads to cheaper alternatives.
Notion co-founder Akshay Kothari shared internal data that aligns with Ramp's findings, summarizing the trend with the phrase, tokens go brr. Costs go ahhh.
A lot of the assumptions that we had about the trajectory of AI spend are starting to break down, Kharazian said. He observed that users are adopting more AI but spending less per unit, which benefits customers but challenges model companies regarding revenue trajectories.
Bloomberg reported that Anthropic's potential valuation has fallen from an estimated $2 trillion in early October to $1.2 to $1.3 trillion, with the company facing a high junk rating. OpenAI also reduced its proposed valuation to $1.25 trillion from $1.4 trillion and lowered its fundraising goals by approximately $10 billion.
Kharazian believes Ramp's dataset reflects these investor concerns. However, there is a speculative silver lining regarding spending trajectories.
The hypergrowth thesis relies on new use cases emerging. While AI coding became a viable product a year and a half ago, driving trillion-dollar valuations, investors hope for another revolution.
Some speculate that personal agents will be the next major driver. OpenAI, SpaceX AI, and Meta have released their own personal agents, alongside startups like Instinct. Despite this, Kharazian remains skeptical that personal agents, which are less economically productive than coding, will provide the necessary boost for uninterrupted growth.
The projection is that there will be new products being introduced that will drive further growth, Kharazian stated. But based on what we can observe today and the products available and what is commercially advanced, AI spend has started to slow down.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.