Companies are deploying more artificial intelligence than ever before, but their financial outlays on the technology are refusing to scale upward. A comprehensive new report from Ramp demonstrates that AI spending flatlined during the late summer months even as overall token volumes continued to climb. This discrepancy signals that intense price competition and increasingly savvy enterprise customers may be capping revenue growth for major players like OpenAI and Anthropic. According to Ramp's tracked metrics, token expenditure grew by a mere 1.5% across August and September, representing a sharp deceleration from July, when spending jumped 42% over the previous month.
Investor Expectations and Market Realities
As Anthropic prepares for its highly anticipated IPO, this data indicates that investors might be harboring misplaced assumptions regarding the sustained breakneck growth of artificial intelligence firms. The looming public offering brings these financial dynamics into sharp focus.
Ara Kharazian, lead economist at Ramp, noted that almost the entirety of AI expenditure originates from venture-backed, high-growth technology companies. While this model functions effectively as long as expenditures continue on an upward trajectory, any market-driven slowdown could cause these firms to decelerate simultaneously, potentially triggering cascading negative effects throughout the sector.
Top Tier Firms Trim Their Budgets
Kharazian pointed to emerging evidence that the top 1% of firms, described as the most AI-pilled organizations, experienced a slight stutter in their spending habits during August and September. These elite spenders dipped just below the $10,000 per-employee threshold they achieved during their summer peaks. He emphasized that a contraction within this specific cohort is sufficient to drag down overall revenue growth, notwithstanding the concurrent rise in usage volumes.
While Ramp's dataset captures spending growth among median enterprises and the top 10% bracket, these increases are insufficient to bridge the revenue gap. Median firms typically allocate between $12 and $15 per employee each month, whereas the top 10% spend between $600 and $1,000.
Kharazian questioned whether standard commercial use cases will ever justify scaling up to the $10,000 monthly per-employee mark established by the top 1% tier. Although strong demand persists for AI coding agents within software development and financial sectors, a medical clinic or a car dealership simply does not require that level of heavy financial commitment to artificial intelligence.
The Era of Price Wars and Buyer Leverage
Although October data registered a modest re-acceleration with token spend climbing 8.9% between September 23 and October 23, Kharazian maintains that strong headwinds continue to suppress expenditure growth. He clarified that this market cooling is not driven by a collapse in demand, given that utilization volumes remain robust. Instead, he attributes the trend to intense marketplace competition.
Aggressive price reductions enacted by OpenAI and Anthropic have successfully cheapened the cost of utilizing artificial intelligence. In September, OpenAI halved its API pricing for GPT-6 Sol and GPT-6 Luna. In direct response to the escalating price war, Anthropic slashed pricing for Fable 5.1 and Opus 5.5, a strategic maneuver that company filings revealed reduced its revenue growth by roughly $600 million. Both companies have additionally relied on enterprise discounts designed to lock in long-term spending, which have ultimately compromised their revenue targets.
This evolving dynamic has firmly established a buyer's market. While legacy technology playbooks often involve subsidizing expensive innovations to capture market share, Kharazian argues that traditional ride-hailing dynamics do not translate to the AI landscape due to hyper-abundant competition.
Major players like Meta and Nvidia are continuously introducing competitive enterprise offerings featuring inexpensive or open-source models. Furthermore, an array of international open-source developers alongside a novel tool called Jev—a lightweight utility designed for routing between predetermined outputs—have rapidly gained traction, emerging among the top performers tracked by Ramp.
Thriftiness and Operational Efficiency
The growing ability of businesses to complete complex workloads at significantly reduced costs represents another primary driver behind the spending slowdown. This efficiency stems from vendor discounts and corporate maturation in thrifty resource management. As organizations gain advanced proficiency in leveraging artificial intelligence, they implement default settings that prevent costly models from executing prolonged tasks, institute strict internal budget caps, and deploy routers to divert standard queries to cheaper lightweight models.
Notion co-founder Akshay Kothari highlighted matching internal data with a succinct observation: token usage surges while operational costs remain a pressing concern. These shifting paradigms challenge foundational assumptions regarding the long-term financial trajectory of artificial intelligence adoption.
Valuation Pressures and Future Outlook
Reflecting these market pressures, Bloomberg reported that Anthropic's prospective valuation dropped from an estimated $2 trillion in early October down to a range of $1.2 to $1.3 trillion. Credit agencies are also reportedly poised to assign the company a high junk rating. Concurrently, OpenAI lowered its proposed valuation from $1.4 trillion to $1.25 trillion and scaled back its latest fundraising goals by approximately $10 billion.
While industry optimists hope that upcoming innovations, such as advanced personal assistants introduced by OpenAI, SpaceXAI, and Meta, will spark a new wave of enterprise spending, analysts remain cautious about whether these tools will match the commercial productivity delivered by automated coding agents.
