Wall Street gets antsy for AI to pay off

The biggest names in tech have poured billions of dollars into turning AI into something companies and consumers can use. But can they make money on it?

Big Tech CEOs seem to be competing for who can fit the most mentions of AI into an hour-long earnings call, and capital spending on data centers and other expensive infrastructure has yet to let up, despite consumer wariness and hurdles for companies aiming to deploy AI solutions.

But over a year after OpenAI’s ChatGPT first burst onto the scene and ignited a flurry of investment, analyst questions in those earnings calls are getting slightly antsier about how all these big AI expenditures are going to ultimately pay off in the form of real revenue. Wall Street investors are clearly excited about AI, but they also seem to be increasingly wondering about the bottom line of it all.

“That’s the only thing they’re looking at,” R. “Ray” Wang, CEO and principal analyst at Constellation Research, told Tech Brew.

The profit picture

Wang said there are a number of areas that investors can look at to gauge how AI investments are impacting business for tech companies. Those include how AI is being used for cost savings, what kinds of new business models are being created around the tech, and whether companies have the “right partnerships to get them to the right use cases” to turn AI into a viable part of their business.

Wang said that at the moment, tech companies that are deploying AI internally are having the most success with using it to reduce costs, mentioning Amazon’s use of AI in its warehouses and Tesla’s use of AI in production as examples. But he stressed that it takes time to build out more complex AI models, and many are far from becoming viable businesses.

“The people that are generating revenue right now are doing cost takeout,” Wang said. “They’re doing optimization and automation that help people get to that level.”

Making money from AI could encompass a number of different consumer and enterprise businesses that cover many different use cases, but companies are still in the “early stages of monetization” according to Arun Chandrasekaran, an analyst at Gartner.

He sorted those into three pathways. One is charging consumers for access to a premium AI-powered service. Another is providing enterprise access to cloud services that allow for the monetization of both the models themselves and the underlying data infrastructure.

“In addition, these cloud providers also offer AI engineering tools to automate, customize, and secure these models, which is monetizable, too,” Chandrasekaran said in an email sent by a spokesperson, Catherine Howley.

Finally, tech companies are “embedding AI into many business applications” for business clients through enterprise products like Microsoft Copilot, Google’s Duet AI, and Salesforce’s Einstein GPT, Chandrasekran said.

Cloud clout

The importance of these cloud services was on full display during the most recent earnings season in late October, in which Microsoft’s strong cloud revenue growth seemed to quell Wall Street concerns about its heavy AI investments, while Google’s flagging growth in the sector spooked investors despite strong core advertising revenue.

Analysts at J.P. Morgan said in a recent research note that Microsoft’s enterprise cloud strength puts it in a strong position, and analysts and investors could be underestimating the importance of the cloud overall.

“Generative AI thematically dominates investor discussions, earnings calls, and analyst days across the software industry,” the analysts wrote in the note. “At this point, we sense investors are under-indexing to the long-term runway for Cloud…We emphasize the potential for Azure to enter a phase of faster new-workload share gains over time due to the halo effect of obtaining the generative AI leadership position.”

But the advent of AI-enabled enterprise cloud services is relatively recent: Microsoft 365 Copilot didn’t become generally available to enterprise consumers until November, and Google’s Duet AI landed just a few months before that. Investors will no doubt be watching to see how products like these fare in earnings reports in the coming quarters.

The price is right?

They will also likely be watching how tech companies set pricing structures for AI services to offset high costs. Adobe, for instance, recently rolled out a pricing structure that allots users a certain amount of AI credits with the option to buy more. And Microsoft and Google are raising price points on new AI-infused versions of their enterprise products after Microsoft reportedly lost money on an initial generative AI offering, according to the Wall Street Journal.

Despite investor worries that these initial pricing structures lacked a cap on usage—could one “very heavy Excel user…cause problems or erode margins”?—J.P. Morgan analysts allowed that it will take time for Microsoft to evolve how it monetizes these expensive new products.

“Yes [Microsoft Copilot] is a revolutionary product, but it will not drive much revenue in its first one to two quarters of availability,” J.P. Morgan’s analysts wrote. “Like a tsunami 5,000 miles out in the ocean, it has no real effect until it reaches the shore. Copilot will take some time to ramp.”