The narrative around artificial intelligence and work has followed a predictable arc: AI takes over tasks, humans lose jobs, and companies save money. It made intuitive sense. It also turns out to be more complicated than anyone anticipated.
The real story emerging in 2026 is messier and more interesting. Some of the world’s largest technology companies are pulling back on AI spending after discovering that the economics don’t pencil out the way they expected. Bryan Catanzaro, Nvidia’s VP of deep learning, put it with unusual candor: “The cost of compute is far beyond the costs of the employees.”
Uber’s chief technology officer said his company burned through its entire 2026 AI coding budget by April after employee leaderboards incentivized maximum token use. Microsoft canceled most of its direct Claude Code licenses months after encouraging mass adoption. One company reportedly spent $500 million on Claude usage in a single month.
These aren’t isolated anecdotes. They represent something larger: a first reckoning with what it costs to replace human capability with computation at scale. And the fallout is pointing toward a conclusion I’ve been tracking in my workforce research for over a decade. The things that make people irreplaceable aren’t disappearing in the AI era. They’re appreciating.
I call it the Human Premium. And I think it’s one of the most important career concepts of our time.
The Economics Are Shifting Under the Narrative
Despite enormous capital investment, including $740 billion in announced capex from Big Tech this year alone, no widespread data is showing AI has driven significant productivity gains across the economy. An MIT study found that AI automation would be economically viable in only 23% of roles where visual work is primary. In the remaining 77% of cases, it was simply cheaper to keep humans doing the work.
That calculation will change over time. Gartner projects that the cost of inference for large language models will fall by more than 90% over the next four years, and McKinsey estimates AI expenditures could reach $5.2 trillion by 2030. At some point, the math tips. But we are not there yet, and the companies that assumed we already were are now recalibrating.
What’s more revealing is what the recalibration is exposing. The most AI-forward companies aren’t retreating from AI entirely. They’re retreating from the idea that AI can simply substitute for people. The ones getting the best returns are using AI to amplify human expertise rather than replace it. According to PwC’s 2026 Global AI Jobs Barometer, an analysis of more than a billion job postings across six continents, the top 20% of AI-exposed companies achieved average labor productivity growth of 163% relative to 2018, nearly five times the average among AI-exposed organizations overall.
The difference between those companies and the ones burning through token budgets without results is what they’re doing with the humans in the room.
Human Connection Is Becoming a Luxury Good
I’ve been thinking about a broader shift that the AI cost crisis is accelerating, and it’s one that goes beyond workforce economics. The more digital the world becomes, the higher the premium people will pay for the experience of a real human interaction.
Think about what’s already happening at the consumer level. We’ve automated customer service, phone trees, chat support, and retail checkout. Efficiency went up. Satisfaction, in many cases, went down. The human on the other end of the line, the one who could read the situation and exercise judgment and empathize with your problem, has become rarer. And as a result, they’ve become more valuable.
Human connection is shifting from a standard expectation to a luxury good. The financial advisor who calls instead of sending an algorithm. The doctor who sits with you instead of routing you to a portal. The manager who has the hard conversation instead of sending the form email. These interactions now carry a premium precisely because they’re scarce. And scarcity, as any economist will tell you, creates value.
In my view, this is one of the most underappreciated career dynamics of the next decade. As AI handles more of the volume, speed, and scale of work, the humans who can navigate ambiguity, build trust, and show up with genuine presence will not just survive the AI era. They’ll command a premium within it.
The Labor Market Is Already Pricing This In
The data is beginning to catch up to the intuition. PwC’s Barometer identified two categories of jobs emerging in the AI economy. The first are what they call “professionalized” roles: jobs where AI handles routine tasks but still relies on human expertise and judgment, such as radiologists and recruiters. The second are “democratized” roles: jobs where AI enables less experienced workers to perform tasks previously requiring more seniority.
Professionalized roles have seen job growth at twice the rate of democratized ones. Salaries in professionalized occupations have risen 42% faster. And demand for AI skills overall carries a 62% wage premium, with some sectors seeing premiums as high as 118%. Jobs requiring AI skills have grown 69% since 2019, nearly eight times faster than overall job market growth.
But here is the part that matters most for understanding where the real career leverage lives: The fastest-growing entry-level roles are now seven times as likely to require skills traditionally associated with senior professionals, including leadership, creativity, and interpersonal communication. Employers aren’t just looking for people who can use AI tools. They’re looking for people who bring what AI can’t.
The World Economic Forum’s Future of Jobs Report found that 39% of workers’ core skills are expected to change by 2030, with analytical thinking, leadership, resilience, and creative thinking among the fastest-growing capabilities employers are seeking. These aren’t technical skills. They’re human ones.
What the Human Premium Actually Looks Like
Over 15 years of workplace research, I’ve watched the same pattern repeat across technological transitions. The workers who thrive aren’t necessarily the early adopters of every new tool. They’re the ones who stay clear about what they bring that the tool cannot.
In the AI era, that means a few things specifically: Judgment—the ability to interpret AI outputs, challenge them when they’re wrong, and make decisions that account for context a model doesn’t have—is becoming one of the most valuable professional competencies in existence. Empathy and relationship-building, long dismissed as soft skills, are now measurably differentiating. Trust, the kind that accumulates through consistent human interaction over time, cannot be generated through prompts.
The CEO pullback on AI spending isn’t just a cost story. It’s a signal about what organizations are discovering works and what doesn’t. What works is AI plus people who know how to use it with judgment. What doesn’t work is AI as a substitute for the human capability that was already there.
The irony of the moment is that the more companies try to automate everything, the more clearly the things that can’t be automated stand out. We are in the process of running a very expensive global experiment in what happens when you remove human judgment, human connection, and human empathy from work at scale. The results are coming back, and they’re telling us something important: People are not the cost to be optimized away. They’re often the entire point.
The Human Premium isn’t nostalgia for the way things were. It’s a thesis about where value is heading. Every major economic transition in history has changed what society rewards most. The industrial age rewarded physical capability. The information age rewarded knowledge. The AI era, I believe, will reward something older and harder to replicate: the distinctly human ability to connect, earn trust, exercise judgment, and lead other people through uncertainty. That is a premium no algorithm can undercut, and its value is only going to rise.