“We have never seen the tobacco industry highlight on its own initiative the dangers of smoking. There are no known airlines that have run communication campaigns saying that their planes could crash. How then can we explain that some of the most important producers of artificial languages have adopted the complaint from a former employee of OpenAi and Anthropic, who resigned declaring in a tweet that everyone in the industry knows that artificial intelligence could kill us all by the end of the decade? Why didn't anyone in the industry point out that the tweet didn't mention no evidence of that disconcerting statement and many have even claimed that we must slow down language research and development?”. The latest episode of the podcast “On the 4th floor" in which Alessandro Fugnoli, strategist at Kairos Partners Sgr, analyses the latest debate on the safety of Artificial Intelligence and its effects on the markets.
AI between catastrophism and politics
Fugnoli starts from a consideration: with the exception of civil nuclear power, no technology has ever been blocked per motivi di sicurezza. Usiamo l’auto anche se ci sono spesso incidenti e costruiamo case, dighe e ponti anche se ogni tanto crollano.
So why was the issue raised now? “It is possible,” observes the strategist, “that the most leveraged and undercapitalized sector of AI, the one who burns cash every day in large quantities sees with concern the approaching of the placing its securities on the market stocks on high valuations and try to organize themselves as an oligopoly, asking for government protection.”
Il tema, insomma, è molto più politico che tecnologico. Organizzarsi in un oligopolio garantisce maggiore protezione dalla concorrenza di nuovi arrivati che in futuro potrebbero affollare il settore, ma anche dai produttori cinesi, “la cui quota di mercato in Occidente è passata dal 2 per cento nel 2024 al 20 per cento alla fine del 2025”, analizza Fugnoli. Né gli uni né gli altri, secondo OpenAi e Anthropic, garantirebbero la sicurezza dei modelli. Ed è a questo punto che arriva la richiesta di “centralize AI production in a few trusted hands that have the concrete support of governments, possibly sealed by public participation in the shareholding and an implicit guarantee of rescue in the event of financial difficulties.” And who are these trusted hands? Always those of the current market leaders.
“Overall, it doesn't seem like the time has come yet to declare the AI boom on its downward spiral,” reassures Fugnoli, who underlines how Donald Trump has no intention of regulating the artificial intelligence sector and that a possible Democratic Congress next year could introduce some form of control, but would still remain in favor of the strong development of the sector, even from an anti-Chinese perspective.
Fugnoli: The stock market rally isn't over yet.
"The sector therefore remains extremely vital. Vitality does not exclude, but rather implies, that internal competition leads to downsizing of the most fragile protagonists and a general decline in profit margins,” continues the economist, according to whom Chinese competition on semiconductors and advanced superchips produced by American hyperscalers could threaten Nvidia's monopoly, also showing that the AI inflation we have seen so far could soon turn into deflation. The benefits, at that point, would go more to end users than to manufacturers.
“The fact that we are not witnessing the final stages of the AI boom also tells us that the stock market rally, also in slight difficulty due to seasonal reasons, it's not sold out. Compared to recent quarters, however, the increase will be less concentrated and better distributed,” concludes Fugnoli.
