Payscale's September 2026 AI Workforce Impact preview found 58% of employers already pay or plan to pay a premium for AI skills, while 49% admit their salary structures haven't kept pace. The people most likely to fall through that gap are the employees HR just paid to train.
Key Takeaways
Most employers now agree AI skills are worth paying for. The trouble is who ends up collecting. When a company can't find the talent it needs, it pays the market rate to bring someone in from outside. When its own people spend evenings and weekends learning the same tools, the reward is often a line added to their job description. The result is a two-track pay system nobody designed on purpose, and the employees on the slower track can see exactly how it works.
Payscale's AI Workforce Impact Report preview, released September 24, found 58% of employers already pay or plan to pay a premium for AI skills, 19% are holding their pay structures steady despite new AI requirements, and 13% haven't settled on a strategy. The demand side is not in doubt: citing Lightcast data, Payscale notes AI job postings have surged 8.7 times in under five years, 61% of employers are rewriting job descriptions because of AI, and 41% say they can't find talent with the AI skills they need.
The pricing side is where it falls apart. 49% of employers say their salary structures haven't kept pace with AI, and 48% say their current market benchmarking no longer reflects the skills their roles actually require. Meanwhile 23% now treat AI fluency as a baseline expectation, with no additional pay attached, even as 56% of employees believe they should be paid more for it. "AI skill demand is outpacing employers' ability to price it," said Ruth Thomas, chief compensation strategist at Payscale. "The result is inconsistent pay decisions at a moment when employees are investing in AI skills and expecting a return."
That inconsistency was already visible at the start of the year. Payscale's 2026 Compensation Best Practices Report, based on 3,413 responses, found 61% of organizations had updated existing roles to include AI skills or competencies, but 55% were not adjusting compensation for them. Where a reward did exist, it was thin: 14% offered higher base pay for AI proficiency, 10% offered bonuses, and 9% offered long-term incentives. Only 10% had added AI competencies to leadership roles at all.
Put those numbers side by side and the risk comes into focus. HR Dive's reporting on the Payscale findings describes a potential "retention time bomb": current employees who upskill on AI watching new hires command premiums of 20% to 40% that they aren't receiving. Employers are, in effect, paying the market price to buy AI skills while expecting to get the same skills free from people already on payroll. 74% plan AI upskilling investments over the next 12 months, which means more employees than ever are about to acquire a skill their employer is actively paying a premium for somewhere else.
Employees have already drawn their own conclusion. A Marsh report covered by HR Dive found more than four in 10 workers now believe they are more likely to get a raise by leaving and rejoining their company than by staying, up from 18% in 2024. Just 31% said they expect to be compensated for upskilling. "A depleted workforce paired with cynicism about pay creates a precarious moment for employers," said Gordon Frost, global rewards leader at Marsh. "Employees are exhausted, they don't believe the system is fair, and the moment the job market improves, they will leave."
The upside of getting it right is just as clear in Marsh's data. Workers who feel fairly paid report being 85% more engaged and 60% more committed. Nearly three-quarters of HR leaders say skills-based pay drives retention, yet only 26% have actually implemented it. That is the gap in one sentence: most HR teams believe the fix works, and most haven't built it.
The math makes this harder than it sounds. A separate Marsh compensation planning survey of more than 1,000 companies projects 2027 merit increases of 3.2% and total salary increases of 3.5%, including promotions and cost-of-living adjustments. A merit pool that size can't close a double-digit gap between an upskilled incumbent and an external hire, and spreading it evenly guarantees it won't. The same survey found 64% of companies provided or planned off-cycle salary adjustments for 2027, which suggests many already know the annual cycle won't be enough on its own.
That leaves three real options for a scarce skill: build it internally and pay for it, buy it externally at market rates, or widen the search beyond the domestic labor market where the premium is highest. Most companies will use all three. What they can't do is pay the premium for the second option while telling employees the first is its own reward. The employees doing the learning will notice the moment a new hire's offer letter circulates, and in a transparency environment, it will.
Guide
When an upskilled employee asks why a new hire with the same AI skills earns more, this is how to explain the decision honestly before pay cynicism turns into a resignation letter.
Download
Guide
Only 10% of organizations have added AI competencies to leadership roles. If the leaders pricing AI skills can't judge them, that is a leadership development gap worth fixing first.
Download
Toolkit
With 41% of employers unable to find the AI talent they need at home, widening the search across borders is one of the three real options. This toolkit covers how to do it compliantly.
Download
87% of employers say they can defend their pay decisions, but only 20% have ever tested that claim.
57% of organizations now post salary ranges while only 51.4% have a formal job architecture underneath them.
Salary increase budgets are settling at 3.4 to 3.5% for 2027, and employers are abandoning across-the-board raises.
