The leaders of the AI world descended on Davos, Switzerland, this week for the World Economic Forum, where they took turns lobbing their best guesses about what the next phase of AI would mean for jobs, as well as whether the AI bubble was real and when it may pop.
In a 30-minute sit-down, Nvidia CEO Jensen Huang told Blackrock CEO Larry Fink that there is no bubble, and pointed out that previous bubbles had encapsulated their markets, whereas AI spending – while it seems big – cuts across nearly every vertical.
“One good test on the AI bubble is to recognize that Nvidia has now millions of Nvidia GPUs in every cloud. We’re everywhere and if you try to rent an Nvidia GPU these days, it's so incredibly hard. The spot price of rentals is going up. Not just the latest generation, but two-generation-old GPUs,” he said during the interview.
“The reason for that are the number of AI companies that are being created, the number of companies that are shifting their R&D budget. (Pharmaceutical giant) Lilly is a great example. Three years ago most of their R&D budget was probably wet labs. Notice the big AI supercomputer they’ve invested in? The big AI lab? Increasingly that R&D budget is going to shift to AI. So the AI bubble comes about because the investments are large. The investments are large because we have to build the infrastructure necessary for all of the layers of AI above it. We need more energy. We need more land, power, and shell. We need more trade-skilled workers. This is the single largest infrastructure buildout in human history. Get involved.”
Microsoft CEO Satya Nadella also said he does not see a bubble, saying that AI has become diffused across industries and economies.
“I think a telltale sign of if it's a bubble would be if all we’re talking about were the tech firms. If all we’re talking about is what’s happening to the technology side, then it's just purely supply side,” he said. “Ultimately if we are not talking about a drug that was brought into the market that was super successful because AI accelerated the clinical trial. By the way, that's happening. This is why I’m much more confident.”
He said the success of AI and the willingness of users to adopt it depend greatly on whether it is capable of producing the surplus that prognosticators have forecast.
“Demand all over the world will only be there if there is a local surplus,” Nadella said. “I think we will quickly lose the social permission to take something like energy, which is a scarce resource, and use it to generate these tokens, if these tokens are not improving health outcomes, education outcomes, public sector efficiency, private sector competitiveness across all sectors, small and large. That to me is ultimately the goal.”
One point where Nadella and Huang differed is around jobs.
Forrester’s most recent AI job replacement research estimates that the technology could uproot 6 percent of jobs by 2030, or about 10.4 million total, through robotic process automation, business process automation, physical robotics and generative AI.
In a more alarming but perhaps not as well-sourced report, minority staff for the US Senate Committee on Health, Education, Labor, and Pensions (HELP) warned that artificial intelligence and automation could put up to 97 million American jobs at risk over the next decade. Staff compiled the report by reviewing economic and corporate data, then asking ChatGPT to analyze federal job descriptions and estimate which occupations are most vulnerable to replacement.
When asked about potential job losses caused by AI, Huang preferred to point to the “tradecrafts” such as plumbers, electricians, and construction workers needed to build out the datacenters and infrastructure inside them.
“Energy is creating jobs. The chips industry is creating jobs. The infrastructure layer, land power, and shell is creating jobs. I mean jobs, jobs, jobs, it's incredible. This is the largest infrastructure buildout in human history and that's going to create a lot of jobs. And it's wonderful that the jobs are related to tradecraft … We’re talking about six-figure salaries for people who are building chip factories or computer factories or AI factories.”
Nadella in a separate interview acknowledged those jobs created by a one-time capital expenditure, but he said they must be separated from the discussion about AI’s eventual diffusion, bringing a surplus to other areas of human life.
“This is a technology that will build on the rails of cloud and mobile, diffuse faster, and bend the productivity curve, and bring local surplus and economic growth all around the world,” he said. “Not just economic growth driven by capital expenses. That’s a narrow point-in-time calculation.”
Nadella said for AI to be a success for humans, it must come with masterable skills that can make people better at earning a living.
“Growing up, there used to be a real relationship between learning Excel skills or Word skills and getting a job,” he said. “That needs to come back. People need to know ‘If I pick up this AI skill, then now I’m a better provider of some product or service in the real economy.’ ”
- AI and automation could erase 10.4 million US roles by 2030
- AI robs jobs from recent college grads, but isn't hurting wages, Stanford study says
- AI may be everywhere, but it's nowhere in recent productivity statistics
- AI layoffs to backfire: Half quietly rehired at lower pay
However, another AI leader, Palantir cofounder and CEO Alex Karp, told Fink that he expects labor and technical trades will be the future of the steady job market for the time being.
“If you went to an elite school and you studied philosophy – use myself as an example – hopefully you have some other skill,” he said. “That one is going to be hard to market. But like technicians. If you’re a vocational technician … those jobs are going to become more valuable. There will be more than enough jobs for the citizens of your nation. Especially those with vocational training.”
One person at Davos sounding a warning on AI was Salesforce CEO Marc Benioff. No one has cheered more loudly for the technology than Benioff, whose company was among the first SaaS enterprises to announce it had deployed AI agents into its stack.
Salesforce has also struck deals with Google Gemini and OpenAI to bring those models into its platform as the brains behind its Agentforce, as well as letting users access OpenAI as a control plane for Salesforce tasks.
“By uniting the world’s leading frontier AI with the world’s #1 AI CRM, we’re creating the trusted foundation for companies to become Agentic Enterprises,” Benioff said in October.
Yet at Davos on Tuesday, Benioff white-knuckle clutched his pearls when describing the need for government regulation as he described the several failure points for AI and a report alleging a chatbot encouraged self-harm and played a role in the lead-up to a child’s suicide.
“I can’t imagine anything worse than that,” Benioff told CNBC. “It can’t be just growth at any cost. There has to be some regulation. Everyone is on a large language model. Everyone knows these things are not that accurate, that they hallucinate a lot. They lie. They don’t really understand what’s going on. They can make a right turn very fast and then when it involves your children. That’s a big deal. In the US we have zero regulation and we fully indemnify all the tech companies. It's kind of the worst of all worlds.” ®