The 2026 Engineering Compensation Report
Where engineering pay actually moved this year, and which markets closed the gap on the Bay Area.
By Daniel Okonjo, Senior Analyst, Work Platforms
The median stalled; the tails did not
Median software engineering compensation grew 6.2% year on year, roughly inflation and unremarkable. That number conceals a widening distribution. Machine learning engineering compensation grew 34.6% at the same time, and UX research fell 2.4%.
The industry is not paying more for engineering. It is paying dramatically more for a narrow band of scarce specialisation and slightly less for everything adjacent to it.
Geographic compression is real but overstated
Remote hiring was supposed to flatten geography. It compressed it. San Francisco still carries a 32% premium over the national band for software engineering, down from 41% at peak, while Austin and Toronto gained ground.
The compression came from companies benchmarking nationally rather than locally, which raised offers in mid-tier US markets and did comparatively little for international ones. London remains at roughly 68% of the US band; Berlin at 54%.
Equity became the negotiation
Base salary bands moved little, but equity treatment changed substantially. Refresh grants, extended exercise windows and secondary liquidity are now routine negotiating points rather than exceptions.
At senior levels in AI-adjacent roles, equity is the dominant component of total compensation, which makes comparing offers between a public company and a late-stage private one a genuinely difficult exercise rather than an arithmetic one.
What this means for hiring plans
Companies staffing generalist engineering have more leverage than at any point in five years. Companies staffing machine learning have essentially none, and are competing against employers whose equity is worth more.
The practical response most organisations have landed on is structural: hire generalists locally, hire specialists globally, and stop pretending one compensation philosophy covers both.
Frequently asked questions
- Which engineering role pays the most in 2026?
- Machine learning engineering, with a median of $202,000 base at national US benchmarks and staff-level bands reaching $342,000 before equity.
- Has remote work eliminated location-based pay differences?
- No. It compressed US domestic differences considerably while leaving international gaps largely intact. San Francisco still carries a 32% premium over the national median.