$100M AI Signing Bonuses
A couple weeks back, Sam Altman made headlines for saying that Meta was approaching OpenAI talent and offering "$100M signing bonuses" to jump ship over to Mark Zuckerberg's new "Superintelligence" Team being formed within Meta. A few days ago, Meta's CTO, Andrew Bosworth ("Boz"), made headlines for saying that Altman was "being dishonest" with such claims.
To be honest, the whole thing is a bit silly because of a higher level point which I'll go ahead and bury down below, but there's also seemingly enough data points now out there to triangulate what the likely truth is here.
First and foremost, the amount of publications just citing the "$100M signing bonus" – or "sign-on bonus" which is the same basic thing – as fact was sort of wild. So I wanted to go back and double check what Altman actually said. In a podcast sit down with his brother Jack Altman 11 days ago, 22:35 into a 37 minute pod, he says the following about Meta's poaching efforts:
"They started making these like giant offers to a lot of people on our team. You know, like $100M signing bonuses, more than that comp per year."
That seems pretty declarative, but in some ways it may come down to how you read "like". In some cases, you could interpret "like" to mean "around". But with Altman, it's clearly, like, a verbal tick like so many of us have. But it also could be both!
It's also entirely possible that someone told him that's what the overall offer was and he misinterpreted that. That seems less likely, but it's possible. More likely may simply be that he misspoke and conflated the signing bonus aspect of the comp with the overall package. It's also possible that he strategically misspoke on purpose, which is essentially what Boz is claiming (more on that below).
Regardless, the "more than that comp per year" would actually seem to be the more outlandish statement! Because it implies that Meta is offering not just $100M signing bonuses, but that ongoing yearly comp will exceed that amount. A $100M signing bonus is crazy. More than $100M a year in comp is insane. I mean, I guess if you're Elon Musk negotiating your comp with the Tesla board it's not, but even there, it's milestone-based. (More on that aspect in a minute.) But $100M+ a year for an employee on an ongoing basis would seemingly be out of bounds, even for trillion-dollar companies, even those trying to catch up in AI. Instead, what Altman likely means with that add-on is simply that Meta is offering over $100M when you include the yearly compensation on top of that signing bonus.
Sorry to get pedantic, but it's being discussed enough that it does seem like what is actually happening matters here. And, of course, it's being rebutted. Alex Heath of The Verge seemingly got either a recording or a transcript of Meta's internal all-hands discussing the matter, and specifically, here's what he's reporting that Boz said in response to a question about it:
Sam is just being dishonest here. He's suggesting that we're doing this for every single person… Look, you guys, the market's hot. It's not that hot.
There's actually quite a lot said – and not said – in those four simple sentences. First, Boz is implying that what Altman is saying about the Meta comp offers just isn't true. But he specifically doesn't say he's "lying" – because he himself notes that there's some level of truth here with the "he's suggesting that we're doing this for every person" which itself suggests that Meta are making such offers for some people – just being "dishonest" as if to suggest that Altman is taking something with some level of truth to it and fudging it a bit to make it less true. Boz goes on:
“And of course, he's not mentioning what the actual terms of the offer are. It's not [a] sign-on bonus. It’s all these different things.”
Here's where the circle seemingly starts to square. And it's probably why Boz is not just saying Altman is lying outright. Because he's suggesting that the $100M number (more on that in a minute) has some level of truth to it. But not as a signing ("sign-on") bonus, but as a part of an overall comp package with "all these different things."
This sort of speaks to the problem that I have with Boz's comments – not to dissimilar from the problem I have with other recent Boz comments on a totally different matter, by the way – they're misdirection at best, and disingenuous at worst. He may as well have said loudly that "look guys, we're not giving out $100M signing bonuses, that's insane..." while following that up with, quietly, under-his-breath "...but we are giving out comp packages in that range".
Who the hell cares if it's a signing bonus or an overall comp package?!1 $100M offers are wild! Perhaps warranted – we'll only know in hindsight – and probably needed given the state of the AI market and where Meta currently sits in it (look no further than the comments at the same all-hands by Meta CPO Chris Cox around actual Meta AI usage, as also cited in the Verge report). But still, wild! Hence all the headlines about the number. And the need to push back against the number – or really, again, misdirect from it.
“Sam is known to exaggerate, and in this case, I know exactly why he's doing it, which is because we are succeeding at getting talent from OpenAI. He's not very happy about that.”
