Most companies aren’t seeing a return on AI investments. This tech CEO wants to change that

By Clare Duffy, CNN

New York (CNN) — The artificial intelligence industry has a big problem: 95% of companies that try AI aren’t making any money from it, according to a report from the Massachusetts Institute of Technology last month. One AI executive believes he knows why.

“There has been this general promise of, hey, you’ll just plug in the (AI) model … and everything will work,” Jason Droege, CEO of startup Scale AI, said in an interview. “The reality is a little bit different.”

Scale is primarily known for helping to make AI work at a fundamental level. AI giants need massive amounts of data to train their large language models, and it needs to be labeled and categorized so the models can learn, for example, what’s a photo of a cat versus a photo of a fish — for years, they’ve turned to Scale to provide that organized data.

It’s such a crucial business that Meta bought a 49% stake in Scale AI back in June for $14.3 billion, valuing the startup at $29 billion in total. As part of the agreement, founder and then-chief executive Alexandr Wang and several other top employees left to work for Meta.

But the deal sparked some concern that rival LLM makers may be wary of working with Scale following Meta’s investment. OpenAI and Google have reportedly scaled back their work with the company.

Scale says that “data labeling” business has continued to grow each month since the Meta deal. But new CEO Droege — who initially joined Scale last year as chief strategy officer — is now also focused on a lesser-known element of Scale’s business, helping all kinds of companies compile their own, custom data sets and create AI tools to automate processes and take on rote tasks. He’s hoping to challenge the impression that implementing AI applications can’t be lucrative.

“I think companies thought it was a bit easier than it actually is,” he said. “But there is a ton of value when you get it right.”

Making AI make money

Despite clamoring from C-suites across Corporate America about the promise of the technology to make businesses run more efficiently — most companies aren’t seeing a return on their AI investments. MIT’s August report contributed to growing concerns that the AI market might be a bubble waiting to burst, even as the industry is propping up the US economy.

Among the prominent clients that Scale is helping to build AI applications are the Mayo Clinic, the Qatari government, Cisco and Global Atlantic Financial Group. And just last month, Scale signed a $99 million contract with the US Defense Department to develop AI applications for the Army.

The companies that fail to see a return on their AI investments are often trying to apply the technology to wrong kind of problem, according to Droege. “I think one of the misunderstandings is that AI is this magic wand or it can solve all problems, and that’s not that’s not true today,” he said.

Problems well suited for AI are ones where humans are “slow or inconsistent or error prone,” Droege said, such as reading through and summarizing or editing many pages of documents.

For example, Scale has helped organizations develop AI systems to process insurance claims and to give doctors a summary of patients’ medical histories ahead of visits.

If the idea of an AI helping to decide whether you get reimbursed for a medical procedure or reminding your doctor about your tricky medical condition seems worrying — Droege says companies also need human experts to contribute to and constantly improve the AI.

“If a healthcare organization is trying to provide a tool that assists a doctor in better diagnosing a patient … you would want your most senior doctors, your senior medical professionals who have expertise in these areas, using the application, giving it feedback, pointing out where there’s problems,” Droege said.

The entire process can take weeks or months, but it can ultimately result in a tool that’s more useful to employees than just a mainstream chatbot, Droege said.

He cited government agencies using AI to evaluate building permit applications ahead of a human review — trained on data from earlier review processes — to make the permitting process faster and smoother.

But some analysts say it could be years before AI tools really makes companies money.

“This is going to take years for large companies to implement AI tools that are broadly useful and generate revenue and save expenses,” said Gil Luria, head of technology research at DA Davidson. However, he added that “once we do figure out how to do AI in the organizational context, those tools will be very valuable and generate a tremendous amount of revenue.”

And Scale AI has plenty of competition, including industry leaders such as Amazon and Microsoft.

“The notion of developing AI applications for companies, everybody’s been chasing that for the last two and a half years,” said Luria. While Scale was first to the data labeling market, it “is now company number 10,000” competing in applications, he said.

Still, Droege says there’s plenty of opportunity for companies that really understand how AI works — and what’s it’s good for.

That sounds like a savvy marketing pitch, but it also aligns with conclusions from MIT’s research, which found the companies least successful at deploying AI were the ones who tried to build AI tools alone themselves, without outside help.

Expertise helps companies understand which problems AI can and can’t solve, Droege said.

Droge told CNN he’s optimistic about both the data side and the applications side of Scale’s business.

“Going into all of this, the application side of our business was already in the hundreds of millions of dollars in revenue,” he said. “And on the data side of the business, we’ve grown every month since the (Meta) deal … It’s a large business for us and we’re very happy with it.”

The-CNN-Wire
™ & © 2025 Cable News Network, Inc., a Warner Bros. Discovery Company. All rights reserved.