The Ramp AI Index shows US corporate AI spending reached a new high in October 2026, but the underlying expansion was modest. Spending among Ramp business customers rose 1.65 percent from September, according to Ramp’s monthly research published on November 11. That combination matters because aggregate records can conceal a market where adoption is broadening slowly, revenue remains concentrated and model prices are falling rapidly.
Ramp AI Index: A record built on uneven spending
The new spending peak was driven less by the typical company than by a small group of heavy users. Ramp reported that the top one percent of companies by AI spending paid a median of 10,200 dollars per employee per month, up nine percent. Firms in the top 10 percent spent 801 dollars, roughly flat, while the median firm spent 16.32 dollars per employee, up 93 cents.
Ramp chief economist Ara Kharazian said the structural risks identified in his earlier analysis persist. One percent of companies drive 80 percent of revenue for OpenAI and Anthropic, according to Ramp. That leading group is largely made up of venture-backed companies in tech, finance and technical services. Outside those sectors, Ramp said AI adoption is barely spreading. AI News Trend summarized the result as a new high alongside a slowdown, a useful distinction for companies planning budgets and model vendors chasing revenue.
Price deflation weighs on revenue
The clearest change in the Ramp AI Index was price. The average price per million processed text tokens dropped 21 percent last month, according to Ramp. Token volume continued to grow faster than token spending, meaning usage expanded even as each unit became cheaper. Ramp attributed the trend to competitive pressure between OpenAI and Anthropic, together with open-weight models from China.
Business spending on OpenAI grew 2.1 percent month over month, compared with 0.5 percent for Anthropic, Ramp said. Chip Dispatch noted that falling unit prices are eating into revenue growth at model vendors unless token volumes rise quickly enough. That matters for evaluating the record headline: more processing does not necessarily translate into proportionate revenue growth when the price per million tokens contracts so sharply.
Despite that pressure, the largest model tier retained most of the wallet. The models identified by Kharazian as Opus, Fable, Sol and Astra accounted for 69.5 percent of total spending in October, up from 67.7 percent. Ramp characterized the move as a slight increase, providing evidence that customers with the heaviest budgets had not yet shifted fully toward less expensive options.
New departmental data reveals a second front
Ramp introduced departmental spending data in the Ramp AI Index for the first time. The new dataset is intended to show which teams are driving net-new expenditure and to help identify use cases spreading beyond technical work. Ramp said weekly departmental tracking would soon become available through the index data dashboard.
Software development teams accounted for 72 percent of all classified AI spending in October, according to Ramp. Their spending remained the largest category, but it had declined from the highs observed earlier in the summer. Ramp described engineering spending as stuttering despite rising usage, an important signal for vendors whose business models have been built around coding agents.
Customer support and customer success delivered the strongest growth. Their share of classified AI spending rose from 0.8 percent in June to 13.3 percent in October, possibly because companies were deploying AI chatbots for customer service. Ramp also recorded slightly higher spending by leadership teams. Kharazian cautioned, however, that customer support had not matched the rapid growth posted by development teams earlier in the year.
New customers may produce lower margins
Ramp expects support workloads to favor cheaper, faster models rather than the highest-priced flagship models. Kharazian argued that these use cases are a good candidate for cost-sensitive deployment, which could mean OpenAI and Anthropic receive less revenue per unit of work from the next wave of enterprise users. The Decoder therefore described new customer support adoption as a potential source of lower margins for the two vendors.
This shift could broaden the customer base while changing its economics. Engineering teams remained responsible for 72 percent of classified spending, but the growth in support and customer success suggests that AI spending is beginning to extend into a second major operating function. AI Bacon said support buyers focus on latency and price per ticket rather than leaderboard position. If that pattern holds, lower-cost models may gain relevance even while flagship models still dominate the existing mix.
Adoption broadens, with caveats
Overall adoption of AI services rose to 56.8 percent among Ramp businesses, an increase of 1 percentage point and slightly faster than September’s gain of 0.66 point. Anthropic adoption increased by 1.1 percentage points to 44.8 percent, while OpenAI adoption held flat at 40.6 percent. Ramp reported that SpaceXAI led breakout growth, with 6.3 percent of Ramp businesses paying it in October, up 2.2 percentage points from the previous month.
Kharazian expects the two-vendor race between OpenAI and Anthropic to quickly become a three-way race, placing greater price pressure on both. Ramp also stated that it had no evidence business adoption of SpaceXAI products would decline following recent market punishment linked to Elon’s relationship with the Trump administration. The Decoder was more cautious about what SpaceXAI’s Ramp figures mean, arguing that the 6.3 percent share most likely does not reflect the broader market and that growth was probably driven by Cursor.
Read the index as a directional signal
The Ramp AI Index draws on payment data from more than 85,000 US companies, giving it a substantial base, but Ramp cautions that its figures reflect market trends rather than the entire economy. Free AI tools and usage through personal accounts may be missing. The customer base also skews toward growth-oriented, tech-savvy and larger companies, so adoption is likely overstated relative to US businesses overall.
There is a further limitation on token volume and spending. Those separate metrics come only from companies using Ramp Token Spend Management and are not representative of overall AI spending on the platform. Ramp had also reported two straight months of virtually no growth in token spend for August and September. AI Bacon noted that the new department classification covers the subset Ramp can attribute to a team, so its first reading warrants caution.
For investors and enterprise buyers, the clearest conclusion from the Ramp AI Index is not that corporate AI demand has stopped growing. It is that the growth mix is changing. A record was set, but concentrated customers and falling token prices mean that record must be assessed alongside adoption, department mix and model economics. Support is emerging, SpaceXAI is gaining customers within Ramp’s sample, and competitive pressure is intensifying. Those shifts suggest a larger but not yet evenly distributed market, with lower margins potentially accompanying the next phase of adoption.