Investors are getting tired of waiting for Big Tech companies like Alphabet (GOOG, GOOGL), Microsoft (MSFT), and Amazon (AMZN) to deliver on their AI revenue promises. That’s the main takeaway from this past earnings season as companies continue to plow money into their ambitious AI offerings without offering much insight into when it will all pay off.

Google parent Alphabet spent $13 billion in capital expenditures in its most recent quarter building out its AI infrastructure, up from $12 billion in the same period last year. Microsoft spent $19 billion, an increase from the $10.7 billion the company spent last year.

Amazon, meanwhile, announced during its earnings report that it spent $30 billion in the first half of the year on spending related to its Amazon Web Services platform.

Despite such massive spending levels, companies have been relatively mum on AI revenue, instead choosing to either offer a percentage of growth attributable to AI or simply pointing to their overall cloud services numbers.

The problem is that the generative AI craze just isn’t living up to the hype. At least not yet.

“[Generative AI] is not yielding benefits that are in tune with … the hype in the market, or in tune even with what customers were expecting out of this,” explained Gartner analyst Arun Chandrasekaran.

“That's kind of the narrative that we see when we talk to enterprise CIOs today.”

Those benefits will come as generative AI technologies continue to mature, but, experts say, it won’t happen overnight.

Too fast with too little benefit

There are two issues at play for generative AI. The first is that tech companies are rolling out new AI models and platforms too quickly for their customers to keep up, and the other is that the technology is still relatively new.

“What is happening today is we see fast-paced and kind of furious innovation, and the CIOs are saying, ‘Hey, this is just coming too fast at us. We want you to slow down a little bit, and we want you to make sure that the products that you're building, or the models that you're building, actually work very well before you go and build that next model', ” explained Chandrasekaran.

It’s not just the speed at which companies are pumping out models, but also whether their customers can understand and trust them.

“It's going to take a while to mature and understand how organizations are going to utilize it effectively,” explained Ari Lightman, professor of digital media and marketing at Carnegie Mellon University’s Heinz College of Information Systems and Public Policy.