Per the WSJ this weekend, OpenAI has now completely eliminated its equity vesting cliff for new employees. Just a few months ago, they’d already shortened it from the industry-standard 12 months down to 6 months. Now? Zero.
You walk in the door at OpenAI, you’re vesting from Day One.For the past 20+ years, the startup equity playbook has been essentially unchanged: 4-year vest, 1-year cliff.
This means:
- You get nothing if you leave before 12 months
- At month 12, you vest 25% all at once
- Then you vest monthly or quarterly for the remaining 3 years
- The cliff exists for a reason. It protects companies from giving meaningful equity to someone who joins and leaves after 3 months. It’s a retention mechanism. It’s also, frankly, a risk-sharing mechanism — you prove your worth for a year, then you earn your shares.
OpenAI just threw that entire model out the window.
Why? Fidji Simo, OpenAI’s applications chief, explained it was designed to “encourage risk-taking by new employees, without the fear of being let go before they get their first chunk of shares.”
Translation: We want elite AI researchers to feel safe taking a risk on us. We don’t want them sitting at Google or Meta or Anthropic, looking at a potential OpenAI offer and thinking “but what if I wash out in 8 months and get nothing?”
The talent is just that scarce.
OpenAI is now projected to spend $6 billion this year on stock-based compensation alone. That’s nearly half of their expected ~$13 billion in revenue.
Let me put that in perspective:
- NVIDIA, a $3.3 trillion public company with 30,000+ employees, spent about $2.2 billion on stock compensation in the first half of 2024
- OpenAI, at maybe 3,000-6,000 employees (the numbers vary wildly by source), is spending 3x that
- The average OpenAI employee is earning somewhere between $400,000 to $2,000,000 in stock compensation alone. The median total compensation is reportedly around $1.37 million. Senior researchers can earn $10 million+ annually.
These aren’t numbers. They’re a war chest.
The Zuck Soup Effect
This move didn’t happen in a vacuum. It’s a direct response to what’s become the most aggressive talent war Silicon Valley has ever seen.
Mark Chen, OpenAI’s chief research officer, revealed something remarkable on a recent podcast: Mark Zuckerberg personally hand-delivered homemade soup to people he was trying to recruit from OpenAI.
Yes, the CEO of a $1.5 trillion company made soup. With his hands. And brought it to people’s homes.
“It was shocking to me at the time,” Chen said. “But over time I’ve updated toward these things can be effective in their own way.”
And then, in the most Silicon Valley response possible, Chen admitted: “I’ve also delivered soup to people we’ve been recruiting from Meta.”
The difference? Chen buys his from Daeho, a high-end Korean soup spot in the Valley. Because of course he does.
But here’s what’s actually happening behind the soup memes:
- Meta is offering packages reportedly as high as $100 million to poach top OpenAI researchers
- Some packages have gone up to $300 million over 4 years for the most senior talent
- Meta has a reported $10 billion annual budget just for AI talent
- They’ve successfully poached at least 10 researchers from OpenAI, including key architects of the o-series models and GPT-4o voice mode
- And it’s not just Meta. Anthropic is preparing its first employee tender offer. Google DeepMind is paying retention packages. xAI shortened their own vesting cliff over the summer in the same arms race.
Why The Cliff Actually Mattered (And Why Killing It Is Risky)
The cliff exists to protect the company from bad hires.
If you hire someone who turns out to be wrong for the role — culturally, capability-wise, whatever — you have 12 months to figure that out before they walk away with equity.
By eliminating the cliff, OpenAI is essentially saying: “We’ll give you equity from day one, and if we have to let you go in month 3, you’re keeping what you’ve vested.”
At their compensation levels, that’s not nothing. Three months at OpenAI vesting equity at a $500 billion valuation could easily be worth hundreds of thousands of dollars.
This only makes sense if:
- Your hiring bar is so high that you’re confident virtually everyone you hire will work out
- The cost of NOT getting top talent is higher than the cost of occasionally giving equity to short-tenured employees
- You’re in a liquidity-rich environment where the equity is actually worth something
- OpenAI checks all three boxes. They’ve run $6.6 billion in employee tender offers. The equity is liquid. The competition is insane.
The Retention Scorecard
Interestingly, according to SignalFire’s State of Talent 2025 report, aggressive compensation isn’t always correlated with retention:
- Anthropic: 80% retention (2+ year employees still there)
- Google DeepMind: 78% retention
- OpenAI: 67% retention
- Meta: 64% retention
- Meta is spending the most aggressively on talent acquisition, yet has the worst retention of the major labs. Maybe missionaries really do beat mercenaries in the long run.
But OpenAI’s 67% isn’t great either. That’s why they’re also throwing retention bonuses at nearly 1,000 employees — up to $1.5 million for certain levels, with most technical staff getting at least $300,000 bonuses vesting over 2 years.
What This Means For The Rest of Us
If you’re building a startup and trying to hire AI talent, this is… not great news.
You’re now competing with:
Zero cliff vesting
$6 billion in annual stock compensation
Regular tender offers at $500 billion valuations
Retention bonuses in the hundreds of thousands
CEOs delivering soup
For most startups, you can’t match this on economics. You can’t even come close.
So what can you do?
1. Mission matters more than ever. Even Chen admitted that when people have Meta offers, “I haven’t heard anyone say AGI is going to be developed at Meta first.” People want to work on the frontier. They want to matter.
2. Speed and scope of impact. At a startup, you might own the whole problem. At OpenAI or Meta, you might be one of 50 people working on a piece of a piece.
3. Founder equity, not employee equity. If you can give someone true founder-level upside (not .05% vesting over 4 years), the math changes.
4. Location flexibility. Not everyone wants to live in SF. Remote-first with occasional offsites is a real differentiator for some candidates.
5. Your own liquidity events. If you’re at scale, consider regular tender offers. Nothing builds loyalty like actual money in the bank.
The Great Financialization of S-Tier AI Talent
What we’re watching is the financialization of talent at a level we’ve never seen before.
OpenAI isn’t just building products. They’re not just a hot startup. They’re a $500 billion financial instrument that happens to employ AI researchers. And they’re using that financial instrument — liquid equity, accelerated vesting, massive bonuses — as the primary weapon in a war for a few hundred people who can actually push the frontier of AI.
The irony is that vesting cliffs were originally designed to align employee interests with company success over time. You had to stay and help build the thing to get your shares.
Now, at the very top of the market, the script has flipped. Companies are so desperate for talent that they’re giving away the thing that used to require patience and commitment.
It’s a seller’s market for AI researchers. The question is how long that lasts.
And for the rest of us watching from the sidelines of this talent war?
We’re playing a different game entirely.
