AI might cost more than human employees right now
Artificial intelligence showed up promising to slash costs and supercharge company productivity. But the script is flipping upside down.
Instead of saving money, many organizations are watching their AI bills surpass, and in some cases by a wide margin, what they spend on their own employees.
The IT budget, which was already one of the most contested line items inside companies, has become the stage for a brand new tension: is it better to invest in people or in AI tokens?
That question, which used to feel far off, is already knocking on the doors of CTOs, CFOs, and CEOs around the world. 👀
And what is happening right now could completely change the way companies think about costs, growth, and the future of work.
When the AI bill outpaces payroll
It is no exaggeration to say the corporate world is stunned by the numbers coming in. Recent reports from mid-size and large companies show that costs tied to artificial intelligence platforms, including language model APIs, automation tools, and dedicated cloud infrastructure for AI, are growing at a pace far faster than any early projection had indicated.
Bryan Catanzaro, vice president of applied deep learning at Nvidia, put it bluntly when he said that for his team, the cost of compute far exceeds headcount costs. That statement, made to Axios, illustrates the scale of the disconnect forming inside tech companies. When an executive at one of the largest chipmakers in the world admits that AI infrastructure costs more than the people operating it, it is clear we are not talking about a one-off issue or an isolated case.
One of the most striking examples of this spending explosion came from Uber. The company’s CTO has already burned through the entire AI budget planned for 2026 because of token costs, according to a report by The Information. We are talking about a company that operates on a global scale and that, even with a robust financial planning framework, could not predict how fast AI spending would snowball. That puts the challenge in perspective: if a company the size of Uber cannot keep this kind of cost under control, imagine what is happening at smaller companies that jumped on the AI bandwagon without the same financial governance structure.
On the flip side, some are riding this wave with pride. Amos Bar-Joseph, CEO of Swan AI, went out of his way to show off his Anthropic invoice in a viral LinkedIn post, claiming he is building the first autonomous business that scales with intelligence rather than headcount. This kind of stance reveals a mindset gaining traction in certain Silicon Valley circles: the idea that hiring fewer people and spending more on AI is, by itself, a sign of innovation. But does that math actually work in the long run?
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
The global numbers do not lie
To understand the scale of the phenomenon, it helps to look at the macro numbers. According to Gartner, global IT spending is expected to hit 6.31 trillion dollars in 2026, a jump of 13.5% compared to 2025. That growth is being driven by what the consultancy calls sustained momentum in AI infrastructure, software, and cloud services. This includes everything from building data centers optimized for language model processing to the cumulative cost of AI platform subscriptions that companies are signing up for.
This increase is not spread evenly. The companies feeling the biggest impact are exactly the ones that positioned themselves as early adopters of generative artificial intelligence. They were the first to integrate models like GPT, Claude, and Gemini into their workflows, and now they are the first to realize that the consumption-based pricing model can turn into a financial trap when internal adoption grows unchecked.
It is worth pointing out that the problem is not with the technology itself. AI genuinely delivers value in many scenarios. The problem lies in how companies are managing adoption and, above all, how they are failing to predict and control the costs tied to that adoption.
The IT budget in the eye of the storm
For years, the IT budget was treated as a necessary but controllable cost. Companies had a pretty good idea of what to expect: software licenses, server maintenance, tech support, and occasionally some digital transformation project with a defined timeline and price tag. Artificial intelligence broke that pattern in a way very few executives were ready to handle.
Consumption-based pricing models, which are the norm for most modern AI platforms, introduced a level of variability that is extremely hard to predict and even harder to control when adoption spreads across the organization. A company that starts with a small pilot can, within just a few months, find itself with dozens of teams using the tool and a bill that has multiplied tenfold without anyone having made a formal decision to let that happen.
This phenomenon is being called AI sprawl, meaning the uncontrolled expansion of AI tool usage inside organizations without centralized governance. It is the modern equivalent of shadow IT, that situation where teams start using unapproved software because they find value in it. With AI, the problem is even more complex because the financial impact can be immediate and significant.
A product team that decides to experiment with an image generation API, a support team that hooks an AI assistant into the ticketing system, a data team that uses language models for document analysis: each of these cases, on its own, might seem small, but together they blow a hole in the IT budget that only becomes visible when the monthly invoice shows up.
Proving ROI is no longer optional
Even companies with the biggest IT budgets on the planet will need to, sooner or later, demonstrate a return on their AI investment. Especially the ones answering to shareholders on quarterly earnings calls. The free experimentation phase, where simply saying the company was investing in AI was enough to excite the market, is coming to an end.
Now, what people expect to see are concrete proof of productivity gains or metrics that show a clear return for all that spending. It is no longer enough to say the team is using AI. You need to show how much that adoption saved, how much it sped up delivery, or how much additional revenue it generated.
