Ramp data shows AI spending stalling as Anthropic nears IPO

A comprehensive new report from Ramp reveals that corporate spending on artificial intelligence stalled during the late summer months, even as overall token volumes continued to rise. This unexpected slowdown points to intense price competition and savvier customer cost controls, which could challenge long-held investor assumptions about uninterrupted AI hypergrowth just as Anthropic gears up for its initial public offering.

• Token spending grew by a modest 1.5% across August and September, sharply down from a 42% jump in July.
• High-growth, venture-backed tech companies drive nearly all AI spending, but top-tier firms saw a slight dip in outlays.
• Price cuts by OpenAI and Anthropic, alongside cheap open-source models from Meta and Nvidia, have created a buyer's market.
• Although October saw a brief rebound in spend growth, underlying market headwinds and efficiency gains remain strong.

Price Wars and Revenue Pressures on AI Leaders

According to Ramp data, the sudden spending slowdown is not driven by a drop in demand, as AI token volumes continue to climb. Instead, the shift is primarily the result of a fiercely competitive marketplace. Both OpenAI and Anthropic have slashed API prices significantly to capture market share, with Anthropic noting that its pricing reductions have reduced revenue growth by approximately $600 million.

This dynamic has firmly established a buyer's market. Major technology players like Meta and Nvidia are introducing enterprise-grade open-source and low-cost models, while specialized tools like Jev offer lightweight routing alternatives for simple tasks. These factors allow companies to bypass expensive frontier models for routine workloads, directly impacting the top-line revenue trajectories of foundational AI providers.

Growing Thriftyness and Strategic Cost Controls

Beyond provider discounts, corporate customers are simply getting better at managing and controlling their AI usage. As organizations gain experience with advanced deployments, they increasingly implement strict internal budgets, default settings that prevent unnecessary calls to expensive models, and automated routers that shift workloads to cheaper light models.

This trend is corroborated by other industry voices, such as Notion co-founder Akshay Kothari, who noted that while token volumes are surging, overall costs are being reined in. While this cost-efficiency is a significant win for enterprise customers, it introduces new structural questions regarding the revenue growth models of AI startups and major labs.

Implications for Valuations and Future AI Growth

The cooling spending trend is already reverberating through private markets and investor sentiment. Reports indicate that Anthropic's potential valuation dropped significantly from early estimates, while credit agencies eye conservative ratings. OpenAI has similarly adjusted its fundraising targets and proposed valuations downward in response to shifting market realities.

While investors continue to pin their hopes on the emergence of revolutionary new use cases—much like AI coding assistants a year and a half ago—current commercial data suggests that spending growth has undeniably plateaued. Whether upcoming product categories like personal agents can reignite hypergrowth remains an open question for the industry.