AI return on investment: Lookie

While one company after another is reporting benefits from AI, most have yet to realize any return on investment.

Sooner or later, the inevitable question about artificial intelligence will arise: Is the investment boom pushing the corporate world into a bubble?

The answer, predictably, lies in the ability of companies to demonstrate a return on their investment (ROI), where the benefits they get from the technology exceed the cost of employing it, expressed as a ratio in which the numerator is greater than the denominator. Yet according to survey after survey, a quotient greater than one has proven elusive.

Yes, companies are realizing the benefits from AI, especially Generative AI (Gen AI), which can create text, images and video from signals using large language models trained on data. Companies including American Express, AstraZeneca, Bank of America, General Mills, Mass General, PayPal, Siemens, Unilever, and Walmart have reportedly seen improvements in research and development, manufacturing, logistics and inventory management, as well as customer and patient care, thanks to AI.

H&M is seeing improvement

Then there’s H&M. Facing high cart-abandonment rates and slow customer response times, the clothing retailer implemented an AI agent that provides personalized product recommendations, answers frequently asked questions, and guides customers through the shopping process. The company reportedly found that 70% of customer queries were resolved autonomously, conversion rates during chatbot interactions increased by 25%, and response and resolution times dropped threefold.

While H&M declined to confirm these figures, a spokesperson says the company’s use of AI “has had a positive impact on resource consumption, but also in terms of inventory, raw materials and emissions. AI also helps us create personalized customer experiences.”

For PayPal, it reportedly saw an 11% reduction in losses in 2023, thanks to the employment of AI models to monitor fraud patterns. A company spokesperson declined to confirm the report, but said PayPal uses AI “to enhance our manual risk controls as well as our fraud prevention and detection capabilities.” The spokesperson said PayPal has seen a meaningful reduction in peer-to-peer transactions related to the scam.

These types of benefits are supported by data. In a McKinsey report published last November, a survey of 1,993 respondents from companies in 105 countries found that the majority reported either cost benefits or revenue benefits from the use of AI. The largest savings were recorded in software engineering, manufacturing, and IT, while revenue gains were typically recorded in sales and marketing, strategy and corporate finance, and product and service development.

big disconnect

Yet despite the benefits companies have gained from AI, only a few report that the benefits outweigh the investment.

“Although AI tools are now commonplace, most organizations have not yet integrated them deeply enough into their workflows and processes to realize material enterprise-level benefits,” the authors of the McKinsey study wrote. Senior partner Alex Singla wrote that most companies introducing AI tools “have not yet produced use cases, redesigned workflows around AI and agentic capabilities, or built the platforms/railroads needed to run them at scale.”

A study of more than 300 publicly disclosed General AI initiatives published last July by the Massachusetts Institute of Technology found that despite $30 billion to $40 billion in venture investment, 95% of the projects produced no returns. A survey of 1,854 executives published in October by consultancy Deloitte found that while 85% of organizations had increased their AI investments in the past 12 months and 91% were planning to do so again by the end of the year, only 10% were realizing a “significant” return on their spend on agentic AI.

“Most organizations have not embedded AI deeply enough to realize enterprise-level benefits.” -McKinsey

And those investments are likely to get more expensive as AI vendors shift from a subscription model, which offers unlimited use at a fixed price, to usage-based pricing.

“Between 2015 and 2024, the number of consumption-based software companies is set to more than double,” McKinsey reported in November, adding that agentic AI has changed the underlying logic of using software.

in an interview with global financeNikolai von Bismarck, partner and leader of McKinsey’s service operations practice, describes the change as “structural and rapid.” He adds, “The cost uncertainty this creates is real: it is showing up as one of the top operational barriers to scaling AI in research.”

So it’s not surprising that banks are wary of lending to operators of data centers that support AI. JPMorgan Chase, Morgan Stanley and Sumitomo Mitsui Banking Corp are among those looking to sell their growing data-center loan exposures to private funds and insurers. For example, banks have been trying to syndicate a $38 billion loan package involving Oracle for six months and are offering it at a discount.

What explains the large gap between expectations and delivery in AI compared to previous rounds of technological innovation?

The simple answer is fear of missing out. Driven by unprecedented hype, the corporate world is vulnerable to a widespread fear, leading to undisciplined investing. As von Bismarck says, “Companies invest heavily in experimentation but struggle to identify use cases that have real impact. Many companies lack a clear road map to connect individual use cases to the broader business strategy, leading to fragmented efforts, duplicative investments, and no initiatives achieving critical mass.”

What will accelerate ROI?

“Organizations that successfully cross the general AI divide do three things differently,” the authors of the MIT report said, drawing on their survey results. “They buy rather than build, empower line managers rather than central labs, and select tools that integrate deeply while optimizing over time.”

But a more fundamental issue can prevent even the most prudent organizations from realizing ROI from AI: context, or, in technology terms, metadata. A recent article in Modern Data 101, a publication of Modern Data Company, a venture-capital-funded firm with offices in Silicon Valley and India, argues that AI agents are ineffective unless they both understand and have access to the data, and for that to happen, “the meaning of the data must travel with the data itself.”

Von Bismarck looks at this issue in terms of organization and culture. “Many of the companies that are leading the way have one common characteristic: They treat AI transformation the same way they treat any broader operating-model change, with strategic discipline, executive accountability, and a clear theory of how value is achieved.”