This week, leaders in the AI world gathered in Davos, Switzerland, for the World Economic Forum, taking turns offering their best guesses about what the next phase of AI will mean for jobs, whether the AI bubble is real, and when it will burst.
In a 30-minute roundtable discussion, NVIDIA CEO Jensen Huang told BlackRock CEO Larry Fink that there are no bubbles, noting that while past bubbles contained markets, spending on AI (albeit seemingly large) is spreading across nearly every industry.
“One of the good tests for the AI bubble is to realize that Nvidia now has millions of Nvidia GPUs in every cloud. We’re everywhere, but trying to rent an Nvidia GPU these days is incredibly difficult. Spot prices for rentals are going up, not just the latest generation, but GPUs from two generations ago,” he said in an interview.
“The reason for that is the number of AI companies coming up and changing their R&D budgets.[Pharmaceutical giant]Lilly is a good example. Three years ago, most of their R&D budget was probably wet labs. Look at the big AI supercomputers they invested in. Big AI labs? R&D budgets are going to shift more and more to AI. So you’re going to have an AI bubble because there’s a lot of investment in AI. We need more land, more power, more shells because we need to build the necessary infrastructure for every layer of the world. This is the biggest infrastructure build in human history.”
Microsoft CEO Satya Nadella also said that AI is pervasive across industries and economies, and that he doesn’t see a bubble.
“I think a clear sign of whether it’s a bubble is if all we’re talking about is technology companies. If all we’re talking about is what’s happening on the technology side, it’s purely a supply-side issue,” he said. “Ultimately, if we’re not talking about drugs that come to market and are hugely successful because AI accelerated clinical trials. That’s actually happening, by the way. That’s why I’m more confident.”
He said the success of AI and the willingness of users to deploy it will largely depend on whether it can generate as much surplus as predicted by forecasters.
“There is only global demand if there is local surplus,” Nadella said. “If these tokens don’t improve health outcomes, educational outcomes, public sector efficiency, private sector competitiveness across all sectors, large and small, then I think the social license to take something like energy, which is a scarce resource, and use it to generate tokens will quickly disappear. That to me is the ultimate goal.”
One difference between Mr. Nadella and Mr. Huang concerns work.
Forrester’s latest AI Job Replacement Study estimates that through robotic process automation, business process automation, physical robotics, and generative AI, this technology could eliminate 6% of jobs by 2030, totaling approximately 10.4 million jobs.
In a more alarming but perhaps less well-sourced report, a small staff member of the U.S. Senate Health, Education, Labor, and Pensions Committee (HELP) warned that artificial intelligence and automation could put up to 97 million U.S. jobs at risk over the next decade. Officials compiled the report by reviewing economic and business data and then asking ChatGPT to analyze federal job descriptions and estimate which jobs are most likely to be replaced.
When asked about potential job losses caused by AI, Huang preferred to point to the “artisans” such as plumbers, electricians and construction workers needed to build data centers and the infrastructure within them.
“Energy is creating jobs. The chip industry is creating jobs. The infrastructure layer, the power of the land, the shells are creating jobs. So jobs, jobs, jobs, it’s unbelievable. This is the biggest infrastructure build in human history, and it’s going to create a lot of jobs. And it’s great that those jobs are associated with trade crafts…The people who are building chip factories or computer factories or AI factories have six-figure salaries.”
In a separate interview, Nadella acknowledged that these jobs were created through one-time capital investment, but said they should be separated from discussions about the eventual proliferation of AI, which will bring surpluses to other areas of human life.
“This is a technology built on cloud and mobile rails that will spread faster, bend the productivity curve, and deliver local surpluses and economic growth around the world,” he said. “It’s not just the economic growth that comes from capital spending. This is a narrow point in time calculation.”
Nadella said that for AI to be successful for humans, it must have learnable skills that can improve human livelihoods.
“When I was a kid, there was a real connection between learning Excel and Word skills and getting a job,” he said. “We need to bring that back. People need to know, ‘If I learn this AI skill, I can better deliver some product or service in the real economy.'”
But Alex Karp, co-founder and CEO of Palantir, another AI leader, told Fink that he expects labor and technology trade to be the future of a stable job market for the foreseeable future.
“If you went to an elite school and studied philosophy, take me as an example, and hopefully you have some other skills,” he said. “It’s going to be difficult to market that job. But it’s like an engineer. If you’re a vocational engineer, your job will be more valuable. There will be more than enough jobs for your people, especially those who have vocational training.”
One of the people who sounded the alarm about AI at Davos was Salesforce CEO Marc Benioff. There’s no one who supports this technology more vocally than Mr. Benioff. Benioff was one of the first SaaS companies to announce the introduction of AI agents into its stack.
Salesforce also signed deals with Google Gemini and OpenAI to bring these models to the platform as the brains of Agentforce, giving users access to OpenAI as the control plane for their Salesforce tasks.
“By bringing together the world’s leading Frontier AI and the world’s number one AI CRM, we are building a trusted foundation for companies to become Agentic Enterprises,” Benioff said in October.
But at Davos on Tuesday, Benioff White-knuckle clutched his pearls as he outlined some of the failings of AI and a report that claims chatbots played a role in encouraging self-harm and leading to suicide in children, and the need for government regulation.
“I can’t imagine anything worse than this,” Benioff told CNBC. “We can’t just grow at all costs. We need some kind of regulation. Everyone rides on a big language model. We all know these things aren’t that accurate, they often hallucinate. They lie. They don’t know what’s going on. You don’t really understand what’s going on. They can turn around very quickly and it could involve your child. There’s no regulation in the US and it’s the worst situation in the world.” ®
