Corporate AI Adoption May Be Hitting a Plateau, Ramp Data Suggests

TL;DR:

• Fintech firm Ramp reports a leveling-off in corporate AI adoption, with its AI Index plateauing at 41% in May.
  
• Large businesses remain the top adopters at 49%, compared to 44% of medium firms and 37% of small businesses.
  
• Klarna’s AI downsizing misfire and increasing abandonment of AI pilots signal growing skepticism.
  
• Nearly 42% of firms have now dropped most generative AI pilots—up from 17% in 2024, according to S&P Global.
  
• Businesses are hitting real-world limits of today’s generative AI capabilities, suggesting a correction after early over-optimism.
  

Ramp’s AI Index Flags Market Maturity Moment

A new dataset from fintech firm Ramp offers one of the first concrete signals that the explosive wave of corporate AI adoption may be stabilizing. According to Ramp’s AI Index, U.S. businesses have started to hit a ceiling in new AI investments—after nearly a year of consistent month-over-month expansion.

Ramp’s AI Index draws insights from corporate card transactions and bill payments across 30,000 companies to estimate AI-related spending. While the data is not exhaustive, it offers a compelling proxy for real-world enterprise adoption trends.

In May 2025, the index plateaued at 41%, suggesting that the rush to onboard AI tools may be slowing as companies move from experimentation to assessment.

Corporate AI Adoption Metrics (May 2025)

A Tipping Point for Corporate AI?

Ramp’s findings may mark the beginning of a correction cycle in AI adoption. Following months of intense enthusiasm driven by tools like ChatGPT, Claude, and Copilot, many companies are now finding that the return on investment (ROI) from generative AI is lower than expected.

According to S&P Global, the percentage of firms abandoning most of their AI pilot programs has skyrocketed to 42%—a sharp rise from just 17% in 2024. This indicates that while AI remains strategically important, the early use cases may not have delivered measurable gains.

“We’re seeing firms take a step back—not because AI doesn’t matter, but because the initial wave may have overpromised,” said a market analyst at S&P Global.

Klarna’s Reversal Highlights AI Limits

A notable example of this reassessment is the case of Klarna, the Swedish fintech company. Earlier this year, Klarna announced plans to replace hundreds of support agents with AI, aiming to streamline customer service with bots.

However, that decision quickly backfired. After a sharp drop in service quality, the company was forced to re-hire some of the laid-off agents, illustrating a misalignment between AI capabilities and user expectations.

“The AI simply couldn’t handle the complexity or empathy required,” one executive familiar with the matter told TechCrunch.

Klarna’s case isn’t isolated. Across sectors, AI deployment is hitting barriers in nuanced roles, especially where customer interaction, ambiguity, or ethical considerations are involved.

Small Firms Trail in Adoption—For Good Reason

Ramp’s data also shows a clear adoption divide by company size. Larger enterprises, with deeper resources and dedicated tech teams, lead the AI adoption curve at 49%. In contrast, only 37% of small businesses have integrated AI tools into their operations.

For smaller companies, the costs, risks, and technical complexity of AI tools remain high. Many are waiting for clearer value propositions and more out-of-the-box solutions before diving in.

“Small firms often don’t have the luxury of failed pilots,” noted a fintech advisor. “They’re conservative adopters for a reason.”

From Hype to Usefulness: The Market’s New AI Phase

Ramp’s leveling data and rising abandonment figures may reflect a maturing AI landscape, not a retreat. As the hype cycle transitions into realistic application, companies are increasingly asking hard questions about ROI, productivity, and integration friction.

This isn’t to say AI is going away. Rather, we’re likely entering a phase where the focus shifts from adoption metrics to long-term optimization.

Analysts predict that success stories will depend on:

• Clear task alignment (matching AI to appropriate roles)
  
• Robust training data and workflows
  
• Human-in-the-loop models
  
• Ethical and compliance safeguards
  

As such, a plateau in new deployments may be a healthy recalibration, weeding out shallow pilots and elevating high-impact use cases.

What’s Next for Enterprise AI?

Looking forward, analysts expect AI to settle into specific verticals like legal, compliance, sales enablement, and internal search—areas where task structure and output expectations are well-defined.

There’s also growing interest in custom LLMs and smaller open-source models that can be fine-tuned for niche tasks at lower cost. Companies burned by general-purpose tools are beginning to explore alternatives like Mistral, Phi-3, and Meta’s Llama 3.

Meanwhile, enterprise platforms like Slack, Notion, and Salesforce continue embedding AI deeper into existing workflows—possibly reducing the need for stand-alone AI pilot programs altogether.

Ramp’s data doesn’t suggest the AI wave is over—it indicates the market is maturing beyond the novelty phase.