The cost of AI in business: What started as “the smart thing to do” has turned into a financial burden

🤖💸 It seems that integrating AI into the corporate world has become a headache for business owners, despite initial promises of a smarter, more cost-effective future. Following waves of mass layoffs intended to favor AI automation, a study from MIT (pdf) suggests that the cost of utilizing AI tools far exceeds human salaries. What exactly does this mean?

Companies are losing twice

Double trouble: The variable costs of AI tools — query fees, processing costs, premium subscriptions, and more — have become a financial weight for companies handling mns of requests and thousands of daily operations. At the current pace, total AI spending is projected to reach USD 5.2 tn by 2030, including USD 1.6 tn for data centers and USD 3.3 tn for IT equipment, according to McKinsey data cited by Fortune. All of this makes the equation of replacing humans with AI a losing wager when compared to the relatively fixed and stable cost of employee wages.

The study revealed that AI is more economical than human labor in only 23% of tasks. In contrast, infrastructure and implementation costs exceed employee wages in 77% of cases. This gap is most evident in computer vision systems, where the cost of installation can be five times higher than hiring an experienced quality control observer for three years, according to the study.

These shifts have rapidly drained the annual budgets of major firms. “I'm back to the drawing board because the budget I thought I would need is blown away already,” Uber’s CTO Praveen Neppalli Naga told The Information, referring to the steep costs the ride-hailing giant faces while adopting AI-powered coding tools like Anthropic’s Claude Code.

The loss isn't just financial: Companies now face a hidden cost known as the “hallucination tax,” which forces business owners to pay twice: once to run the large language models, and a second time to fund the human salaries of those who monitor and edit the output quality. On average, employees spend over four hours a week auditing AI outputs and correcting factual or logical errors, translating to estimated losses of USD 14.2k per year per employee, according to Forrester Research data.

A course correction

Learning the hard way: Regardless of the on-paper losses following layoffs, companies won't feel the gravity of the situation until it manifests as inflated figures in their ledgers. This has prompted a noticeable shift in the methodology of CFOs. According to a Gartner report, they're moving away from adopting and experimenting with AI at any cost toward measuring AI unit economics to determine exactly where money is being spent and what the real return is.

Some companies have moved past the initial infatuation phase and are now pivoting toward strict scrutiny of ROI, cloud spending, and computing costs. In addition to the variable costs draining budgets, companies face supplier instability — pricing structures and contracts change almost constantly, making AI budgeting a complex and difficult task. This gives the human element the advantage of relatively fixed costs.

Businesses are currently staging a tactical retreat from comprehensive automation policies, moving instead toward limiting these tools to high-value tasks that do not consume massive amounts of data processing. Moreover, due to the high cost of data filtering and continuous maintenance, humans are now being re-hired to evaluate model outputs and address hallucinations. Despite this, companies like Meta and Microsoft continue to lay employees off in droves.

So, what about us?

Good news for our region: These shifts represent a window for companies operating in Egypt and the Gulf. While labor costs remain relatively competitive across many sectors, the financial break-even point — where the cost of replacing an employee with AI becomes equal — is much further off than in global tech hubs like Silicon Valley. This means that for regional companies, particularly SMEs, relying on human talent is the wisest strategic choice to protect margins in the medium term.

The bottom line? There is less pressure to rush into total AI systems. Instead, the focus should be on tools that assist employees in their tasks rather than replacing them entirely, especially given a regional infrastructure that still requires further development.

(** Tap or click the headline above to read this story with all of the links to our background as well as external sources.)