MIT Report: 95% of Enterprise AI Pilots Fail to Deliver ROI

Enterprises worldwide are spending billions on artificial intelligence and getting almost nothing back. That is not a hot take. It is the finding of a major research effort from MIT's NANDA Initiative.

The MIT "GenAI Divide" report, based on 150 executive interviews, 350 employee surveys, and analysis of 300 public AI deployments, reached a damning conclusion: 95% of enterprise generative AI pilots fail to deliver measurable financial impact. This is after companies collectively poured US$30 to US$40 billion into AI in 2024 alone.

The numbers keep stacking up. RAND Corporation's 2025 analysis found 80.3% of AI projects fail to deliver intended business value. According to NTT DATA, 70-85% of generative AI deployments are missing their ROI targets. And Gartner predicted that 30% of generative AI projects would be abandoned after proof-of-concept by end of 2025. In reality, abandonment rates jumped from 17% in 2024 to 42% in 2025.

The Problem Is Not the AI

Here is what makes this failure rate alarming. The tools themselves are not broken. MIT found the core issue is a "learning gap." Generic AI tools like ChatGPT perform well for individual tasks. They stall in enterprise environments because they cannot learn from or adapt to how a specific organization actually works.

The $11B Market Betting on AI Orchestration Infrastructure

Orkes raises $60M Series B for AI workflow orchestration platform. Only 6% of enterprises see AI ROI; 89% of AI agents never reach production.

Mission MediaMission Media

In other words, dropping a powerful AI tool into a company does not automatically make that company smarter. The AI still needs to connect to the right data, the right workflows, and the right decision points. Without that, it just becomes a productivity gadget for individuals, not a business transformation engine.

McKinsey's State of AI 2025 found that while 78% of organizations use AI in some capacity, only one-third scale it across the enterprise. Only 39% report any measurable effect on overall business earnings. Most of those say AI accounts for less than 5% of that impact.

The APAC Gap Is Even Wider

For companies operating in Asia-Pacific, the numbers are particularly stark. Accenture and Deloitte research shows 91% of APAC organizations report productivity and efficiency gains from generative AI. Yet only 45% report any commercial impact.

The region's data foundations are a key reason. Only 10% of APAC respondents in Infosys's GenAI Radar APAC survey claim their data is fully governed. Nearly half of APAC businesses expect existing technical debt to slow down their AI ambitions. Despite this, APAC enterprises plan to increase AI spending by 15% in 2026.

Spending more on a broken approach is unlikely to fix it.

What the 5% Are Doing Differently

The companies that do see results from AI share a common approach. They embed AI directly into structured, governed business workflows rather than treating it as a separate tool or experiment.

Hershey Cuts Marketing Analysis Time to 3 Weeks with AI Platform

Hershey partnered with Mutinex and Tracer to automate marketing measurement, cutting analysis from months to three weeks. The real bottleneck isn't AI—it's data readiness.

Mission MediaMission Media

Appian, which bills itself as "The Process Company," is among the platforms built on this philosophy. The company connects AI agents to end-to-end business processes, rather than deploying them as standalone tools. Its implementation with TELUS, a Canadian technology company, saved 7,200 employee hours annually by embedding AI directly into core workflows rather than bolting it on as a standalone experiment.

At Appian World 2026 on April 28, the company unveiled new capabilities including AI-assisted development, Model Context Protocol integration for AI agents, and the general availability of its AI Document Center. These developments point toward a broader industry shift: AI that earns its keep by being wired into real work, not sitting alongside it.

The ROI Accountability Pressure Is Building

89% of enterprises have adopted AI tools. Only 23% can accurately measure their return on investment, according to a KPMG report. That measurement gap is closing fast. Investor pressure for proof of AI ROI jumped from 68% of organizations in Q4 2024 to 90% in Q1 2025.

Boards and executives are increasingly asking for evidence, not experimentation. The era of "we're exploring AI" as a satisfying answer is ending. What replaces it requires a harder question: is our AI connected to real business processes, or is it still just a pilot?