AI vs. B2B SaaS: Trends Reshaping How Companies Pay

Compensation benchmarks rely on a reasonably stable market. For most of the past decade that assumption held. A Series B software company in London or New York paid within a recognizable range for a given role and level, and that range moved a few percent a year.

That is no longer the case. AI is not simply another sector competing for the same engineers. It has become an overlay on the entire market, and it is reshaping the pay curve for companies that would never describe themselves as AI companies at all.

The trends below are what we are seeing across high-growth companies in the US, the UK and Europe. Some of them are cash and benchmarking questions. Most of them eventually become equity questions, because equity is the lever that still moves when cash stops.

AI Is an Overlay on the Whole Market, Not a Sector

Capital has concentrated quickly. AI took 54% of venture dollars in early 2026, up from roughly 40% across 2025. Equity grants have followed. Artificial intelligence (AI) and machine learning (ML) engineer grants rose 31% between 2024 and 2026, roughly three times the rate of the broader workforce, and 64% at the smallest startups, where grants are priced off early-stage valuations.

At the same time, hiring has slowed sharply. January 2026 was the quietest January for startup hiring since 2018, down 65% from the 2022 peak. Average Series D headcount was 131 in 2025, down 29% from the 2023 peak.

Taken together, these two trends explain most of what founders are feeling. Teams are smaller, and each person is more expensive. Fewer hires do not translate into a cheaper plan. It usually produces a more concentrated one, where a small number of offers carry a lot of weight and there is less room to get any of them wrong.

The part that matters most for companies outside AI is that they hire against AI-priced compensation anyway. A B2B SaaS company at Series A recruiting a senior backend engineer competes with an AI-native company at the same stage and with an AI lab several stages later. The benchmark that matters is the one the candidate is actually holding.

The AI Premium Widens as You Go Up the Ladder

AI-focused engineers out-earn their peers at every level, but the premium is not flat. In 2025 total compensation data, the premium runs:

• Entry: +6.2%
• Engineer: +11.9%
• Senior: +14.2%
• Staff: +18.7%

In absolute terms, an AI-focused software engineer averages around USD 326K in the US and GBP 184K in the UK. UK and EU pay follows the same shape at lower absolute levels.

The direction of travel matters more than the levels themselves. The junior premium is already compressing, from around 11% at entry a year ago to around 6% now, while the senior and staff premium is holding. In other words, the durable premium sits exactly where hiring is hardest and where a mis-hire is most expensive.

Traditional Benchmarking Is Breaking Down

The harder problem is that the market median is becoming a less useful concept.

In an AI study across one venture fund’s portfolio, a single engineering department spanned three distinct populations. Niche AI specialists with scarce, lab-adjacent skills sat at the top of the range at around GBP 350K. Strong AI adopters, fluent with the tooling and generating high leverage, sat in the mid-band at around GBP 150K. Everyone else, solid but undifferentiated, formed the foundation of the range at around GBP 100K.

That is a GBP 100K to GBP 350K range inside one department, under one job family, at one company. A blended median would overpay the bottom of that range and lose the top of it. Job architecture still matters for structure, but the pay decision increasingly depends on which population a person belongs to rather than which box the org chart puts them in.

New Roles the Benchmarks Do Not Have Yet

AI is also creating roles that did not exist 18 months ago, and they are priced like the scarce skills they require.

Forward deployed engineers sit inside the customer and drive adoption and implementation. If a client uses only 40% of the product, value is being left on the table, and companies are increasingly valuing the person who closes that gap.

GTM and RevOps engineers are RevOps on steroids: still a revenue role, but with far more technical leverage, building the AI-native go-to-market engine. It is one of the fastest-rising titles we see across portfolios.

AI product managers now price like AI labs rather than like software companies. In one case, a candidate’s competing offer escalated from roughly GBP 410K to roughly GBP 1.5M in a matter of weeks. Senior AI product packages in the GBP 0.9M to GBP 1.5M range are no longer anomalies.

