Super Intelligence Engineer Salary: The Real Numbers

Job boards haven't minted a new title overnight. What exists are two separate pay realities wearing the same search term.

The first is the broad AI engineer market: tens of thousands of roles at normal companies, with pay that looks like any senior engineering track plus an AI premium.

The second is a tiny set of frontier labs, Meta Superintelligence Labs (MSL), Safe Superintelligence Inc. (SSI), and Microsoft's MAI Superintelligence Team, that already used "superintelligence" as a brand before the government rename and pay on an entirely different scale.

Anyone selling you a single "super intelligence engineer salary" number is flattening two markets that don't resemble each other at all.

Start with the market that will actually hire you.

What Glassdoor shows

Glassdoor's AI Engineer data, built from over a thousand submitted salaries, puts average total pay at $146,277 a year, with a typical range of $118,000 to $184,000 between the 25th and 75th percentile, and top earners reaching $225,474.

By seniority, Glassdoor shows entry-level averaging $128,013, senior averaging $287,515, and principal averaging $323,278, with top principal earners up to $527,686.

Anyone selling you a single "super intelligence engineer salary" number is flattening two markets that don't resemble each other at all.

What Coursera and RDU's own research show

Coursera's 2026 guide lands in a similar place from a different angle: a median total salary of $138,000, a BLS-sourced annual median of $145,080, and a pay curve by experience that runs from $103,015 at 0 to 1 years up through $121,513, $138,185, $155,008, $172,361, and $185,709 at 15-plus years.

On this site's own AI engineer jobs research, the bands run junior $95,000 to $130,000, mid-level $130,000 to $180,000, senior $180,000 to $260,000, and staff or principal $250,000 to $400,000, which lines up closely with Glassdoor and Coursera once you account for base-only versus total-comp reporting.

This is the real range for the overwhelming majority of people who will ever hold an AI or "super intelligence" job title. The frontier-lab numbers below are the exception, not the baseline.

MSL is the one frontier lab with enough public postings to show real numbers.

Posted base salaries

Pay-transparency-mandated listings on metacareers.com show posted base salaries of $219,000 to $301,000 for an AI Research Scientist on the Product and Applied Research team, $154,003 to $217,000 for an AI Research Engineer on the Media team, and roughly $122,000 to $257,000 for an AI Research Scientist on the Robotics team, all before bonus and equity, which at MSL can roughly double or triple the posted base.

Total comp estimates

Total comp estimates fill that gap.

Compensation site ctaio.dev puts standard MSL bands at $525,000 to $700,000 for E5 (Senior), $750,000 to $1,000,000 for E6 (Staff), and $1,000,000 to $1,500,000 for E7 (Principal), describing these as a 10 to 20 percent premium over equivalent core-Meta engineering levels.

Levels.fyi data for the "AI Researcher" title at Meta shows a wider spread: $386,000 at E4 up to $1.26 million at E6, with a reported median between roughly $414,500 and $454,000 depending on how the page's figures are sliced.

The related "AI Engineer" title at Meta runs lower, $359,000 at E4 to $645,000 at E6, a useful reminder that titles inside the same lab carry very different pay.

SSI and Microsoft's superintelligence effort are harder to pin down, by design.

SSI, founded by Ilya Sutskever, Daniel Gross, and Daniel Levy, hit a $30 billion valuation in a March 2025 round led by Greenoaks Capital, six times its $5 billion valuation six months earlier, without generating revenue.

Public data points are thin: reported compensation for senior researchers exceeds $1 million annually, and one Member of Technical Staff listing showed a base range of $100,000 to $500,000.

Microsoft's MAI Superintelligence Team, formed in November 2025 under Mustafa Suleyman and pursuing what he calls "Humanist Superintelligence," hasn't published superintelligence-specific compensation bands as of this writing.

The team is still young, having only shipped its first in-house model family at Microsoft Build 2026, and Microsoft's standard AI and machine learning engineering pay elsewhere in the company is the closest public proxy until bands for this specific group surface.

The number driving most "super intelligence salary" searches isn't a salary at all.

Wired reporter Zoe Schiffer revealed that Mark Zuckerberg personally offered more than 10 OpenAI staffers packages up to $300 million over four years, with more than $100 million in year-one compensation.

Subsequent reporting named researcher Matt Deitke's deal at roughly $250 million over four years, and described other offers in the $200 million to $500 million range, with one package reportedly reaching $1 billion before the researcher turned it down.

OpenAI's own CEO, Sam Altman, confirmed the scale publicly, describing it this way:

“$100 million signing bonuses, more than that in compensation per year.”

— Sam Altman, OpenAI CEO

Meta's own CTO, Andrew Bosworth, has pushed back on treating these as representative. He told employees the nine-figure offers went to a very small number of people in senior leadership roles.

For context, Levels.fyi data puts Meta's E7 engineers at an average of about $1.54 million a year, and most engineers never reach E7.

These packages go to a handful of named researchers with years of public frontier-AI reputation behind them, recruited directly by executives, not posted, applied to, or negotiated through a normal hiring process.

If a course or job board is pricing "super intelligence engineer salary" using these numbers, it's selling you a fantasy.

Here's every verified range from this guide in one place, with the source behind each number.

Four things separate a $140,000 AI engineer from a $1 million-plus frontier researcher, and none of them are the word "superintelligence" on a job posting.

• Research pedigree. Publications and hands-on experience training foundation models open the research-scientist tier, which pays multiples of applied engineering work.
• Reputation. SSI and top-end MSL recruit people they already know, not applicants, so visible public work replaces a cold application.
• Equity structure. Comp is stock and bonus heavy, so headline numbers assume continued vesting and company performance, not guaranteed cash.
• Competing offers. The $100 million-plus packages exist because one named person held simultaneous Meta, OpenAI, and Anthropic offers, leverage almost nobody else has.

If you're earning inside the typical $95,000 to $260,000 AI engineer range and want to move toward frontier-lab pay, the applied and research-engineer seats are the realistic bridge, not the research-scientist titles that dominate headlines.

• Specialize. Pick what frontier labs hire for right now, in MSL's own posting language: model evaluation, multimodal training pipelines, agent orchestration, data curation at scale.
• Build in public. Don't wait for a posting; reputation is how SSI and senior MSL actually recruit.
• Negotiate from real numbers. Use this guide's ranges to ask for what your level actually pays, not the $300 million headline a news story sold you.

For the hiring side of this, the super intelligence jobs guide covers who's hiring and how to search, the broader AI engineer jobs guide covers the market that will actually hire most developers reading this, and the super intelligence engineer hub ties both together with what the role requires day to day.

If you want the wider picture beyond AI specifically, the highest-paying software engineer jobs list and the developer salary negotiation guide cover how to push your number up no matter which track you're on.