According to a new report, many businesses expect to increase their budgets for generative AI in 2025.

If there was one lesson that was extraordinarily clear at this year’s CES, it’s that generative AI is poised to be a massive force for businesses in the coming months and years. You couldn’t walk 10 steps through the Las Vegas Convention Center halls without encountering a new product featuring or promoting artificial intelligence in some form or fashion. AI spending doesn’t look to be slowing down anytime soon, either. Businesses spent $13.8 billion on genAI in 2024. That’s five times the $2.3 billion spent in 2023, according to data from Menlo Ventures. But determining what to spend on generative AI, whether it’s as a tool to help employees or a feature to include in your product or service, can be a challenge. Unfortunately, there’s no simple answer—and even some experts say they’re stumped, since AI is still such a new field. KPMG’s latest AI Quarterly Pulse Survey shows that 68 percent of large companies plan to invest between $50 million and $250 million over the next year. And a growing number of leaders are prepared to spend. One year ago, just 45 percent of companies planned investments in that price range. Among small businesses, the number is, of course, much lower. A report from ServiceDirect found over half of small businesses plan to spend more than $10,000 per year on AI tools. The amount they plan to spend varies by the size of the company. Some 58 percent of businesses with fewer than 10 employees plan to increase their AI budgets by more than $5,000 over the next 12 to 24 months, while 67 percent of businesses with 10 to 50 employees plan to increase their budgets by that amount or more in the same time frame. Among companies with more than 50 employees, 77 percent expect to increase their budgets by more than $5,000 on AI solutions in the next 12 to 24 months—on top of their existing heavy investments in the technology. ROI and pricing The big question, of course, is what sort of return will businesses see on their investment? The frenzy surrounding the technology has prompted some companies to adopt AI even when the benefits are questionable at best. (At the aforementioned CES, AI was being incorporated into everything from air fryers to plants.) Only one-third of leaders surveyed by KPMG say they expect to be able to measure the return on their investment in the next six months, with none believing they have yet reached that stage. “Leaders are putting real dollars behind agents, but with mounting pressure to demonstrate ROI, getting the value story right is critical,” said Steve Chase, vice chair of AI and digital innovation at KPMG, in a statement. “The dynamic nature of AI demands new ways to measure value—beyond the limits of a conventional business case. As leaders work to define the right metrics, those measures must be tightly aligned with the business strategy and should account for the cost of not investing.” Part of the problem is the cost of generative AI in the first place. James D. Wilton, a former associate partner at McKinsey & Company and founder of Monevate, a pricing and monetization consulting firm, makes the argument that the pricing models AI firms currently use are hurting the industry’s adoption. There are two types of pricing models used most frequently these days among AI companies: license fees and per-query charges. Neither is ideal, says Wilton. The subscription model assumes a one-size-fits-all approach, but that license fee is often so high that it is unaffordable for smaller users. The charge-per-query might make more sense, but users don’t get value every time they ask a genAI system a question. It often takes several iterations before the technology gives you the answer you’re looking for. “The challenge is it’s not very value aligned,” says Wilton. “It’s aligned to the costs the AI generators will incur, but that’s not necessarily where the value is for the user.” One alternative, he says, is outcome-based pricing models, which charge businesses per satisfactory resolution. (Zendesk offers something like this currently, he notes.) “The more directly you can tie your pricing to the way the product creates value, the lower the ROI you need to give the customer, because the customer will do less work in order to realize the value,” he says. BY CHRIS MORRIS @MORRISATLARGE