What happened
Ramp has published an AI Index that quantifies how U.S. companies adopt and spend on artificial‑intelligence products and services. The index draws on transaction data from Ramp’s corporate‑card, invoice and ACH processing platform, covering over 85,000 firms. Firms are counted as AI adopters when a paid line‑item for an AI‑related merchant appears in a given month. The index also reports spend per employee and ‑volume metrics for the subset of customers using Ramp’s Token Spend Management (TSM) product, which imports usage and billing data directly from AI providers. Ramp notes that its methodology likely underestimates total AI use because free tools and personal accounts are not captured.
Ramp’s AI Index is built on a of corporate‑card, invoice, and ACH transactions processed through its financial platform. The company’s internal models extract line‑item text from receipts and bills, then match merchant names and description fields to a curated list of AI‑related vendors. When a transaction is identified, the associated firm is flagged as having purchased an AI product or service for that month.
The index categorises firms by industry using NAICS codes predicted by Ramp’s internal models, and by size using revenue and employee‑count data from both internal and third‑party sources. This enables the calculation of adoption rates (the share of firms with at least one AI‑related payment) and spend per employee (total AI spend divided by employee count) for each segment.
A separate ‑spend metric is offered to customers of Ramp’s Token Spend Management product. This product pulls usage and billing data directly from AI providers, tagging spend by model identifier. While the token‑spend sample is not representative of Ramp’s entire AI spend, the company says its trends are directionally similar to the broader .
Ramp explicitly acknowledges that its figures likely understate true AI adoption because many firms use free AI tools or personal accounts that do not generate a corporate transaction. Consequently, the index should be interpreted as a lower‑bound estimate of corporate AI investment.
Why it matters
The Ramp AI Index provides a more granular, transaction‑based alternative to survey‑based estimates of AI adoption, which have historically suffered from under‑reporting and lag. By leveraging actual spend data, the index can reveal real‑time shifts in corporate AI investment, helping investors, policymakers, and competitors gauge the health of the AI market. The per‑employee spend metric offers a normalized view that can compare AI intensity across firms of different sizes, while ‑volume data sheds light on usage patterns for large‑language‑model services. However, the index’s coverage is limited to businesses that pay for AI through Ramp’s platform, and it excludes free‑tier usage and personal accounts, meaning the figures likely represent a lower bound on total AI activity. Understanding these limits is crucial for anyone using the index to forecast demand for AI infrastructure, talent, or related services.
Survey‑based AI adoption studies have struggled with response and timing delays; a transaction‑based approach offers a more objective, near‑real‑time signal of corporate behavior.
Investors can use the adoption rate and spend‑per‑employee metrics to identify sectors where AI is gaining traction, informing allocation decisions in AI‑related equities or venture funding.
Policymakers interested in the economic impact of AI can track whether corporate spend is accelerating or stalling, which may influence regulatory or workforce‑development initiatives.
The ‑spend data, though limited, provides early insight into which large‑language‑model families (e.g., GPT‑4, Claude, Gemini) are being consumed at scale, helping AI providers anticipate capacity needs.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
Which component of an AI application is the machine-learning model itself?
What to watch next
Future releases of the Ramp AI Index will be watched for trends in AI spend growth or slowdown, especially as major AI firms approach IPOs or announce pricing changes. Analysts will also monitor how the ‑spend segment evolves, since it can signal shifts in model usage (e.g., movement from older to newer LLMs). Finally, any methodological updates—such as expanding coverage to include free‑tier usage or integrating data from additional payment processors—could substantially alter the index’s accuracy and industry relevance.
Quarterly updates of the AI Index to see whether the adoption curve continues to rise, plateaus, or declines, especially in the context of broader macro‑economic trends.
Changes in the ‑spend composition that could indicate a shift toward newer, more capable models or a migration back to older, cheaper alternatives.
Potential expansion of Ramp’s data collection to include free‑tier usage or integration with other payment processors, which would improve coverage and reduce the current underestimation .
Reactions from industry analysts and AI vendors to the index’s findings, which could affect market sentiment and corporate AI budgeting decisions.