“The key question is, what problems will a trillion-dollar AI solve?” said Jim Covelo, head of research at Goldman Sachs. “Many people like to compare today’s AI to the early days of the internet. But even in the early days of the internet, e-commerce was a low-cost technology solution that could replace expensive legacy solutions.”
Generative AI Faces Profitability Challenges
Given the high barriers to entry and the high cost of key components like GPUs, there is skepticism that AI costs will ever come down enough to automate most tasks. “The market is overly optimistic about the potential for cost reductions,” Covelló added.
MIT professor Darren Acemoglu is also skeptical. Acemoglu estimates that “only a quarter of the jobs at risk from AI could be cost-effectively automated over the next decade,” meaning that AI would affect less than 5% of all activities. He also points out that it is not yet known whether AI models will become cheaper over time.
“It’s still unclear how long investors will be satisfied with the ‘build it and they will come’ mantra,” says Covelo, adding that he expects investor enthusiasm to wane if significant use cases don’t materialize in the next 12 to 18 months.
But more importantly, the company’s profitability. As long as the company’s profits remain solid, AI experiments will continue. Therefore, given the positive economic development in the U.S., Covelo does not expect companies to reduce their spending on AI infrastructure and strategies.
Wrong comparison with the Internet or mobile
Senior equity analyst Eric Sheridan is more bullish on the current situation. “Anyone who argues that the current euphoria is irrational is only looking at the amount being spent today compared to the two previous investment cycles: the expansion of network infrastructure that enabled the development of Web 1.0 and desktop computing in the late 1990s and early 2000s, and the mobile expansion from 2006 to 2012, including the introduction of mobile spectrum, 5G network equipment, and smartphones,” Sheridan writes.
It’s like comparing apples to oranges. “A more appropriate metric is the ratio of investment to revenue generated,” Sheridan said. “Currently, cloud computing companies are spending more than 30% of their cloud revenue on CAPEX, and most of that is going to AI initiatives. For the IT industry as a whole, that’s not much different from previous investment cycles that have driven changes in IT usage by businesses and consumers.”
“All IT cycles follow a sequence called IPA: infrastructure, platform, applications,” analyst Kashi Rangan added. “The AI cycle is still in the infrastructure phase, so it will take time to find the killer application, but I believe it will get there eventually.”
editor@itworld.co.kr
Source: www.itworld.co.kr
