Portuguese 
English 
Spanish 
3 min of reading 
0 comments 
High Vertiginous Sparks Alerts
On September 10, 2025, Oracle’s stock rose 35.95%, adding about US$ 244 billion to the company’s market value.
This jump represented the largest daily valuation increase since 1992, according to Times Brazil.
Despite the euphoria, the increase raised concerns among investors wary of speculative bubbles.
This is because the gains were driven by the growing demand for cloud services linked to artificial intelligence.
For this reason, the scenario resembles the excesses of the dot-com bubble in the early 2000s.
Megamodels Impress, but Are Economically Fragile
Currently, systems like ChatGPT and Gemini dominate the headlines with impressive advancements.
However, they consume billions in infrastructure and show clear signs of diminishing returns.
Meanwhile, less flashy solutions demonstrate concrete efficiency across various areas.
For example, in Austin, Texas, a local AI system optimized the construction licensing process.
The process that used to take months is now resolved in just a few days, with a significant practical impact.
Despite the lack of fanfare, productivity gains have been solid and sustainable.
Moreover, companies in the healthcare sector have been using specific models for medical diagnostics.
These models outperform generic LLMs in accuracy and adaptability.
In the financial sector, BloombergGPT provides more reliable results than conventional platforms.
Thus, the return on investment becomes more tangible and less dependent on technology popularity.
-
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
Local Models Offer Efficiency, Low Cost, and Greater Security
Smaller solutions, tailored for specific tasks, show clear advantages over generalist megamodels.
In addition to reducing costs, they ensure greater speed and accuracy in delivered results.
While LLMs spend billions to train and operate, local models require far fewer resources.
Therefore, various companies have adopted on-premise or edge systems, reducing reliance on remote servers.
This has made operations faster, cheaper, and much more secure.
By keeping data close to the operation, companies gain control and protection over sensitive information.
Thus, this approach reduces regulatory and privacy risks.
At WebAI, for example, engineers managed to reduce models to one-third of their original size.
Even with this compression, the accuracy of the results was maintained.
This way, operational costs fell drastically while autonomy increased.
Investors Should Avoid Euphoria and Focus on Applied Intelligence
History offers important parallels.
Netscape led the internet boom in the 90s but was ultimately swallowed by infrastructure solutions.
In the same way, AI megamodels may lose ground to simpler and more efficient technologies.
Despite being impressive, platforms like ChatGPT provide more entertainment than direct business value.
In other words, they work well for the public but have limited utility in corporate processes.
The launch of GPT-5 illustrates this transition.
Even with technical advances, the market reacted coolly, as scalability is no longer a novelty.
Thus, the search for practical solutions replaces the fascination with gigantic systems.
Executives should avoid investments in technologies that prioritize spectacle.
Instead, the focus should be on solutions that deliver measurable and sustainable results.
Intelligence Will Not Disappear, It Will Transform with Strategic Focus
Even if the AI bubble bursts, intelligence will continue to evolve.
The difference is that it will follow more discreet and efficient paths.
Companies investing in specific solutions will not suffer from speculative crashes.
They are already reaping concrete benefits, away from the headlines and hype cycles.
Just as the internet survived the dot-com bubble, AI will also move forward.
However, the model will be different: less centralized, leaner, adapted to the reality of each sector.
In other words, the future will not be about grandiose AI at any cost, but about practical intelligence with staying power.
To business leaders, the recommendation is clear: think small, execute with focus, and build real value with AI.
In an uncertain scenario, what remains is efficiency.
-
Uma pessoa reagiu a isso.