Jensen Huang wants you to believe that artificial intelligence is the greatest job creation engine since the Industrial Revolution. During a Monday night conversation with MSNBC at a Milken Institute event, Nvidia’s CEO painted a remarkably rosy picture of AI’s impact on American workers. According to him, we should all stop worrying about displacement and start celebrating the dawn of a new industrial era powered by GPU factories and AI infrastructure.
I find this messaging deeply frustrating, not because it’s entirely wrong, but because it’s selectively true in a way that serves Nvidia’s interests perfectly.
The Hardware Salesman’s Perspective
Let’s be clear about the context here. Huang runs a company that sells the pickaxes in this gold rush. Every AI model that gets trained, every inference cluster that gets built, every “AI factory” that springs up needs Nvidia hardware. Of course he’s going to be optimistic about AI adoption. His quarterly earnings depend on organizations believing that AI implementation is both necessary and beneficial.
The argument he makes about jobs versus tasks is actually interesting though. He claims that just because AI automates a specific task doesn’t mean the entire job disappears. The broader function an employee serves remains intact. This sounds reasonable until you actually think through what happens in practice.
When you automate enough discrete tasks within a role, you don’t need as many people doing that role. Maybe you go from needing five customer service reps to needing three who manage AI systems. Sure, those three people still have jobs. But what happened to the other two?
The Doomer Discourse Problem
Huang’s criticism of “doomer” rhetoric is where things get really interesting. He worries that fearmongering will make Americans so afraid of AI that they won’t engage with it at all. There’s actually something to this concern. Productive technology adoption requires some level of public trust and willingness to experiment.
But here’s the irony that the article points out perfectly: much of that doomer rhetoric came from AI companies themselves. Remember when AI labs were warning about existential risks and comparing their work to nuclear weapons? That wasn’t external critics. That was the industry hyping its own capabilities to an almost absurd degree.
It was marketing disguised as concern. Create the narrative that you’re building something so powerful it could destroy humanity, and suddenly your chatbot sounds a lot more impressive than it actually is. Now that the hype cycle has generated enough investment and attention, the message conveniently shifts to “actually don’t worry, this is all going to create jobs.”
The 15% That Nobody Wants to Talk About
Financial institutions and academic researchers suggest around 15% of US jobs will be eliminated in the coming years due to AI. That’s not some fringe doomer prediction. That’s mainstream economic analysis from reputable organizations.
Fifteen percent isn’t an apocalypse, but it’s not nothing either. That’s tens of millions of people who will need to find new work, retrain, or accept different roles. The idea that all of this displacement will be smoothly absorbed by new AI-adjacent job creation is optimistic at best and deliberately misleading at worst.
I build software for a living. I’ve seen how automation actually plays out in organizations. Yes, new roles emerge. DevOps became a thing. Data engineering became a specialty. But the math doesn’t work out to one-to-one replacement. One senior engineer with good tooling can now do what required a team of five just a decade ago.
The “re-industrialization” argument Huang makes is particularly interesting. He’s right that AI infrastructure requires physical manufacturing and that those facilities need workers. But modern factories are dramatically more automated than their predecessors. The chip fabs and hardware manufacturing plants that power AI employ far fewer people per dollar of output than traditional manufacturing did.
What This Means for People Actually Building Things
If you’re a developer or engineer working with AI systems, Huang’s perspective should feel familiar. It’s the same pitch we’ve been hearing from every technology vendor forever: adopt our platform, it’ll make you more productive, don’t worry about the downstream effects.
The reality is messier. AI tools do make individual developers more productive. GitHub Copilot and similar assistants genuinely speed up certain coding tasks. But companies don’t typically respond to productivity gains by keeping the same headcount and just letting people work less. They respond by expecting more output from fewer people.
This doesn’t mean AI is inherently bad or that we should resist its adoption. But it does mean we should be skeptical when billionaire CEOs of AI hardware companies tell us there’s nothing to worry about. Their incentives are clear, and they’re not aligned with the average worker trying to figure out if their job will exist in five years.
The question isn’t whether AI will create any jobs. Of course it will. The question is whether it will create enough jobs, of sufficient quality, to offset the displacement it causes. And whether the people who lose work will be the same people who gain it, or if we’re creating a new class divide between those who can leverage AI and those who get replaced by it.
Maybe Huang is right and this will all work out fine. But I’m not taking labor market advice from someone whose company just posted record profits selling the infrastructure that enables the very automation we’re discussing.