Artificial intelligence didn’t arrive as a tech trend—it arrived as a wave moving faster than any previous revolution. In less than two years, generative AI went from being a laboratory curiosity to becoming a tool that shows up in your WhatsApp, your ERP, your email, and even in the software you didn’t know you were using.
And yet, the new MIT study The GenAI Divide — State of AI in Business 2025 presents a striking finding:
Only around 5% of companies are achieving real, measurable impact with AI. The remaining 95% are seeing no return.
This doesn’t mean that AI “doesn’t work.”
It means most companies are implementing it without strategy, without clarity, and—above all—without understanding that AI is much more than a tool: it’s a new way of thinking about processes.
I work with companies that move millions: industrial parks, financial institutions, factories, printers, credit companies. And what I see every day fits PERFECTLY into this divide. Here’s what’s happening and how to avoid falling into that 95%.
1. The “I want AI… but I don’t know what for” phenomenon
In family businesses, mid-sized companies, and even regional corporations, something curious is happening:
Many want to implement AI just to say they have AI.
I don’t mean this disrespectfully. It’s human nature.
- The ghosts of Kodak, Blockbuster, and Nokia are still alive.
- No one wants to be “the business that didn’t adapt in time.”
So this happens:
A) They implement AI to show off, not to transform
To use it as a business card at the networking lunch, at the board meeting, on the golf course.
The problem? Showing off AI doesn’t generate ROI.
B) They invest budget without strategy
I see this constantly: companies with cash flow want to “do something with AI” and go straight into buying tools, copilots, chatbots, or platforms without understanding what problem they’re solving.
Which leads us to the third scenario:
C) When they do hire consulting… everything changes
Because then we discover:
- which processes can actually be automated,
- which problems don’t even require AI,
- which areas are losing money without realizing it,
- and what the first step is that will actually move the needle.
95% of failed projects come from skipping this diagnosis.
2. Adoption is becoming MANDATORY— and no one is talking about it
When the Internet appeared, you weren’t forced to use it.
It took years to reach mass adoption.
That’s not what’s happening with AI.
AI is being pushed onto every user—whether they want it or not.
The big players are embedding it everywhere:
- WhatsApp offers assistants even if you didn’t ask for them.
- Microsoft integrated Copilot across its entire suite.
- Meta inserts models in automated messages.
- Google added AI to Gmail, Calendar, Drive, and Workspace.
I see it weekly:
- Older adults thinking “Mr. Meta is writing to me.”
- Teams confused because their software changed on its own.
- Companies using AI without a plan simply because “it already comes integrated.”
No technology in history has been imposed this fast.
And that explains why so many companies are starting off on the wrong foot.
3. Why so many AI projects fail (according to MIT and what I see in the field)
MIT identified several patterns among companies that fail to achieve impact.
And they match EXACTLY what I see with my clients:
1. They start with the tool, not the process
They want “a chatbot,” “a model,” “a copilot.”
But they haven’t defined which task, area, or pain point they’re targeting.
2. They focus on what is visible, not what is profitable
Marketing and sales receive almost all the AI budget.
Meanwhile, areas like finance, purchasing, operations, or credit—the ones that actually move money—are left out.
That’s where the real ROI is.
3. Projects stay stuck in pilot phase
Amazing demos get built…
But they never integrate with:
- the ERP,
- the CRM,
- real workflows,
- real teams,
- internal data.
That’s where they die.
4. They try to build everything in-house
And MIT confirms it:
Internal AI projects fail twice as often as those developed with external experts.
4. What the successful 5% do differently
On the right side of the GenAI Divide are companies that understand this:
1. AI is implemented by processes, not by features
The question is not “What tool should I buy?” but:
“Which part of my operation can I redesign to gain speed, precision, or savings?”
2. They work with consultants who understand both business and technology
They are not looking for a pretty chatbot.
They want measurable impact:
- Time
- Cost
- Errors
- Friction
- Real productivity
3. They start small—but in strategic places
One properly automated process is worth more than ten directionless pilots.
4. They scale only when they already have impact data
Not before.
5. Predictions for 2025–2027: What’s coming for large companies in Mexico
Based on what I see in the field and MIT’s trends, here are my predictions:
A) AI will become cheaper… but not simpler
There will be Wix-style tools that “solve the basics.”
But none replace a solution designed around your processes, your data, your systems.
B) Companies that adopt early will take the lead
Even if it’s more expensive today, it’s more profitable in the medium and long term.
Why?
- they gain experience earlier,
- optimize earlier,
- learn earlier,
- scale earlier,
- free up budget earlier.
C) The gap between the 5% and the 95% will widen
MIT confirms it:
- Companies that already achieved ROI are the ones adopting even faster.
- The others will remain stuck in pilots, demos, and frustration.
Conclusion
AI doesn’t transform businesses because it’s trendy.
It transforms them when it’s integrated into processes, data, and real decision-making.
The GenAI Divide is not just an MIT statistic:
It’s a mirror of what’s happening today in Mexican companies.
The next few years will separate those who implement AI strategically…
from those who adopt it out of pressure or simply “to have something to show.”
This blog is the first in a series of analyses where I will explain how to bring AI into real operations—without noise, hype, or confusion, but with technical clarity and business focus.
FAQ – Artificial Intelligence in Business
What is MIT’s GenAI Divide?
The GenAI Divide is a finding from MIT showing that only around 5% of companies achieve real, measurable impact from AI. The remaining 95% fail to get ROI because they implement without strategy, without clear processes, or without integrating AI into internal operations.
Why do so many AI projects fail?
Most fail because companies start by buying tools without diagnosing their processes, focus on visible but low-impact areas, leave projects stuck in pilot phase, or try to build everything in-house without technical or strategic experience.
How can my company get ROI from AI?
ROI appears when AI is integrated into critical business processes: finance, operations, purchasing, credit, logistics, data analysis, or document review. Impact comes from automating repetitive tasks, reducing errors, accelerating operational cycles, and freeing up team time.
Which companies benefit most from AI?
Companies with complex processes and high volume: factories, industrial parks, financial institutions, credit companies, real estate firms, printers, manufacturing, and businesses where operational efficiency directly affects profitability.
Will artificial intelligence be cheaper in the future?
Yes, but not simpler. Generic tools will become accessible, but truly effective solutions—those integrated into internal systems, data, and real workflows—will continue requiring experts who understand both business and technology.
Is it better to adopt AI now or wait for lower prices?
Early adoption usually generates more medium-term profitability: companies that start earlier learn earlier, automate earlier, and gain competitive advantage before their competitors.
