The Trimodal Nature of Tech Compensation in 2026

This article is inspired by Gergely Orosz's Pragmatic Engineer trimodal analysis (updated March 2025). The data here is synthetic but calibrated to levels.fyi public compensation records and the levels.fyi 2025 year-end report. Refreshed July 2026.

The Salary Gap is Real

Two software engineers with the same title, same years of experience, and similar technical skills can earn wildly different amounts — sometimes 4x apart. This isn’t noise or negotiation luck. It’s structural.

What does “structural” look like in data? Plot total compensation for thousands of software engineers and you don’t see a single bell curve. You see three overlapping distributions — a trimodal pattern that reflects three separate labor markets operating side by side. And as of 2026, a fourth, smaller spike has appeared above all of them.

The left cluster (blue-gray) peaks around $100-140K. The middle cluster (orange) peaks around $200-260K. The green mass centers near $340K with a long tail. And the violet spike — frontier AI labs — starts where Big Tech tops out and runs past $1M. Same job title. Same country. Completely different compensation realities.

So what are these markets, and how do you tell which one an employer belongs to?

The Three Tiers (Plus One)

Following Gergely Orosz’s taxonomy, we can classify the tech labor market into three tiers, defined less by prestige than by who a company benchmarks its pay against. That single question — “whose offers does this company try to match?” — sorts almost every employer cleanly:

Tier 1 — Traditional / Local Tech

Companies that compete for local talent: IT departments inside non-tech companies, outsourcing firms, consultancies, and banks. Compensation is almost entirely base salary. Note the surprise entries — levels.fyi data shows bank software engineering roles land here (Goldman Sachs SWE median is roughly $170K, far below its front-office pay), and Orosz’s list of top Tier-1 payers includes names you’d guess were “tech tech”: Indeed, HubSpot, Workday, Capital One, eBay.

Examples: Accenture, Infosys, IBM, Goldman Sachs, JPMorgan Chase, Capital One, Visa, Disney, Workday, Indeed.

Tier 2 — Competitive Tech

Recognized tech brands that pay significantly above the local market and offer meaningful equity, but that don’t quite benchmark against Google and Meta. They compete for talent with Tier 3 and lose their best people to it.

Examples: Adobe, Salesforce, Atlassian, Shopify, Spotify, Uber, Airbnb, LinkedIn, Snap, MongoDB.

Tier 3 — Big Tech + Top Startups + Quant

The classic top of the market: companies with massive revenue per employee that compete globally for engineering talent. Equity often exceeds base at senior levels; quant firms pay the same totals mostly in cash. Tier boundaries blur here — levels.fyi has Netflix at roughly $460K median and Databricks near $490K, well above the Google all-level median of ~$295K.

Examples: Google, Meta, Apple, Amazon, Microsoft, Netflix, NVIDIA, Stripe, Databricks, Jane Street, Citadel, Two Sigma.

Tier 4 — Frontier AI Labs (the 2026 spike)

This is the new development. Orosz’s framework is still officially trimodal, but 2026 compensation benchmarks increasingly describe a de facto fourth mode: frontier AI labs paying 2-3x strong Big Tech medians. levels.fyi puts OpenAI’s SWE median around $800K; Anthropic lands somewhere in the $420-600K range depending on which population you count. Equity — often private, infrequently liquid — makes up more than half of these packages, reaching 60-70% at the biggest labs.

Examples: OpenAI, Anthropic, Google DeepMind, xAI, Thinking Machines Lab, Mistral AI, Cohere.

The median Tier 3 engineer earns $339K — roughly 2.6x the median Tier 1 engineer at $132K. Tier 2 at $253K is nearly double Tier 1. And the AI-lab spike at $579K median sits another 70% above Tier 3. These aren’t small differences.

What Creates the Gap: Equity

What mechanism produces a 2.6x gap between markets for the same work? Not base salary — bases differ by maybe 40%. The driver is equity compensation. At Tier 1, equity is essentially nonexistent. At Tier 4, it’s the majority of the package.

This is the key structural insight: Tier 1 companies pay salaries. Tier 3 and 4 companies pay ownership stakes. When the stock appreciates, the gap widens further; when it drops, upper-tier engineers feel it more acutely.