New roles create two problems at once. There is no reliable benchmark, so pricing is set candidate by candidate, and the resulting packages land inside a band structure that was never designed for them, which creates internal fairness issues that surface long after the offer is signed.

Comp Levers No Benchmark Has a Column For

Alongside the new roles, companies are inventing pay mechanics.

Token allowances are emerging as a genuine compensation line. For top AI talent, token budgets run from half a salary to a full salary on top of base, reaching GBP 370K to GBP 745K, or USD 500K to USD 1M, per year. The open question is whether this spreads to non-AI roles as tooling cost becomes part of how any knowledge worker does the job.

Companies are also exploring a move from role-based to skill-based compensation. The question is shifting from “What is your role to?” to “What can you do?” Job families still anchor benchmarking, but premiums attach to specific skills rather than to the title.

The premium for top contributors is widening alongside it. Companies are differentiating sharply between top-tier and standard contributors, with roughly a four to five times pay premium at the top and headcount concentrating at the high end.

Furthest out are early experiments tying pay to units of work, to value delivered, or to return on investment as agents takes on more of the execution. These are immature, and defining a 10x contributor objectively is legally and practically difficult. They are still worth watching, because they are the first serious attempt to price output rather than time.

Leadership Cash Is Plateauing, So Ownership Becomes the Lever

One of the clearest patterns in the data is that leadership cash compensation is flattening across stages. A Series A company now competes on cash with a Series D for the same VP. The curve barely moves.

That cannot be solved with cash. If a Series A cannot outbid a Series D on base, the only lever that scales with stage is ownership, and the case has to be made properly. The outcome of joining a GBP 40M company and joining a GBP 800M company is very different. The smaller company has a much higher multiple ahead of it, and that is the argument to use.

Senior hires should be treated like investors. Give them the full picture, including dilution, the terms attached to their grant, and what has to happen for the equity to be worth something. Vague upside is not a competitive offer in a plateaued cash market. This is the single biggest reason to get equity strategy right early: for the roles that matter most, ownership is the only lever that scales with the stage you are at.

Equity Pools: AI Runs Larger Than SaaS, and the US Runs Larger Than Europe

Pool sizes reflect all of the above. As a percentage of total equity, allocated and unallocated, employee pools currently sit roughly as follows.

Two points are worth drawing out. Pool sizes are going up for most AI companies, and the companies that held their pools in line with SaaS benchmarks have generally had to cut their headcount projections to make the numbers work. 

Vesting Is Moving from Time to Performance

New-hire grants have not moved much. Four-year vesting with a one-year cliff remains the standard in both public and private companies.

Refreshers and executive structures are where the movement is. Later-stage companies are shortening refresher vesting to two to four years to conserve equity, and grants are becoming more frequent but smaller. 

At executive level, the shift is toward performance. Performance-based executive refresh grants rose from 9% to 20% between 2023 and 2026. 

Performance vesting only holds with board and remuneration committee support, because someone has to set the metrics and then defend them. That is the practical constraint on this trend. The structure is not the difficult part; the governance is.

Conclusion: What These Trends Mean for How You Pay

Very little about this market rewards making compensation decisions at the point of hire.

Three implications follow from the trends above. First, benchmarking has to reflect the fight a company is actually in. A generic median misleads when a single department spans GBP 100K to GBP 350K, and the relevant comparison is the specific talent pool a candidate belongs to rather than the job family they sit in.

Second, when cash cannot win, ownership has to. That requires equity to be well designed and clearly explained rather than simply generous. A plateaued leadership cash market makes this unavoidable rather than optional.

Third, the decisions that cost the most are the ones made earliest: the first grant, the first leveling framework, the pool size the hiring plan depends on. Those decisions compound. Made early with good data, they keep working all the way through to exit. Made late, they get expensive.

AI has not changed the underlying discipline. It has raised the cost of getting it wrong and shortened the time before you find out.

Stats above are based on 2026 Optio Incentives client data from approximately 100 B2B SaaS and AI companies across the U.S. and Europe, covering relevant job roles and locations.

For more resources, visit the Private Company Stock Plans section on NASPP.com.