How Expensive is AI Talent?

Pave Data Lab
March 15, 2024
1
min read

I was recently told that Sundar Pichai, CEO of Google and self-made billionaire, has been taking 1:1 recruiting phone calls with AI & ML software engineer candidates in their mid-20s who have competitive offers from OpenAI that tip the scale at nearly $1M annually. Meanwhile, Walter Isaacson’s biography on Elon Musk talks about Musk throwing around $1M+ sign-on bonus packages for top AI & ML candidates he was trying to poach to join xAI.

This being said, at a recent dinner with Total Rewards leaders, I was told that while there are certainly “outlandish” comp stories from the outliers…the rest of the market for AI talent is “high but not crazy”. Especially outside of the Bay Area Bubble.

What are we seeing in the compensation market data? Have million-dollar packages become the 'new normal,' or are such cases merely outliers and part of the mythology that fuels the AI hype cycle?

ASSUMPTIONS
  • Assumptions: P4 (senior) and P5 (staff) levels, USA, $200-500M Capital Raised.
  • Analysis: Let’s look at the software engineering generalist vs machine learning (ML) job families. ML is the best job family proxy we currently have for AI skills.
  • Caveat: Some TR leaders are viewing AI as a new job family while others are viewing it as a skill set (with a comp premium) within the SWE job family. Meanwhile, other companies are going even deeper with specialty job families such as “model optimization”. For the sake of this analysis, we will take a look at “ML vs SWE”, and we will do a deeper skills-related analysis soon based on your feedback.

RESULTS

In this market data analysis, Machine Learning ICs are indeed paid a bit higher than Software Engineering Generalists.

However, the “comp premium” for ML (as the best available proxy for AI) appears to be somewhat limited (~7-8%) at this point in time across the broader market. This suggests that the hype cycle stories of the “$1M comp packages” are from a vocal minority of extreme outliers rather than the “new normal”.

P4 (Senior) Median Benchmarks:

1️⃣ Machine Learning => $210,000 | High Confidence (n = 51-100)

2️⃣ Software Engineering Generalist => $196,000 | Very High Confidence (n = 1,001+)

P5 (Staff) Median Benchmarks:

1️⃣ Machine Learning => $244,612 | Confident (n = 11-30)

2️⃣ Software Engineering Generalist => Very High Confidence (n = 1,001+)

Learn more about Pave’s end-to-end compensation platform
Matthew Schulman
CEO & Founder
Matt Schulman is CEO and founder of Pave, the complete platform for Total Rewards professionals. Prior to Pave, he was a software engineer at Facebook focusing on user-centric mobile experiences. A self-proclaimed "comp nerd," Matt is known for sharing data-driven thought leadership around all things compensation and personal finance.

Become a compensation expert with the latest insights powered by Pave.

(function (h, o, t, j, a, r) { h.hj = h.hj || function () { (h.hj.q = h.hj.q || []).push(arguments) }; h._hjSettings = { hjid: 2412860, hjsv: 6 }; a = o.getElementsByTagName('head')[0]; r = o.createElement('script'); r.async = 1; r.src = t + h._hjSettings.hjid + j + h._hjSettings.hjsv; a.appendChild(r); })(window, document, 'https://static.hotjar.com/c/hotjar-', '.js?sv='); !function () { var analytics = window.analytics = window.analytics || []; if (!analytics.initialize) if (analytics.invoked) window.console && console.error && console.error("Segment snippet included twice."); else { analytics.invoked = !0; analytics.methods = ["trackSubmit", "trackClick", "trackLink", "trackForm", "pageview", "identify", "reset", "group", "track", "ready", "alias", "debug", "page", "once", "off", "on", "addSourceMiddleware", "addIntegrationMiddleware", "setAnonymousId", "addDestinationMiddleware"]; analytics.factory = function (e) { return function () { var t = Array.prototype.slice.call(arguments); t.unshift(e); analytics.push(t); return analytics } }; for (var e = 0; e < analytics.methods.length; e++) { var key = analytics.methods[e]; analytics[key] = analytics.factory(key) } analytics.load = function (key, e) { var t = document.createElement("script"); t.type = "text/javascript"; t.async = !0; t.src = "https://cdn.segment.com/analytics.js/v1/" + key + "/analytics.min.js"; var n = document.getElementsByTagName("script")[0]; n.parentNode.insertBefore(t, n); analytics._loadOptions = e }; analytics.SNIPPET_VERSION = "4.13.1"; analytics.load("0KGQyN5tZ344emH53H3kxq9XcOO1bKKw"); analytics.page(); } }(); $(document).ready(function () { $('[data-analytics]').on('click', function (e) { var properties var event = $(this).attr('data-analytics') $.each(this.attributes, function (_, attribute) { if (attribute.name.startsWith('data-property-')) { if (!properties) properties = {} var property = attribute.name.split('data-property-')[1] properties[property] = attribute.value } }) analytics.track(event, properties) }) }); var isMobile = /iPhone|iPad|iPod|Android/i.test(navigator.userAgent); if (isMobile) { var dropdown = document.querySelectorAll('.navbar__dropdown'); for (var i = 0; i < dropdown.length; i++) { dropdown[i].addEventListener('click', function(e) { e.stopPropagation(); this.classList.toggle('w--open'); }); } }