Compensation by Level

Does the gap stay constant as you climb? No — it widens dramatically at senior levels. An entry-level engineer at Tier 3 makes roughly 2.4x a Tier 1 counterpart. At staff level the multiple is over 3x — and against Tier 4 it’s 4x.

Why does the gap widen? Because equity grants at the top scale superlinearly with level. A senior engineer at a Tier 3 company might get $200K/year in RSUs; a staff engineer $300-400K. Meanwhile, the Tier 1 equivalent gets a modest raise in base salary. Same mechanism as before — the tiers that pay in ownership can scale ownership; the tiers that pay in salary cannot.

The market data confirms the widening is accelerating: in the levels.fyi 2025 year-end report, Staff-level pay rose +7.5% year over year while Entry-level rose just +1.6%. The top of the curve is pulling away from the bottom even before you count the AI labs.

US vs Canada

Is this an American artifact, or does the market structure travel? Canada is the cleanest comparison: same companies, same roles, different country. The answer: the same tiers appear, compressed downward — still multimodal, just with lower absolute numbers across the board.

Takeaway: the tier structure survives the border — what changes is the exchange rate on it, roughly a 25% discount everywhere except the AI labs.

The discount is roughly 24-26% for the three classic tiers, aligning with exchange rate and cost-of-living adjustments. The AI-lab tier is the exception: frontier labs benchmark globally rather than locally, so the Canada discount shrinks — a Toronto offer from a frontier lab (or homegrown Cohere) is much closer to US parity than a Toronto offer from a bank. Which market you’re in even determines how much your country matters.

Interactive: Salary Tier Explorer

Use the controls below to explore the distribution by country and experience level (it opens at Senior, the most representative view). Hover over bars to see exact counts per bin — and click any tier card to open it up and see which companies make up that slice, with each company’s median for the selected level.

Try clicking Tier 3 — you’ll see the quant firms (Citadel, Jane Street) and top scaleups (Databricks) sitting above the Big Tech names inside the same tier. At Staff level the tiers separate almost completely; at Entry level they overlap far more.

What Changed in 2025-26

Why does the trimodal picture need a 2026 refresh at all? Because four shifts changed the shape of the curve, not just its height:

1. The AI talent war created the fourth spike. Frontier labs now pay 2-3x Big Tech medians, with a 56% wage premium for AI skills (up from 25% a year earlier). The headlines are wilder still — Sam Altman publicly claimed Meta dangled “$100M signing bonuses” at OpenAI staff, and one reported Meta package totaled ~$1.5B over six years. Read the fine print: these are heavily back-weighted with retention vests and performance gates, not guaranteed cash.

2. The top is pulling away from the bottom. Staff +7.5% vs Entry +1.6% year over year (levels.fyi 2025). The curve is stretching vertically even within tiers.

3. The entry-level squeeze is real. Entry-level postings fell from 8.1% to 7.4% of the tech job mix while senior postings grew, “junior” roles increasingly ask for 2-5 years of experience, and roughly 148,000 tech workers were displaced in 2026 with AI cited in about a quarter of recent layoffs. Getting into the funnel is now the hard part.

4. Geographic arbitrage is eroding. Return-to-office roles grew +12% year over year, pulling pay premiums back toward hub cities. London still pays about 70% of San Francisco; Bangalore about 25%.

What Tier Am I In?

If you’re wondering where you fall:

• Tier 1 signals: Your company’s core business isn’t technology. Your compensation is almost entirely base salary. Your total comp is within the ranges you see on general job boards.
• Tier 2 signals: Your company is a recognized tech brand. You receive meaningful equity (RSUs or options). Your comp is notably above generic salary surveys, but you know peers at Big Tech are earning more.
• Tier 3 signals: Your company is a household tech name, a top scaleup, or a quant firm. Equity (or quant cash bonus) is a major portion of your comp. Your total package would surprise most people outside tech.
• Tier 4 signals: You work at a frontier AI lab. Most of your compensation is private equity whose value depends on the next funding round. Your offer letter made you re-read it.

The tier you’re in matters more than your negotiation skills. Moving between tiers is the single highest-leverage career move in tech compensation — far more impactful than switching teams, getting a promotion within the same tier, or negotiating a signing bonus.

Key Takeaways

References