Brad Owens, vice president of digital work strategy at Asymbl, a company focused on workforce orchestration, captured this shift in mindset well. According to him, the conversation is shifting toward figuring out what the real value of a worker is, whether that worker is human or digital. That is the kind of reflection that forces companies to be more deliberate and less swept up in excitement when deciding where to allocate resources.
The battle between the big AI labs
One aspect that a lot of people are not following closely yet is how the escalation of costs is shaping the competition between the leading artificial intelligence labs. The surge in corporate AI spending has created an interesting dynamic: enterprise customers are starting to compare not just model quality, but also the cost efficiency of each platform.
An OpenAI investor told Axios that this shift could benefit the company, since they consider Codex superior to Claude Code when it comes to maximizing efficient token usage, bringing down utilization costs for the end user. In other words, the war between the major AI labs is migrating from the field of raw capability to the field of economic efficiency. Whoever delivers more results per dollar spent will come out ahead.
Anthropic, for its part, has already adjusted its pricing structure to deal with a surge in demand. This shows that the labs themselves are reacting in real time to market dynamics, and that AI prices are far from stabilizing. For companies that rely on these platforms, this means living with a financial uncertainty that simply did not exist when the main technology cost was an annual software license with a fixed price.
Jobs versus automation: the tension that will not go away
At the center of this entire financial discussion sits a question that hits close to home for a lot of people: if AI is costing more, does it still make sense as a replacement for human jobs? The initial narrative was straightforward. Automate repetitive tasks, reduce the need for human labor in certain roles, and redirect the investment toward higher-value areas.
But what companies are finding out in practice is that the relationship between AI and employment is far more complex than that simple logic suggested. In many cases, companies that adopted AI at scale did not reduce their headcount. Quite the opposite, they needed to hire new, highly specialized profiles to manage, train, and oversee the artificial intelligence systems that were deployed. The result is that instead of replacing people, AI added a new layer of technical jobs that was never part of the original plan.
That does not mean AI’s impact on the job market is not real or significant. In more operational roles, like document processing, first-tier customer support, and standardized report generation, automation is already reducing the need for human intervention in a very tangible way. But what we are seeing is that while some jobs disappear or transform, others emerge, and the new ones tend to require much more specific qualifications and therefore cost companies a lot more.
Prompt engineers, model fine-tuning specialists, AI solution architects, ethics and governance professionals in machine learning: these are the profiles in high demand, and none of them come cheap.
Productivity under the microscope
Productivity, which was the big argument in favor of accelerated AI adoption, is being reassessed more carefully. Recent studies show that productivity gains are real, but concentrated in specific tasks and among specific worker profiles.
A developer using AI to write code can be significantly more productive in certain activities, but a manager who needs to make strategic decisions based on context, relationships, and intuition does not necessarily see the same productivity boost from AI tools. That means companies need to be much more surgical in how they implement AI, pinpointing exactly where the gains are real and where the investment is not justified by the return.
The situation has become so critical that some of the largest consulting firms in the world are already advising their clients to create what they call AI centers of excellence, dedicated structures designed not just to develop and deploy artificial intelligence solutions, but primarily to monitor and control the costs tied to them. These teams act as budget guardians, evaluating every new AI project from a financial perspective before greenlighting implementation.
The critical point: when spending becomes risk
The most important takeaway from all of this might be the simplest one: when AI labs raise their prices, heavy spending on artificial intelligence can stop being a show of strength and turn into a liability. Companies that brag today about their multimillion-dollar invoices from AI providers might soon find themselves explaining to investors why they spent so much without a proportional return. 📊
Human labor, which was being treated as the cost to be eliminated, might end up being the more efficient option in many scenarios. It is not irony, it is the reality of the numbers crashing head-on into market enthusiasm.
What changes from here
The current landscape is forcing a maturity that the market needed from the start, but that the hype around artificial intelligence ended up delaying. Companies that rushed to adopt AI without a clear cost and governance strategy are now reviewing their contracts, consolidating tools, and making tougher choices about where it truly makes sense to invest.
This rationalization process is healthy, even if it is painful for those who need to explain to the board why the IT budget grew so much without the expected returns. The good news is that with more data and hands-on experience available, AI decisions are likely to get smarter and more aligned with the financial reality of each organization.
The future of artificial intelligence in business is not about abandonment, but about selectivity. The organizations that will come out ahead are the ones that manage to balance the real productivity promise that AI offers with rigorous management of the costs involved, keeping their eyes open to both the opportunities and the financial pitfalls this technology brings with it.
What is worth watching now is how rising costs will impact corporate spending at the major AI labs and whether the competition for efficiency will finally bring more rational pricing to the market. Until then, maybe the best strategy is the one no startup wants to hear: sometimes, hiring a person is still the smartest investment you can make. 🤷
