A single engineer just reset the going rate for AI talent. Meta's Superintelligence Labs reportedly offered Andrew Tulloch, a co-founder of Mira Murati's Thinking Machines Lab, an equity package worth up to $1.5 billion over six years. No, that's not a typo. But the billion-dollar headline isn't the part that should worry you.
Tulloch was one of five Thinking Machines founders Meta recruited in a single raid. Meta disputes the figure. A spokesperson called the Wall Street Journal's report of it "inaccurate and ridiculous," and the company notes the number assumes full milestone payouts and years of stock appreciation. The figure is already doing its work in the market anyway.
Sam Altman watched Meta come straight for his OpenAI bench:
"They [Meta] started making these giant offers to a lot of people on our team. $100 million signing bonuses, more than that comp [compensation] per year. It is crazy. I'm really happy that, at least so far, none of our best people have decided to take them up on that."
— Sam Altman, CEO, OpenAI, on the Uncapped podcast, via Entrepreneur
Anthropic's median total compensation, per self-reported Levels.fyi data, runs around $600,000. A senior engineer there earns $316,000 base plus $247,000 in stock, before you get anywhere near the frontier research tier.
This isn't a story about billionaires being weird. It's a story about what happens to your hiring pipeline when compensation at the frontier resets the entire market's frame of reference.
How Extreme Numbers Cascade Down
When you're trying to recruit a mid-career AI/ML engineer with four years of production experience and solid model fine-tuning skills, you're not competing with Meta Superintelligence Labs. That person doesn't have a $1.5 billion offer in hand.
But they do have a number in their head.
They read the news. They talk to peers. They see Anthropic comp in the mid-six figures and wonder what your total package says about how seriously you take AI. When your recruiter calls with a $220K all-in offer (solid by pre-2024 standards), the candidate isn't measuring you against what they made last year. They're measuring you against the market's current story about what AI talent is worth.
The data backs this up. AI/ML compensation commands a steep premium over traditional tech roles. Estimates run from 28% to over 50% depending on the dataset, with PwC's analysis of close to a billion job ads landing at 56%. Comp at frontier labs has bifurcated into two worlds: enterprise ML engineers earning $170K–$245K total, and a frontier cohort commanding $600K–$1M+ for the same job title on a different letterhead.
Here's the problem: your candidates don't always know which tier they're in. And neither do you, until you lose them.
What Meta Actually Bought
Meta spent $14.3 billion last year to acquire a 49% stake in Scale AI and brought in Alexandr Wang as its first chief AI officer. Then it built Meta Superintelligence Labs from scratch and went recruiting with a checkbook that would embarrass a sovereign wealth fund.
The result: Muse Spark, Meta's first model out of the new lab, launched in April. It's competitive with leading models from OpenAI, Anthropic, and Google, but doesn't clearly beat them. By most aggregate benchmarks, Meta spent an extraordinary sum to nearly match, not beat, the competition, with a notable gap on coding.
Which raises the question every CTO should be sitting with: if nine-figure packages produce parity, not dominance, what exactly is being bought?
The answer is optionality and speed. Meta isn't paying these sums because it has verified that any particular engineer produces that much value. It's paying to not be locked out of the capability race. It's buying the possibility that this person builds the thing that matters. That's a different logic than ROI. It's closer to the logic of a strategic land grab.
That logic doesn't translate to your org. You can't operate that way, and you shouldn't try. But your talent market is being shaped by it whether you like it or not.
Three Things to Do Now
Start by resetting your comp benchmarks. The salary surveys your HR team is using are probably 18–24 months old, and the AI talent market has moved faster than any other in recent memory. If you're benchmarking against 2024 data, you're quoting candidates rates that read as out of touch. Get current numbers from sources tracking real AI offer-acceptance data, not self-reported job-board listings.
Then make the non-cash story explicit. You are not Meta. You do not have Meta's $125-to-$145 billion AI capex budget, and in most cases you can't compete on raw comp. What you can compete on is technical ownership, scope of impact, speed of iteration, the absence of bureaucracy, and meaningful equity in a company where individual engineers visibly move the needle. That pitch matters to the engineers who are genuinely excellent and genuinely motivated by craft. Make it early, not at the offer stage.
And treat speed as a comp substitute. Top AI candidates are fielding multiple offers at once (the good ones are usually gone within two weeks of going active). Every week your process drags, you lose people to whoever moved faster, even at lower total comp. Cut your process to three rounds. Designate someone who can approve an offer in 24 hours. The engineers you want aren't waiting around for your hiring committee to align.
The Engineer You're Actually Losing
When Meta reportedly puts $1.5 billion on the table for one hire, it doesn't just set a new ceiling. It sets a new frame. Every AI engineer on the market now has a reference point for what their skills might theoretically be worth at the frontier. You're not competing at the frontier. But you're recruiting from a pool that knows it exists.
Your job isn't to match that number. Your job is to understand the market it's reshaping and build a comp and hiring process that doesn't make your best candidates feel like an afterthought.
The $1.5 billion engineer isn't your problem. The $240K engineer you're losing because your process is slow and your offer is stale — that one is.
VC5 Partners places senior AI/ML engineers, engineering leadership, and technical specialists for companies building in the AI era. If your comp benchmarks or hiring process need a reset, we can help.