Over the first 3 years of the current AI boom, it’s fair to say that most IT leaders’ biggest worry about AI was that they weren’t using it enough. They feared missing the boat, one which might have already left the dock with their competitors happily ensconced on board.
Now a growing concern is that IT teams and enterprises more generally may be using AI too much. The building storyline is, basically, that human employees may actually be cheaper than AI after all, at least for many use cases where AI was originally assumed to have an edge.
This article from The Next Web sums up some of the major recent reports and developments. The biggest issue appears to be the token-based pricing model employed in most AI scenarios. Some companies that encouraged widespread AI experimentation, as well as those that went all-in on specific use cases, most notably coding, report costs far exceeding expectations. The culmination came in this month’s report that none other than Microsoft, which has driven the AI push as hard as any company in the world, is telling its developers to use internally-owned GitHub tools instead of paying for Claude Code. But it’s not just Microsoft. Uber reportedly exhausted its full-year budget for AI coding by the end of April, according to The Information.
The Next Web insists this situation doesn’t reflect badly on AI technology. Rather, “the Claude pullback is the most credible signal yet that the unit economics of enterprise AI coding do not, at current token prices, work,” they write. “Not because the tools are bad. The opposite: they are good enough that engineers use them constantly, and the constant use is what breaks the maths.”
So the issue is budgeting for a world where technology providers are trying to earn back their capital investments in ways (and at a scale) that may not align with the way enterprise IT shops run their business. The Next Web writes, “The [budget] forecast was wrong because the variable being forecast, token consumption, behaves nothing like the licenses and seats that finance teams know how to model. A traditional enterprise software deal is denominated in users.”
The issue of AI’s cost as compared with that of human employees has direct relevance to the contact center, of course. To date, we’ve seen CX vendors as well as enterprise adopters trumpet the success of AI CX by citing basic call deflection statistics, in which every successful self-service interaction is treated as a win for the AI application that enabled it. But given the realities we’re seeing from the world of coding, it might behoove enterprises to dig a little deeper on these call deflections.
If the self-service interaction required a long string of back-and-forth between the customer and the AI, could the token consumption reach a point where it would have been more economical for a competent agent to spend a few seconds on the phone with the customer to answer their question? Or should AI-driven self-service automatically default to an agent interaction if it’s not resolved within a given number of exchanges? (Presumably this would also rescue users from a likely-frustrating exchange.)
Enterprise IT leaders can no longer afford to assume that AI is cheaper than humans, especially as the hyperscalers grow more focused on securing returns on their massive capital investments, likely by raising token costs. The economics of AI transactions and interactions will require a closer look.
Reimagine Collaboration and Communications with Enterprise Connect
All In on What’s Next
For over 35 years, Enterprise Connect has been the premier event where IT leaders, solution providers, and industry analysts come together to define the future of enterprise communications. As the most trusted conference in the industry, we deliver cutting-edge insights, world-class networking opportunities, and hands-on solutions that drive real business success.
Enterprise Connect is more than an event—it’s an experience that empowers enterprise communications leaders to reimagine the future of work. With elevated content, curated networking, and an engaging atmosphere, it remains the must-attend conference for professionals shaping the industry’s evolution.
Our mission is to equip you with the tools, strategies, and actionable insights needed to address your organization’s unique challenges. With industry transformation in mind, this year’s program introduces fresh new conference tracks, including:
- Legacy Unleashed: Optimizing Your Communications Backbone
- Workplace Reimagined: Engineering Tomorrow's Collaboration
- Tech Disruption Zone: Implementing Without Imploding
- Achieving Transformation: Architecting Your 3-5 Year Roadmap
- Future Shock: Communications Tech Beyond the Horizon
Featuring the most comprehensive program of objective content and an expansive vendor-neutral expo hall, Enterprise Connect is your ultimate resource for advancing collaboration, CX, and AV strategies. Join us to stay ahead, connect with industry trailblazers, and lead the transformation of enterprise communications.