I watch about a thousand companies make hiring decisions in real time as the CEO of Paraform, the world’s largest recruiting marketplace. The pattern I’ve seen over the last 18 months is so consistent that it’s no longer just a trend.
The standard story about AI and jobs is that automation is hollowing out entry-level work, but that’s not entirely true. What I’m seeing is more consequential: Artificial intelligence is making “good enough” workers redundant, while the remaining demand is concentrating around a much smaller group of exceptional people, and the gap between those two groups is widening faster than most leaders realize.
Our data shows the top 12% of candidates we surface now capture more than 25% of all offers extended, and the top 10% of engineers we place earn roughly three times what the bottom 10% earn for nominally the same title. Both gaps have widened sharply over the past year. Companies aren’t hiring fewer people because they’ve run out of work, but instead because they’ve decided the marginal good-enough hire isn’t worth what they used to pay for it, and the exceptional hire is worth substantially more.
Two years ago, fewer than 4% of engineering roles offered $300,000 or more at the top of their salary band. This year, that share is over 21%. One in five engineering roles is now competing at a level that, until recently, was reserved for the rarest hires. “Staff engineer” and “member of technical staff” postings have grown faster than any other engineering title. Roles paying $400,000 or more, which barely existed two years ago, now appear weekly.
A founder I work closely with cut his head count targets by 60% last quarter after seeing four engineers using AI produce more output than a team of 10 had the year before. He used the budget from those eliminated roles to increase compensation for the engineers he retained and the senior talent he still wanted to hire. More leaders are starting to rethink team composition this way.
The bottleneck has not moved up the org chart the way you’d expect. Mid-level roles are being posted much less than staff or principal openings. Companies have stopped hiring the good-enough worker and shifted resources toward senior talent, evaluating more candidates and saying yes to fewer of them.
This isn’t just engineering. Anthropic is hiring an enterprise copywriting lead with 10 years of experience and a salary band of $225K to $320K, roughly three to four times the average copywriter’s annual salary, rather than a team of mid-level writers. Similarly, Meta has reportedly offered individual AI researchers signing bonuses of $100 million and packages topping $300 million over four years, in some cases hundreds of times what a peer-level engineer at the same company would earn. The math is the same in each case: When one exceptional person can move the trajectory of the company, the marginal premium for getting them is effectively uncapped.
The strategic mistake I am watching leaders make is treating this as a head count problem when it’s a discernment problem. Cutting your team only works if the people who remain are genuinely exceptional. Concentrating spend on fewer hires only works if you can identify the right ones with confidence. Most companies still evaluate candidates against a job description, which is reasonable in a stable environment and insufficient when the role itself is changing faster than the description can. The right question is whether a candidate changes what the team is capable of. Almost no interview process or recruiting tool today is designed to answer that.
Most hiring infrastructure was never built for this environment. Internal recruiting teams sized for steady-state hiring, agencies working within narrow networks, and automated outreach tools built around scale all came out of a market where talent was more evenly distributed and the gap between a good hire and a great one was relatively small. In that world, volume worked. Now, a single exceptional person can meaningfully change a company’s trajectory, while a huge percentage of candidates compete for a shrinking share of offers. Hiring advantage increasingly comes from the ability to consistently recognize and secure candidates with outsize impact.
The retention side compounds this. When responsibility is concentrated in fewer people, losing one of them creates a hole that is genuinely hard to fill. A few years ago, strong operators sometimes stayed in roles longer than made sense for them, out of inertia or loyalty. That dynamic has weakened. Options for exceptional people have expanded faster than most companies’ ability to keep them, and the cost of a wrong hire, or a missed one, has never been higher.
The “10x engineer” used to be Silicon Valley folklore, but today it’s the baseline assumption for any role with real leverage: engineers, operators, marketers, designers, product leaders. The gap between people who can work at that level and people who can’t is widening into a chasm, and the people on the right side of it have more leverage than I’ve ever seen.
Every leader I talk to understands that the stakes have changed, but most have not yet changed how they hire. Over the next few years, hiring quality and talent density will become a durable competitive advantage, especially for companies that approach recruiting with the same rigor they apply to product development or capital allocation. Meanwhile, competition for exceptional talent will keep intensifying, compensation will continue rising, and smaller high-performing teams will keep outpacing much larger organizations.
Talent has always been the constraint. AI just made it the only one that matters.