The Ramp AI Index measures AI adoption among American businesses using transaction data rather than traditional surveys, which can underreport adoption rates. The methodology analyzes billions of aggregated, anonymized transactions from over 30,000 U.S. businesses using Ramp's corporate card and bill payment platform.12 Firms are considered AI adopters if they have a positive transaction for an AI product or service in a given month, with AI tools identified through merchant names and line-item details from receipts and bills.1

Unlike survey-based approaches, this transaction-based methodology provides more timely and accurate measurements of business AI adoption.1 Companies are categorized by industry sectors using NAICS standards and segmented by business size based on revenue and employee counts.1 The index likely underestimates actual adoption rates since it doesn't capture businesses using free AI tools or employees using personal accounts for work tasks.1 This approach builds on previous academic work by Bonney et al. (2024) and offers a more data-driven perspective on how AI is penetrating the corporate landscape.12

The failure rate of AI initiatives has increased dramatically, with 42% of companies expected to abandon their AI projects in 2025, up from just 17% in 2024, according to S&P Global Market Intelligence analysis12. Organizations are discarding an average of 46% of AI proof-of-concepts before implementation, primarily due to challenges like high costs, data privacy concerns, and security risks1. Gartner's research similarly predicts that at least 30% of generative AI projects will be abandoned after the proof-of-concept stage by the end of 202534.

The underlying issues behind these failures aren't typically technological shortcomings but rather organizational challenges. These include poor data quality, inadequate risk controls, unclear business value, and the significant financial burden of developing and deploying AI models, with upfront investments ranging from $5 million to $20 million34. Industry experts suggest that successful AI integration depends on prioritizing tailored use cases rather than pursuing every AI opportunity, with some even recommending that organizations embrace these "failures" as valuable learning experiences in the experimental process25.

While OpenAI dominates the generative AI landscape with its ChatGPT platform reaching 500 million weekly active users as of April 20251, the company is experiencing market share challenges in specific segments. In the enterprise large language model (LLM) space, OpenAI's share fell significantly from 50% in 2023 to 34% in 2024, according to Menlo Ventures data2. This decline comes despite the company's impressive revenue growth trajectory, with annualized revenue hitting $10 billion in mid-2025, up from $5.5 billion in December 20243.

The competitive landscape varies dramatically depending on how the market is measured. While OpenAI controls nearly 80% of all generative AI tool traffic (approximately 190 million of 240 million daily visits)4, its position in the broader artificial intelligence software market is more modest at 7.63%, trailing behind competitors like Grok (51.40%) and Optimole (10.95%)5. In the foundation models and model management segment, OpenAI holds about 9% market share6, highlighting how the company's dominance varies significantly across different AI market segments and measurement methodologies.