Generative artificial intelligence (GenAI) has moved past its peak in the Gartner hype cycle, but it hasn’t quite hit the mark, as analysts pointed out at the Gartner conference in Barcelona.
Alicia Mullery kicked things off at the European Gartner Symposium, diving into the two big races in AI: one among tech providers and the other focused on delivering AI outcomes safely. She really drove the point home, telling IT leaders, “This is your race.”
One key takeaway hit hard: there’s a real risk of burning through cash with GenAI. Mullery and Daryl Plummer, chief research analyst at Gartner, stressed the need to keep a close eye on the budget. “You’ve got to understand the bill and stay on top of it,” Mullery warned.
Plummer added that most organizations Gartner has surveyed aren’t ready for AI. “They’re not prepared emotionally, technologically, organizationally, or management-wise,” he said bluntly.
To steer clear of failure, Gartner suggested two strategies. One targets organizations looking to boost productivity, while the other aims at leveraging AI for transformational changes.
Running a proof of concept isn’t cheap—it can set you back anywhere from $300,000 to over $2 million. Even though leaders grasp the hefty price tag of training AI models on costly GPU hardware, Plummer pointed out that the expenses for AI inference can spiral quickly.
“Processing isn’t just costly; it’s extreme because AI models rely on matrix multiplication to handle the predictions. This requires GPUs—whether you buy them for your own data center or lease them from a cloud provider, both options hit the wallet hard,” he explained.
Plummer cautioned that tech providers are too focused on their own advancements without guiding customers on how to achieve the potential of these AI systems. He highlighted a big misstep from giants like Microsoft, Google, and Amazon: “They show us what we can do but not what we should do,” he said.
With many organizations feeling unprepared for the advanced AI solutions from major providers, Plummer noted that a staggering 75% of their budgets go toward IT consulting just to figure out how to harness the technology effectively.
“Getting a proof of concept off the ground demands even more budget,” he added, warning that costs are likely to keep climbing until organizations begin putting enterprise AI systems into production. Only then will they start to grasp how to manage ongoing costs effectively.
The analysts stressed that IT leaders need to pinpoint their desired outcomes. Those aiming for efficiency improvements—dubbed “AI-steady” organizations—typically run a handful of pilots, say 10 or fewer. In this case, it’s feasible to have a small team ensuring the AI systems function smoothly.
On the flip side, organizations that see GenAI as a game-changer tend to launch many more pilots. Gartner calls these “AI-accelerated” groups. The analysts expressed doubt about the ability of such organizations to manage the volume of AI systems they want to implement.
As a result, they predict an uptick in a concept known as TRiSM (trust, risk, and security management), which will play a crucial role in keeping AI systems compliant.