fin3·Business·September 4, 2025 at 10:49 AM

Why Most Enterprise Generative AI Pilots Are Failing

MIT finds 95% of generative AI pilots fail to scale, with startups thriving while enterprises struggle due to misaligned investments and adoption gaps.

(8395x4315)_shutterstock_2652728871.jpg
Most-Enterprise-Generative-AI-Pilots-Are-Failing / Shutterstock.com
MIT Report: Why 95% of Generative AI Pilots Are Falling Short

A new study from MIT’s NANDA initiative, The GenAI Divide: State of AI in Business 2025, reveals a sobering reality: while generative AI continues to capture headlines and boardroom attention, the vast majority of enterprise pilots are failing to deliver on their promise.

According to the research, which draws on 150 executive interviews, a survey of 350 employees, and an analysis of 300 public AI deployments, only 5% of pilots have achieved rapid revenue acceleration. The remaining 95% either stall or show minimal impact on company performance.

Aditya Challapally, lead author of the report and a contributor to MIT’s NANDA project, explained the disconnect. “Some large companies’ pilots and younger startups are really excelling with generative AI,” he said. Startups founded by 19- and 20-year-olds, for example, “have seen revenues jump from zero to $20 million in a year. They succeed because they target one pain point, execute with focus, and partner strategically.”

For most enterprises, however, the challenges go beyond the models themselves. The report identifies a “learning gap” between what AI tools can do and how organizations integrate them. While executives often cite regulation or model limitations, MIT researchers point to flawed enterprise adoption. Generic tools like ChatGPT thrive with individual users but underperform in business settings when they fail to adapt to established workflows.

Misaligned investments

The study also highlights a mismatch in resource allocation. More than half of generative AI budgets are being funneled into sales and marketing tools, even though the highest returns appear in back-office automation—streamlining operations, reducing outsourcing, and cutting agency costs.

Adoption strategies matter as well. Companies that purchase AI solutions from specialized vendors and form strong partnerships see a 67% success rate, while those developing tools internally succeed only one-third of the time. This finding is especially relevant for heavily regulated sectors like financial services, where many firms are pursuing proprietary builds—often with disappointing results.

Challapally noted that few companies were willing to discuss their failure rates. “Almost everywhere we went, enterprises were trying to build their own tool,” he said. “But the data showed that purchased solutions delivered more reliable outcomes.”

The workforce impact

The report also explores how generative AI is reshaping workforces. Companies aren’t rushing to mass layoffs, but they are increasingly leaving roles unfilled as employees depart—particularly in customer service and administrative jobs once outsourced.

Meanwhile, “shadow AI” use remains widespread, with employees adopting unsanctioned tools like ChatGPT to fill gaps. Measuring AI’s effect on productivity and profitability remains a major challenge for leadership teams.

Looking ahead, MIT researchers say the most advanced organizations are already experimenting with “agentic AI”—systems that can learn, remember, and act autonomously within boundaries. These emerging models may define the next phase of enterprise AI adoption.

Big Deal: AI in Manufacturing Cybersecurity

Modern manufacturing is increasingly reliant on connected devices and industrial control systems—making the industry a prime target for cyberattacks. According to the State of Smart Manufacturing Report by Rockwell Automation, AI is emerging as a critical line of defense.

The survey of more than 1,500 leaders across 17 countries ranked cybersecurity as the second-biggest external risk, trailing only inflation and economic growth. Nearly half of cybersecurity professionals (48%) identified securing converged IT and OT systems as essential for success over the next five years—well above the 37% of all respondents who agreed.

However, challenges persist: a shortage of skilled talent, rising labor costs, and training needs remain major hurdles. As manufacturers recruit the next generation of workers, cybersecurity and data analysis skills are becoming top priorities.

Going Deeper: The Cost of Losing Black Women in the Workforce

In a new Fortune opinion piece, Katica Roy, CEO and founder of Pipeline, warns of the economic fallout from the departure of nearly 300,000 Black women from the U.S. labor force this year.

“This isn’t a seasonal fluctuation or statistical footnote. It’s a strategic failure with long-term consequences,” Roy writes. Historically, Black women have had the highest workforce participation rate of any group of American women. Their absence, she argues, threatens corporate succession planning, innovation, and overall economic stability.

Roy underscores that Black women have long been “a cornerstone of America’s economic engine—driving participation, powering key industries, and anchoring family incomes.” The current exodus, she cautions, fractures that foundation at a critical time for the economy.

Get our newsletter for the inside scoop on today's big stories and join us at X!
your-email@mail.com