A study on trillion-dollar bets on AI shows how difficult it is to assess their return on investment. Jessica Wachter, a finance professor at the University of Pennsylvania, started with an undisputed fact: several hyperscalers are investing heavily in building data centers. Instead of making predictions about the scale of model adoption, she posed the question differently – how quickly must these companies' profits grow to justify expenses by 2027, when they are expected to reach nearly $1.1 trillion. The results are sobering: to break even by 2030, AI companies will need exceptional productivity growth.
Separately, it became known that the OpenAI Foundation, the nonprofit parent organization of OpenAI, will fund the idea of analyst Ruxandra Teslo. AI lacks biological data for breakthroughs in disease treatment, and Teslo proposed using information from bankrupt biotech companies: participating in bankruptcy proceedings allows access to regulatory documents, production strategies, and safety data. She calls this the 'lost archive of biotechnology.' The foundation will fund the creation of 'high-quality scientific datasets.'
In the latest issue of MIT Technology Review, the threats of AI extinction, a new 'blood rejuvenation' drug from the startup Generation Lab, and other topics were also discussed. Among the day's tech news: Nvidia and Meta heads Jensen Huang and Mark Zuckerberg rejected calls for coordinated slowing of AI development, with Zuckerberg stating that competition will itself push labs towards safety; the FTC head warned against granting antitrust exemptions to AI companies; a Chinese hacker firm used AI to analyze stolen data; in cancer research, 'smart' nanoparticles delivered mRNA to tumors and reprogrammed cells to attack in mice; a digital brain of a fruit fly, replicating 166,000 neurons, was used to control a car, trade Bitcoin, and play Doom; the AI agents platform iLand announced sending 1.6 million messages; ByteDance founder Zhang Yiming's fortune exceeded $105 billion, making him the richest person in Asia; and a fully AI-generated sitcom was described by a critic as 'lifeless wax figures.'
Related articles
