Beware the Hidden Costs of AI: Why “AI Included” Doesn’t Always Mean “AI is Free”

In a manner reminiscent of the arms race during the Cold War, businesses across the spectrum of commerce today are investing massive amounts of money in Artificial Intelligence (AI) in an effort to gain advantages over their adversaries in the marketplace. Organizations racing to deploy AI to improve productivity, automate tasks, and gain that competitive advantage have been barreling down the racetrack with blinders on, impervious to the potential downside of “too much too soon” with AI. Indeed, many companies have turned a blind eye to the possible existential consequences of AI today: namely, that AI can come with significant and unexpected costs.

Those costs are not limited to software subscriptions. Depending on how AI is deployed, organizations may incur expenses related to model usage, cloud infrastructure, implementation, third-party integrations, and consumption-based pricing. Understanding where those costs originate is becoming just as important as understanding what AI can do.

And those costs can vary wildly.

The Rising Cost of AI: Real-World Lessons

Fast Company recently reported that one business inadvertently spent nearly half a billion dollars in a single month on Anthropic Claude usage. After failing to implement proper controls and usage limits, this company has become a cautionary (and quite baffling) tale around AI usage: deploying AI without oversight or budgeting can turn out to be a recipe for disaster.

There are multiple comparable stories. TechCrunch also reported that Uber’s CTO revealed in April that its entire annual AI budget had been blown through in just four short months after Uber encouraged staff to use AI “as much as possible” – all while ranking internal usage on leaderboards to create a competitive environment around its AI usage. Microsoft, as well, reportedly restricted its engineers’ access to Claude Code due to soaring costs.

And in a "which came first, the chicken or the egg" type of fashion, Axios had this to share:

Companies are citing AI's ability to automate jobs as a cause for layoffs, though Anuj Kapur, CEO of CloudBees, told Axios that workforce cuts may simply be "the only lever they can pull" to offset their AI bills. The enterprise is undergoing a "healthy swing" away from AI overuse — or "tokenmaxxing," the push to burn as many AI tokens as possible — Ali Ansari, CEO of model training firm Micro1, told Axios.

As AI becomes embedded into everyday business operations, organizations are finding themselves part of a growing trend: the realization that the true costs of AI extend beyond the initial decision to utilize it. Understanding where AI costs originate – and how quickly they can accumulate – is essential to successfully managing the unique risks of AI today.

ERP AI Is Different, But the Hidden Costs Still Exist

These same concerns are beginning to emerge in the Enterprise Resource Planning (ERP) world, although they have yet to generate the same alarming headlines. As Oracle, SAP, Microsoft, and other enterprise software vendors rapidly embed AI into their platforms, understanding which AI capabilities are included in existing subscriptions and which trigger additional licensing or consumption-based charges is imperative. These ERP behemoths have never been known for their transparency in pricing, and the introduction of AI into so many aspects of ERP software is adding another element of instability and uncertainty.

Even in these early stages, there are some notable vagaries and potential for hidden costs with regard to Oracle’s AI capabilities. For example, Oracle has stated that many Fusion AI capabilities are included with existing subscriptions. However, customers are discovering that "AI included" often means that while the software capability is included, what’s required to make it useful frequently is not. There may still be additional costs associated with cloud infrastructure, AI consumption, data services, custom AI agents, implementation services, and integrations with third-party large language models.

In other words, "AI included" does not necessarily mean "AI at no additional cost." Organizations should understand not only what functionality is available, but also what infrastructure, services, and consumption models may be required before those capabilities deliver meaningful business value.

Enterprise software pricing models can change quickly, so organizations should remain cautious and not assume that AI capabilities within ERP will have little or no financial impact.

The Bottom Line: Understand the True Cost of AI Before You Deploy It

AI introduces a new category of ERP spending that many organizations are only beginning to understand. The organizations that manage AI costs most effectively will be those that evaluate licensing, infrastructure, implementation, and ongoing consumption before deploying AI at scale – not after unexpected invoices begin to arrive.

Whether your business has already incurred unforeseen ERP-related AI bills or you're simply concerned about your exposure to hidden AI costs, contact us. We can help you sort that out! Understanding your exposure before it becomes a budget problem is half the battle.

Published on July 16, 2026