Jensen Huang Channels Dwight Schrute… Wants to Pay Engineers in AI Tokens (No, Seriously)

Jensen seems to be taking his own swing on Dwight’s “Shrute Bucks” idea…

The “you’re getting paid in robot money now” era of Silicon Valley Nvidia has officially arrived… And leading the charge is none other than leather-jacketed GPU godfather, Jensen Huang.

At Nvidia’s latest GPU Technology Conference, Huang flippantly threw out what might be the most crazy comp plan since Dwight Scrute tried to convince Stanley to care about “Scrute Bucks” on that one episode of The Office.

Engineers, he says, shouldn’t only get salaries… or crypto… or even stock options. They should get (checks notes in disbelief) AI tokens.

Tokens, more specifically for AI agents.

Translation: instead of just paying you $300K a year to code, Nvidia might hand you another $150K worth of “AI firepower” so you can deploy your own army of digital interns.

So what are these “tokens” exactly?

Jensen wants us to think of them like credits you spend to run AI tools… writing code, analyzing data, drafting reports, (all the things he has his nerdy engineers doing).

The way he sees it, the more tokens you have… the more AI agents you can spin up… and the more productive you become… the more money you (and most importantly) Nvidia makes.

And if Huang’s vision plays out, this doesn’t stop at a few helpful copilots. He’s talking about Nvidia eventually running hundreds of thousands of AI agents alongside its 42,000 human employees. 

Naturally, this is where things go from “cool productivity hack” to “slight existential crisis.” Because while Huang is pitching this as a productivity cheat code… a lot of people are hearing something else: “Congrats on your promotion… you now manage the software that replaces you next quarter… anyways, here are your Chuck E Cheese tokens.”

The numbers backing that fear aren’t exactly comforting. Goldman Sachs estimates AI could automate 25% of all work hours in the U.S. They also see a 15% productivity boost… paired with up to 6-7% job displacement along the way.

Even Howard Marks chimed in, warning that AI’s ability to act autonomously is the difference between a “nice $50B market”… and a full-blown multi-trillion dollar labor shakeup.

So yeah… that’s not scary at all.

But here’s where Huang flips the script. According to him, AI agents won’t actually kill software demand… they’ll supercharge it.

More agents = more code, more infrastructure, more compute… aka more reasons to buy Nvidia chips like they’re the last case of water before a hurricane. But to play devil's advocate, of course he’s gonna think that… because it benefits him BIGLY.

Still, the real world outside of Nvidia HQ looks a little more complicated. Countless Fortune 500 companies are already dealing with what’s being called a “talent paradox”… where executives expect to cut headcount while simultaneously struggling to find people who actually know how to use this stuff.

Entry-level roles are first in line for extinction. The classic “learn by doing the boring work” ladder is getting pulled up… which is great if you’re already experienced, and horrible if you’re just getting started.

And just to keep things grounded… about 80-85% of AI projects have failed since 2018.

So while everyone’s busy hiring digital employees by the thousand… there’s a very real chance half of them end up being expensive, glitchy interns that break things and hallucinate answers.

Which raises a question nobody seems too eager to ask:

If every engineer has an army of AI agents… And every company is deploying thousands of them… Who’s managing the chaos when the robots start stepping on each other’s toes?

At the time of publishing this article, Stocks.News doesn’t hold positions in companies mentioned in the article.