Jensen Huang didn’t say “prefer the one who uses AI.”

He said “hire the one who uses AI.”

At GTC 2026, the CEO of a $3 trillion company was explicit: between two candidates with the same credentials, the one who masters AI gets the job. The other doesn’t. No caveats.

And he wasn’t just talking about software engineers.

Accountant. Hire the one who uses AI. Lawyer. Hire the one who uses AI. Marketing. Supply chain. Sales. Customer service. Function by function, the same answer.

Then he said something that should concern anyone who still thinks AI is a Silicon Valley tool: “If you’re a carpenter, if you’re an electrician, go use AI. If I were a farmer, I would absolutely use AI.”

A farmer who uses AI to optimize soil, predict weather, and manage yields isn’t competing with other farmers. They’re operating at a level that used to take an entire department.

That’s already happening.

The math of the team changed

For decades, productivity scaled roughly linearly with headcount. Ten people produced about ten times what one could. There were variations, but the curve was predictable.

AI broke that curve.

It’s not that one person with AI works faster. It’s that they can operate across disciplines that used to require separate specialists. A marketing manager who masters AI tools can now run competitive analysis, write content, build metric dashboards, and model pricing scenarios — tasks that used to fill four distinct roles.

McKinsey documents 20-60% productivity gains with agentic AI. But that range only tells half the story. Speed is the obvious gain. The bigger one is breadth — a person with AI mastery starts doing work across disciplines they never could have touched before.

And they don’t need anyone’s permission to cross departmental boundaries. AI doesn’t understand org charts.

AI agents extend the multiplier

If you think the 100x employee is an exaggeration, look at what Anthropic just shipped.

Claude can take over your Mac desktop remotely. You send a command from your phone, and the AI starts clicking, typing, and operating applications on your computer to complete the task. This isn’t a text assistant that answers questions — it’s an agent that executes complete workflows on your machine.

Four weeks before that launch, Anthropic acquired Vercept, a startup specializing in autonomous agents. That acquisition-to-launch timeline — four weeks — should tell you how fast the autonomous agent stack is maturing.

And it’s not just Anthropic. Open-source tools like Understand Anything turn entire codebases into interactive knowledge graphs — a new developer can map out a codebase in hours that used to take weeks of reading.

Jensen also named farmers, electricians, carpenters. Not office roles. People who build things with their hands. An electrician who uses AI to model electrical loads, simulate wiring, and quote jobs in seconds doesn’t compete with other electricians. They compete with firms.

One person with command of the model replaces the output of a team without it.

Your next job posting is written wrong

If your last open role asked for “5+ years of experience in digital marketing” with no mention of AI, you’re hiring for a world that already moved.

Look at the role from an output perspective. Before AI, you needed those 5 years because experience was a proxy for productivity. Someone with 5 years knows where to find data, knows the tools, has intuition about what works. All of that took time to build.

With AI, much of that accumulated knowledge is accessible in seconds. Not all of it — business intuition, judgment for prioritization, customer knowledge remain irreplaceable. But the technical tasks that took years to master now accelerate dramatically with the right tool.

So the hiring math flips. If one AI-literate professional produces what five used to, you don’t need five hires. You need one exceptional one — and a budget for AI tools.

Jensen mentioned a revealing number at GTC 2026: for a $500,000-per-year engineer, he expects a minimum $250,000 annual investment in AI tools. We wrote about what that means for your specialization strategy. The relevant point here is different: the tools budget per person can be nearly as large as the salary.

If your hiring process doesn’t evaluate AI mastery, you’re filtering on the wrong criteria.

Training vs. hiring: the decision most companies skip

Jensen said every college student should graduate as an AI expert. Not familiar with it. Not aware of it. Expert.

But universities haven’t caught up. They still train students to execute the work. The market already moved — it wants the person who directs the machine that executes.

That means talent with real AI mastery is scarce. And if the market doesn’t produce it in sufficient volume, you have two options: compete with everyone for the few who exist, or build the capability inside your current team.

The second option usually wins.

Your current employees already know your business. They understand your processes, know your customers, see where the real problems are. What they lack is the tool. Giving them that tool — and the skills to use it — delivers returns faster than finding someone from the outside who masters AI but needs to learn everything else. The numbers back this up:

Factor	Hiring new talent	Training current team
Time to productivity	3-6 months (onboarding + business context)	4-8 weeks (already know the business)
Upfront cost	$80K-$150K+ (competitive salary in a tight market)	$5K-$15K per person (programs + tools)
Knowledge retention	High risk (competitive market, turnover)	Distributed across existing team
Compounding advantage	Individual	Organizational

AI training doesn’t mean sending your team to a 4-hour ChatGPT workshop. It means building the ability to direct AI agents through real business workflows — from research to execution, through analysis and validation.

Casually using ChatGPT to summarize emails is like using a calculator. Directing AI agents through multi-step business processes is like writing software. Both are useful. One of them rewires what a single person can deliver.

This doesn’t apply to every role

If you read our previous post on when AI is the wrong tool, you know that not every problem calls for AI. The same applies to people: not every role multiplies equally with AI tools. Roles with high ambiguity, multiple disciplines, and constant decisions benefit the most. Roles with fixed processes and heavy regulation have a lower ceiling.

The point isn’t to replace your entire team with one professional and an AI subscription. It’s to understand that the equation of how many people you need — and what kind of people you’re looking for — has changed.

---

Jensen wasn’t giving career advice to students. He was describing a hiring policy for the entire world.

If you’re not sure how much of your team is ready to operate with AI — or how wide the gap is — that’s the first question we answer in our AI maturity assessment.

Assess my team’s AI readiness