When Every Company Calculates Correctly and In the End Everyone Loses
5 min read
In the first quarter of 2026, tech companies cut around 80,000 jobs-nearly half due to automation. Two recent studies reveal why this calculation adds up for individual businesses but becomes a trap for the economy as a whole. For SMEs currently deciding on AI investments, a clear-eyed look at both perspectives is worthwhile.
Key Takeaways
- The individual calculation works. When a company replaces a job with AI, it keeps the full savings. The cost is borne by the market, as lost income reduces demand.
- The collective loses out. If everyone automates at once, demand collapses faster than the economy can replace it. The study calls this the “automation trap.”
- The impact falls short. According to an MIT analysis, 95% of AI pilots deliver no measurable returns. Maturity determines the outcome.
Related:The real cost of interchangeable AI-generated content / Asia sourcing: what it really costs SMEs
The calculation that works for every company individually
Economists Gerry Tsoukalas from Boston University and Brett Hemenway Falk from the University of Pennsylvania modeled the mechanics behind this dynamic. Their core finding seems straightforward at first: when a business replaces an employee with AI, it saves the full wage costs. The follow-up costs, however, land elsewhere.
After all, the laid-off employee was also a customer. Their lost income reduces demand, impacting all businesses that sell products or services-far beyond their former employer. The individual company barely feels this gap, as the effect is spread across the entire market.
Why, in the end, everyone still loses
The sum of many rational individual decisions creates a dilemma. Every business automates because it pays off for them. The result? Purchasing power declines faster than the economy can generate new jobs. The researchers describe this as a prisoner’s dilemma: each optimizes for themselves, yet in the end, everyone is worse off.
Their proposed solution is uncomfortable. They advocate for a levy on automation, comparable to a tax on environmental pollution. Companies replacing jobs would pay for the lost demand, with the revenue funding further education. Traditional remedies like universal basic income or retraining programs, according to their model, aren’t enough on their own. Whether such a tax will gain political traction is another question entirely.
The Second Finding: Lots of Money, Little Impact
While one study looks at the macro level, a MIT analysis provides a view into the companies. Around 300 corporate projects were examined. The result is sobering: 95 percent of AI pilots yielded no measurable return, while only five percent noticeably boosted revenue.
Interesting is what this is due to. Purchasing from specialized providers succeeded in the evaluation about twice as often as in-house development. The greatest return came from silent automation in the back-office, while visible marketing projects lagged. Those who start where processes are already expensive and repetitive get more out of it than with the next chatbot on the homepage.
What the mid-sized sector should take away
For mid-sized decision-makers, two clear lines emerge by which AI investments can be measured.
Costly, recurring back-office processes with clear rules. Purchased, proven tools instead of in-house development. Measurable before‑after numbers before rollout.
Visible prestige projects without a clear process behind them. Workforce reduction as the sole goal. Pilots that never deliver a number yet continue running.
The automation trap is an argument for taking a closer look where efficiency ends and harm begins. The mid-sized sector decides project by project, rarely in one big sweep.
Frequently Asked Questions
What is the automation trap?
The automation trap describes a situation where every business rationally automates to cut costs, but collectively, society loses because the disappearing purchasing power reduces demand faster than new jobs are created.
Who authored the study?
Gerry Tsoukalas from Boston University and Brett Hemenway Falk from the University of Pennsylvania. They model the effect as a competitive demand externality.
Why aren’t universal basic income or retraining enough?
According to the authors’ model, these tools don’t address the root cause: the incentive for each business to automate more than is economically beneficial overall. Only an automation tax tackles this directly.
What does the MIT analysis say about AI’s ROI?
Out of around 300 projects examined, 95 percent delivered no measurable returns. Success was mainly seen in companies that purchased proven tools and applied them in back-office operations rather than high-visibility marketing projects.
Should SMEs delay AI projects as a result?
Delaying is rarely the answer. A better approach is selectivity: automate where processes are costly and repetitive, with clear metrics set from the start. Where the goal is simply to cut jobs, a second look is warranted.
More from the MBF Media Network
Also available in