One thing is that AI pessimists tell us that there is a bubble. Another very different is that Sam Altman suggested, CEO of Openai. But it is what has happened, and that is a worrying indication of the situation in which this segment is located. More and more expert voices warn of the danger of an AI bubble, but there are not only voices: there are data that raise a potential crisis. One that could be even more harmful than that of the Puntocom.
What has Altman said. The head of OpenAI, the company that develops ChatgPT, invited a group of journalists to comment on the launch of GPT-5. During that meeting, they indicate in The Verge, he said the following:
“When bubbles occur, intelligent people are excessively excited about (which is actually so alone) a pinch of truth.
Are we in a phase in which investors in general are too excited about AI? My opinion is that yes.
Is AI the most important thing that has happened in a long time? My opinion is also that “
Remembering the story. Altman compared the current dynamics that were experienced during the Bubble of the Puntocom in the late 2000s. Between March 2000 and October 2002, the Nasdaq lost about 80% of its value: many of the companies that pointed to the Internet fever and the web failed to generate income or benefits.
The value of the 10 most important companies of the S&P 500 index is today much greater than the one in the 90s, and that points to a potential (and huge) bubble. Source: Apollo Global Management / Tornsten Slok.
Worse than the bubble of the Puntocom. Economic analysts and experts have long offered arguments that point in the direction of a potential bubble of AI. The chief economist of the investment firm Apollo Global Management, Torsten Sløk, indicated in a report that this bubble could be worse than that of the Puntocom: the 10 most important companies on the S&P 500 index have a value well above the 10 that occupied those positions in the late 90s.
Too much speculation. Ray Wang, director of the Futurum Group semiconductor firm, showed two faces of the same currency. As commented on CNBC,
“From the perspective of a broader investment in AI and semiconductors … I do not see it as a bubble. The foundations of the entire supply chain remain solid, and the long -term trajectory of the trend of AI supports the continuous investment”
But at the same time, he stressed that there is a problem with this segment: there is too much speculative investment in companies that have less solid bases and in which there is only one perception of its potential without real foundations – Hello, Safe Superintelligence, Hello Thinking Machines.
It’s hard, but bubbles have their good side. As Alberto Romero points out in his Newsletter, “in a way, bubbles are an inevitable and welcome phase between short -term selfishness and long -term progress.” In his opinion and that of other experts such as Mills Baker, manager in substock:
“He hype It is acceptable under the premise that only an optimistic character, prone to exaggeration and hyperbole, can build the new world for which a bubble is only the starting point, his big bang. The cynical and pessimistic character is a useful counterweight to excessive optimism (…). While optimism is an active creation force, pessimism is a reactive modulation force. “
Source: Paul Kedrosky.
When the trains were the AI. Or what is the same: for the world to advance, bubbles are (or can be) necessary. We saw it with the Puntocom: it is true that the crisis existed, but that uncounted optimism in the future of the Internet ended up making sense. Of course, only a few companies (the great current empires) ended up benefiting. But something very similar to the railroads happened at the end of the 19th century. At that time the investment and the capex in these infrastructure was colossal – five times greater than what is now lived in AI – and although many companies broke, but from that bubble we came out with an absolute revolution both at the level of transport and economic and social.
But this bubble can be very, very large. As Romero points out, the difference here between optimistic speeches and pessimists (or realistic) is growing, and that is worrying. The expectations that are creating the companies of AI and their CEOs (with Altman in front, the man-hype) are increasing. They constantly tell us how we are close to reaching the AGI, but the reality is that there are no real indications that this is so and in fact there is a slowdown of AI. Faced with the promises of the revolution that theoretically should have begun to generate AI, the reality is that the advances do not seem extraordinary. In fact, a MIT study discovered that after asking 150 entrepreneurs and 350 employees of companies that have integrated AI in their processes, 95% had not seen any benefit in doing so.
Better Wait for GPT-6. It has been demonstrated by GPT-5, a model for which we expected a historical leap and in the end raises an improvement that is now discreet and that introduced changes that were highly criticized. The launch of this model has been a small disaster that has forced the company to reverse several of its decisions. As Walter Bloomberg pointed out, Altman himself admitted that GPT-5 had been a failure, and now bets on GPT-6.
Taste: Michael A. Arouuet.
The data worries. The seemingly excessive expense in data centers or talent is not the only concern. There is also that absolute concentration of companies that concentrate the value. An analyst named Michael A. Aouet published these days a graph in which he showed two income growth trends. On the one hand, that of the 490 companies of the S&P 500 excluding the 10 most important. On the other, that of those 10 companies. Until Chatgpt was launched, both lines went to the time, the growth was very modest, but later, the 10 most important companies – among them, “the seven magnificent” – fired their income and their assessment in the stock market.
Fuente: Bloomberg.
Too much value concentration. At the same conclusion, Bloomberg analysts arrived, who also included a worrying graphic (in the image). It shows the normalization of all the values of the actions to show how the evolution of the S&P 500 index is seen if the largest companies in the world will not concentrate the entire value. In that vision the trend is clear. Chatgpt was launched in November 2022, just where the lead fall of that trend begins, and as they say in Bloomberg, “it is unheard of 2% of the index companies represent practically 40% of the value (of the S&P 500).”
If the bubble explodes, few will survive. Thus, we are certainly worrying at a time. The one that unfortunate optimism and that belief of investors in expectations do not seem to have a brake. Robin Li, CEO of Baidu, already indicated months ago that the bubble exists and believes that it will end up exploding. And if he does, he said, only 1% of companies will survive and dominate a market that, yes, will end up creating enormous value. As the railroad did. Or internet.
image | Techcnch | WorldOfSoftware with freepik picture
In WorldOfSoftware | Rumors about the huge goal data centers point to a problem: the Big Tech are playing everything to AI
