Portuguese 
English 
Spanish 
5 min of reading 
0 comments 
The Question Changed: From Technological Enchantment to Direct Profit Demands
In recent years, the most common question in boardrooms was simple: “What can we do with artificial intelligence?”.
Today, however, the focus has clearly shifted. Executives want to know something much more straightforward: where’s the money?
This shift reflects not just a market perception. On the contrary, it serves as a concrete financial alert. Recent data from Forrester shows that the majority of companies betting on quick returns with AI must cut or halt investments still this year.
-
Artificial intelligence is skyrocketing energy consumption, raising emissions from tech giants and pushing Google, Microsoft, and Meta closer to natural gas.
-
Mercor paid $1.5 million per day for doctors, lawyers, and former Goldman Sachs bankers to teach artificial intelligence to do their jobs, and in 17 months, it went from zero to $500 million in annual revenue while its own contractors accelerated the replacement of their own work.
-
Casio Unveils Moflin, Robotic Pet With Artificial Intelligence Designed To Provide Emotional Comfort And Simulate A Permanent Affectionate Bond
-
Google Maps Revolutionizes GPS Navigation With Gemini Artificial Intelligence, Ask Maps Feature, and Advanced 3D Visualization for Route and Travel Planning
Furthermore, many organizations have realized, in practice, that there is a significant gap between creating a well-presented pilot and putting AI into production within legacy systems. When this happens without planning, cash flow suffers and compliance is at risk.
In Brazil, this scenario worsens. With high Selic rates and pressure on the cost of capital, the cost of error has become unfeasible. In this context, innovation has ceased to be a corporate hobby.
Either the technology directly impacts profit, or it leaves the budget before the second quarter.
Why Only 5% of Companies Achieve Real ROI with Artificial Intelligence
According to recent studies, only about 5% of companies can achieve measurable ROI from AI projects. Meanwhile, the rest face pilots that don’t scale, accelerated cash consumption, and internal frustration.
As an entrepreneur in technology for decades, I view this movement with some relief. Artificial intelligence has never been a magic solution to inflate financial results. Still, many companies have tried to sell it that way.
In practice, AI acts as a process reengineering tool, not as an automatic profit generator. Research from MIT and Wharton indicates that only organizations that redesign end-to-end workflows capture sustainable value.
On the other hand, companies that merely “plug in” AI to old processes end up accumulating costs without a clear return.
Data, Structure, and the Most Common Mistake in AI Initiatives
Many managers still ask why their AI projects are not taking off. In most cases, the answer lies right at the base: lack of organized data.
Implementing advanced AI without the appropriate structure is akin to racing a Formula 1 car with adulterated fuel. The engine may be powerful, but it fails quickly.
An article published by MIT revealed a striking fact:
despite global investments between US$ 30 billion and US$ 40 billion, 95% of organizations have not achieved measurable returns with AI.
In light of this, the difference between “having AI” and “making money with AI” becomes evident.
The Invisible ROI: Why Accounting Doesn’t Recognize the Value of AI
Another recurring mistake involves how results are measured. Many companies believe that the value of AI appears only as increased revenue or direct cost cuts. However, this rarely happens this way.
In practice, much of the ROI from AI emerges as cost avoidance, meaning costs avoided.
A clear example appears in customer service. By using chatbots, one team can handle up to 20% more demands without hiring new employees. The real value lies in the hires that didn’t happen.
However, accounting does not record unpaid salaries. At the same time, cloud, token, and AI tool costs appear as immediate expenses.
Consequently, the company becomes more efficient, but short-term financial indicators worsen.
The Shift from Capex to Opex and Its Impact on the Balance Sheet
In the past, investing in technology meant buying servers and licenses. This model entered the balance sheet as an asset and depreciated over time.
Today, the logic has reversed. AI infrastructure operates on demand, in the cloud. Each interaction generates immediate Opex, reducing EBITDA and net profit.
Thus, even when the company builds future competitive advantage, short-term financial indicators suffer. Therefore, many leaders mistakenly interpret that AI “doesn’t work”.
Where Artificial Intelligence Really Makes Sense: The Strategic Matrix
Given this scenario, the right question shifts from “What does this return?” to “Where does this make sense?”.
The article The Gen AI Playbook for Organizations from Harvard Business School proposes a clear matrix based on two factors:
type of task and cost of error.
No Regrets Zone: Immediate Efficiency
This is where tasks with low cost of error and structured data come in, such as resume screening, drafting emails, and summarizing meetings.
In a selection process with 4,800 applications, AI helped select the 200 best candidates, drastically reducing human effort.
Additionally, Unilever reported that it reduced its hiring time from 4 months to just 4 weeks using a similar approach.
Copilot Mode: AI Produces, Human Validates
In this quadrant, the cost of error is high, but the data is structured. Programming and legal contracts fit here.
AI generates quickly, while the human reviews. This way, productivity increases without compromising safety.
Alternative Creator: Creativity Without Financial Risk
When the cost of error is low and knowledge is tacit, AI works well as a generator of options. Marketing, design, and brainstorming benefit from this model.
On Christmas 2025, Coca-Cola, in partnership with OpenAI and Bain & Company, launched the platform Create Real Magic.
The result was significant: over 1 million users in 43 markets.
Human Command Zone: Decisions That Cannot Be Automated
Finally, strategic decisions, layoffs, complex medical diagnoses, and high-risk investments should not be delegated to AI.
In this context, technology only supports analysis. The final decision remains human.
Wrong Metrics Kill Projects Before Maturity
AI projects rarely fail due to technical issues. Most often, they die from wrong metrics.
Evaluating AI solely by immediate ROI eliminates promising initiatives too early. More mature companies use intermediate metrics, such as cycle time, adoption rate, and exception reduction.
ROI emerges as a consequence, not as an initial criterion.
Conclusion: AI Doesn’t Destroy Balance Sheets, It Exposes Bad Decisions
In the end, artificial intelligence does not destroy balance sheets. It exposes poorly structured decisions, mistaken metrics, and unrealistic expectations.
The secret is not to stop investing, but to know how to steer.
Ignoring basic automation leads to inefficiency. Automating strategic decisions leads to value destruction.
The difference lies in correctly choosing where AI should operate.
In your company, is artificial intelligence generating real value or just consuming budget without clear metrics?
Seja o primeiro a reagir!