A New Bubble? From the Dot-Com Era to Artificial Intelligence
A few days ago, Oracle (NYSE: ORCL) —a company where I had the privilege of serving as CEO in Peru— announced the layoff of up to 30,000 employees, delivered through nothing more than an impersonal email. Not because of a crisis: its quarterly revenue reached $17.2 billion, a record. The layoffs, executed en masse and "virtually" at 6 a.m., were driven by a different reason: funding its bet on artificial intelligence infrastructure. Oracle's stock has fallen 25% this year under pressure from investors concerned about the level of debt this bet demands.
A Recurring Pattern
This scenario feels familiar. In the late 1990s, I witnessed the dot-com bubble firsthand, when companies invested fortunes to avoid missing the e-commerce wave that was supposedly going to destroy traditional retail — and it didn't. What actually happened was that e-commerce took longer to mature, and a great deal of money was wasted. More recently, the digital transformation wave — accelerated by the pandemic — repeated the pattern: fear, euphoria, massive investment, and the urgency of not being left behind.
The Numbers No One Wants to Hear
With artificial intelligence, we are repeating the same cycle, but amplified. The numbers are telling: a study by the National Bureau of Economic Research from February of this year found that 90% of companies report no impact whatsoever from AI on their productivity, despite the investments made. Sam Altman himself, CEO of OpenAI — the company behind ChatGPT — has acknowledged that a bubble exists. The IMF has warned it could burst. And analysts identify the 2026–2028 period as the window of greatest risk for a significant correction.
The Cost Paradox
There is a paradox that is seldom mentioned: companies are investing large sums of money at a time when the costs of implementing AI are the highest they will ever be. This is nothing new in technology. The prices of emerging technologies always decline with volume and maturation, and they do so in ever-shorter timeframes. Those who bought e-commerce infrastructure in 2000 paid several times what it cost just five years later. The question is whether the urgency to implement AI today justifies investing at prices that will inevitably drop even faster, or whether a more gradual adoption would allow companies to do so with better aim and lower risk.
The Board's Dilemma
This reflection is not just for the big tech companies. In private companies in our country, the pressure to adopt AI is moving from management to the boardroom, often driven by consultants or vendors with a direct commercial interest. The board, composed of professionals with vast business experience but not necessarily in emerging technologies, faces a delicate dilemma: approve significant investments in something they don't fully understand, or be perceived as a brake on innovation. It is an uncomfortable position that I know well from my experience on boards, in the IT/TELECOM industry, and from my HR practice in executive and board member search.
The Right Question
The question a board member should ask is not "Should we implement AI?" — the answer is probably yes — but rather "Should we do it now, at these costs, with a technology that is still maturing?" Designing a gradual adoption without eliminating proven existing processes, with limited pilots and clear metrics, may be smarter than a massive bet driven by the fear of being left behind.
What History Teaches Us
History teaches us that these bubbles do not destroy the underlying technology. The dot-com bubble did not kill the internet; e-commerce did not replace traditional commerce — it complemented it. AI will not disappear, but capital invested prematurely can be lost, just as it was lost before and to an even greater degree. And Oracle, with its 30,000 layoffs to fund AI data centers, may be showing us the first cracks in a euphoria that deserves to be examined with the same prudence with which we should examine any investment: by looking at history before betting on the future.