Why Billions in AI Investment Can Be a Pitfall for Some Companies

(The Epoch Times)—The artificial intelligence (AI) gold rush has reached American businesses, but despite billions being spent, many companies aren’t seeing a return on their investment in the emerging technology.

The United States is the world’s leading investor in AI technologies. Tech giants such as Amazon, Google, Meta, and Microsoft have led the way in private sector investment and announced more than $100 billion in additional AI expenditures this year.

An analysis from CMS developer Storyblok noted that eCommerce businesses are spending, on average, nearly $400,000 on AI solutions for enhanced customer service experiences. However, only 32 percent reported even a “slight improvement” in operations resulting from their AI investment.

Massachusetts Institute of Technology (MIT) published research showing that despite U.S. companies spending upwards of $40 billion on AI investments, 95 percent have seen zero monetary return.

The study found that only 5 percent of integrated AI pilot programs are producing millions of dollars worth of value. Businesses stuck in the start-up phase of integration suffer what AI developer and vice president of Vapor IO, Kamil Mansuri, called “magic wand” thinking.

“Companies stuck in pilot hell usually have three issues: unclear success metrics, trying to solve everything at once, and treating AI as the goal instead of the solution,” Mansuri told The Epoch Times.

Mansuri said the MIT study findings didn’t surprise him because, in his experience, companies tend to treat AI like a magic wand instead of a tool for specific problems. Mansuri said the best way to avoid this pitfall is to steer clear of what he called “vague AI transformation.”

“At Vapor IO, we saw real ROI [return on investment] because we targeted concrete use cases like infrastructure optimization and automated failover systems,” Mansuri said.

“The difference is focus …. We cut cloud spend from $1.5 [million] to $800,000 by using AI for resource optimization because we knew exactly what problem we were solving.”

Mansuri believes the key to monetary return on AI investments comes from starting small and choosing an area with a measurable impact.

“The companies seeing results pick one specific pain point, prove value there, then expand,” he said.

Trial and Error

Mansuri is among many trying to peel back the corporate hype to expose what lies behind the investment-returns gap: a disconnect between integration and workflows. The MIT report also pointed to a lack of feedback loops or a misalignment with individual business needs.

“We invested a significant amount of money in making use of AI at Ranko Media and, overall, replacing humans didn’t work at all. Enabling our team to produce more output is where we found the best ROI,” Nick Rubright, CEO of Ranko Media, told The Epoch Times.

“For example, we create lots of content for our clients … We tried to automate content with AI, but the problem for us was that in GEO [generative search optimization] and SEO [search engine optimization], it’s winner-take-all. So we had to go back to leveraging human writers with real-world subject matter expertise because we needed to create content that would be competitive on the internet,” Rubright said.

He added that his company still uses AI to create content, but only the tools that are useful to his workforce and with the specific goal of expediting repetitive tasks.

By taking the focused integration versus human replacement approach, Rubright said AI has significantly improved his company’s content profitability.

“We make about 4x margin on content now, and the performance of that content has improved significantly across the board. I think it’s because humans have instinct from prior experience, but AI just sort of does what everyone else is doing and uses data, not experience,” he said.

Rubright also believes executives who view AI as a cheap replacement for human labor likely won’t see the returns they expect.

“There’s a lot of talk about AI being a human replacement, and lots of AI startups claim their tools can replace workers, but I’ve never found this to be true because these new tools still need management,” he said.

The phenomenon of high AI-tool adoption and low industry disruption rates is something the MIT report observed across 300 publicly disclosed AI projects, interviews with 52 organizations, and responses from 153 senior leaders at four key industry conferences.

So far, industries showing the most successful AI transformation include telecommunications and professional services.

The report also noted that established large language model AI programs such as ChatGPT have higher rates of successful corporate deployment than their custom-made counterparts. Many of the failed attempts to integrate custom AI tools into businesses were attributed to “brittle workflows, lack of contextual learning, and misalignment with day-to-day operations.”

Mansuri said he sees three main barriers come up regularly for companies struggling to get a return on their AI investment.

“First, data quality. You can’t build reliable AI on messy data. Companies rush to implement models without cleaning up their data infrastructure first. This is like trying to cook gourmet meals with spoiled ingredients,” he said.

The second one he noticed is unrealistic expectations at the executive level. Mansuri said many CEOs expect to see an “immediate transformation” after sinking money into AI technologies.

Finally, he said, many leadership teams try to retrofit existing roles instead of hiring people who actually understand both the tech and the business applications.

“You need engineers who can bridge the mercurial gap between cutting-edge AI capabilities and practical business value,” Mansuri said.

Fixing Bottlenecks

Starting small and having clarity around what AI tools are being used for has helped many business owners steer clear of an investment sinkhole, experts said.

“We’ve seen tangible ROI from AI because we started small and applied it to specific bottlenecks rather than chasing a big project,” Eric Turney, president of custom product manufacturing firm The Monterey Company, told The Epoch Times.

Turney said his company uses AI to generate SEO-optimized content and streamline customer responses, which has significantly reduced their cost per lead and improved lead response times.

“Unlike companies that stall, we’ve turned AI into a revenue driver by tying it directly to measurable outcomes,” he said.

Nick Strada, the founder of ad agency Bruiser Creative, is also seeing a fast return on his AI investment because he put it into production immediately. It has paid for itself, even with big projects.

“AI tools aren’t lab toys, they’re embedded in workflows that affect both cost and revenue,” Strada told The Epoch Times.

He said on the cost side, automation through AI tools has saved money on labor hours with tasks such as parsing briefs, scraping campaign data, and generating research reports. In terms of revenue generation, Strada said his company is seeing returns there, too.

He used a recent example of when a client came to his company with a seemingly impossible schedule and modest budget, which AI was able to help solve.

“By combining human craft with AI-enhanced workflows, including image generation, scaling tools, [and] automated asset preparation, we produced work that reached Cannes Lions scale. That success generated new client opportunities and reinforced relationships,” he said.

Strada said reaping the rewards from smart AI investments isn’t a theoretical concept: “It shows up as reduced operational overhead, new project wins, and the ability to make ideas come to life in ways that were previously not possible.”

Turney said successful AI integration in a business requires clarity on its place within the company workflow.

In his experience, Turney has noticed “companies often overinvest in experimental tools or broad strategies, but fail to integrate AI into their daily operations with accountability effectively.”

He said another corporate pitfall is treating AI like a “magic bullet,” rather than continually refining its role.

Mansuri said the formula for successful AI investment is simple: Industries with clear, measurable processes see faster returns. For example, he said logistics and supply chain optimization manifest revenue quickly because route planning and inventory management have direct cost savings that can be measured in real-time.

Alternatively, Mansuri said industries that are heavily regulated, such as health care, can take longer to show returns on AI investment.

“Industries with quantifiable processes and clear success metrics see faster ROI than those with subjective or heavily regulated outcomes,” Mansuri said.

In a March report, Morgan Stanley noted many company executives are optimistic about seeing a return on their AI investments amid rosy forecasts of hundreds of millions in profit gains over the next few years.

However, the investment bank tempered this enthusiasm by acknowledging that AI-investments with long-run payoff are challenging to identify.

“AI adopters are already outperforming the broader market,” Andrew Pauker, a director at Morgan Stanley Equity Research, said in a statement.

“Companies discussing AI adoption have been rewarded in fourth-quarter earnings.”