Ramp data shows AI spending flatlining as Anthropic nears IPO

TechCrunch reports that corporate AI spending has effectively stalled, with Ramp data showing token spend growing only 1.5% across August and September, a sharp deceleration from the 42% month-over-month growth seen in July. This divergence between rising usage volumes and flat revenue signals a structural shift in the AI market, where price competition and customer efficiency are capping the top-line growth that investors have long assumed for OpenAI and Anthropic. As Anthropic prepares for its initial public offering, this data suggests the era of breakneck hypergrowth may be ending, forcing a re-evaluation of the economic thesis underpinning the industry’s trillion-dollar valuations.

The details

Ramp’s lead economist, Ara Kharazian, attributes the slowdown not to a lack of demand, but to a buyer’s market dynamic driven by aggressive price cuts and improved cost management. In September, OpenAI halved API prices for GPT-6 Sol and GPT-6 Luna, while Anthropic significantly reduced pricing for Fable 5.1 and Opus 5.5, a move that reportedly slashed its revenue growth by approximately $600 million. Kharazian notes that the top 1% of firms, the most "AI-pilled" customers, have seen their per-employee spend stutter slightly below the $10,000 monthly peak they hit in the summer. While median firms spend between $12 and $15 per employee per month, and the top 10% spend $600 to $1,000, the aggregate growth from these cohorts is insufficient to offset the revenue loss from the high-end segment. October data showed a slight re-acceleration, with token spend up 8.9% from late September to late October, but Kharazian warns that headwinds remain strong.

Context

This spending pattern mirrors the early days of ride-hailing, where companies subsidized usage to lock in customers before raising prices. However, Kharazian argues this playbook fails in AI due to intense competition from Meta, Nvidia, and open-source models from China and the United States. The market is also seeing the rise of lightweight tools like Jev, which routes simple tasks to cheaper models, further suppressing average spend. Bloomberg reported that Anthropic’s potential valuation has dropped from an estimated $2 trillion in early October to $1.2 to $1.3 trillion, with the company expected to receive a high "junk" rating from credit agencies. Similarly, OpenAI lowered its proposed valuation to $1.25 trillion and cut its fundraising goals by about $10 billion. These adjustments reflect a broader market correction where investors are questioning the sustainability of revenue trajectories in the face of commoditizing inference costs.

What it means for developers

For developers and engineering leaders, this shift creates a buyer’s market that favors cost optimization over raw capability. The data suggests that simply increasing token volume no longer guarantees business value; efficiency is the new metric. Teams should audit their AI workflows to identify tasks that can be offloaded to lighter, cheaper models via routers. As Kharazian notes, advanced users are increasingly imposing budgets on employees and using defaults that prevent expensive models from running on long, low-value tasks. This is particularly relevant for industries outside of software and finance, where the economic justification for high per-employee AI spend is weaker. Developers should prioritize building robust routing logic and monitoring cost-per-task rather than just total spend, as the competitive landscape now rewards those who can deliver outcomes at the lowest possible inference cost.

What to watch

• Anthropic’s IPO filing: The final prospectus will reveal how the company frames its revenue growth and margin structure in light of the recent price cuts.
• OpenAI’s enterprise pricing: Whether OpenAI introduces new tiers or discounts to counter Anthropic’s moves will determine if the price war continues or stabilizes.
• Adoption of routing tools: The market share of lightweight routing platforms like Jev will indicate how quickly enterprises are shifting to cost-optimized architectures.
• New use case emergence: Investors are betting on the next "coding agent" revolution, such as personal agents from OpenAI, Meta, or startups like Instinct, to drive future spend. Watch for commercial traction in these areas to see if they can reverse the spending slowdown.