Assuming the above so far is correct, I guess technically it is an exaggeration. But again, it's still directionally accurate enough in terms of what Meta is offering (and, as Boz notes, OpenAI itself may be counter-offering) to some people. Sure, it may not be the standard package for all the talent Meta is trying to bring on board, but even if it's just for a few folks, it's wild. So wild that it's easy to frame as an "exaggeration".2
So why not just come out an say exactly what the comp offers are? Well, presumably that would be an HR violation. And full comp transparency is still a touchy subject in Silicon Valley – perhaps now so more than ever thanks to the situation with AI. Boz's entire framing here is clearly to try to cut off any growing internal discontent at Meta where some AI talent is getting paid far more than their peers – even other non-AI engineering talent. Such compensation dichotomies are obviously tinder boxes for internal problems with both morale and retention. That's hardly just a Meta problem, but when the new high comp bar is set at $100M... It's a potential bonfire.
I should also note that the above may be a reason why someone might want to put such a number – exaggerated or not – out there! It's like lobbing a grenade of strife into a company. "Did you hear we're paying some people $100M?! I'm making $250k!" That kind of thing... It's a potential counter-measure, something Altman has seemingly already been doing publicly.
Anyway, while Boz never specifically confirms the $100M number, only suggesting it's not true as the signing bonus while also implying that it might be in the ballpark for some small number of overall packages, it's worth looking at some of the actual reporting on the matter.
On June 22, a few days after Altman's comment, Meghan Bobrowsky, Berber Jin, and Ben Cohen reported the following for The Wall Street Journal:
Zuckerberg has offered $100 million packages to some people. He doled out $14 billion for a stake in AI startup Scale and its CEO Alexandr Wang, who is slated to run the new AI team that Zuckerberg is assembling. At that price, he essentially made the 28-year-old one of the most lucrative hires in history.
You'll note the "packages" element here. That, of course, suggests the compensation goes far beyond a "signing bonus". WSJ also included the $100M in their headline with "pay packages".
While "$100M signing bonuses" may sound good on paper – certainly to anyone getting such an offer or even just reading about such an offer! – it also would likely be a nightmare for any company to actually implement. It's so much money up front that you would have to worry about incentives going forward. If someone got all that money right away, what stops them from leaving the minute they're no longer happy? Like say, if Meta doesn't keep giving the mouse their cookie? Or if they get another, similar offer with a massive signing bonus, elsewhere? (It's also, by the way, unlikely to be the most tax-efficient way to do compensation, for either the company or the employee!) If you just step back and think about it, it doesn't make a ton of sense to structure such a package in this way. It has to be mostly incentive-based or at the very least time-based in some way. Obviously.
As such, it seems far more likely that Meta would offer a merely "healthy" slug of money up front to entice people to leave behind their current equity stakes and come over but that most of the incentives would still be tied to performance-based earn-outs. And because Meta is a public company with fully liquid stock (aside from the reporting windows were employees cannot trade), they can easily sell that against an OpenAI, where employees have to wait for a tender offer of some sort to cash in (which, to be fair, seems to happen a lot at OpenAI!).
While signing bonuses have been soaring in sports – in particular in the NFL, where every game brings the chance of never playing again and thus, losing a lot of the value of the contract you signed (aside from whatever insurance you took out), tech is a much less risky business. Sure, you may still be an "at will" employee, but there's a reason these companies are all battling over the same people. The market is backstopping the prices here. (There's also perhaps a tangential conversation around "guaranteed" money – something you also hear in sports – and which Altman also casually mentions in the podcast, and how a company might structure that without just using a "signing bonus".) Which is a long-winded way of saying that $100M total compensation offers feel more likely than $100M signing bonuses.
And yeah, those getting such offers seem to back that up – though such public pronouncements also feel a bit coordinated, TBH. "Fake news" – what is this, 2017?
Yesterday, Cade Metz wrote the following in The New York Times:
The tech industry’s giants are building data centers that can cost more than $100 billion and will consume more electricity than a million American homes. Salaries for A.I. experts are jumping as Meta offers signing bonuses to A.I. researchers that top $100 million.
That just matter-of-factly states that "signing bonuses" "top $100 million". Presumably he's just taking Altman's word here, but if that's the case, "top" is interesting because Altman didn't actually say that – unless Metz is also roping in Altman's comments about the comp per year exceeding that. But that would be weird to rope into the "signing bonus" number, since again, that's not actually what Altman said.3 So perhaps NYT has their own sources on the matter? Further down, Metz matter-of-factly brings up the number again, but with a different context:
Mr. Zuckerberg has been offering compensation packages worth as much as $100 million a person. He and his company made more than 45 offers to researchers at OpenAI alone, according to a person familiar with these approaches.
Aha! This links to an earlier NYT report from Metz and Mike Isaac. It's actually from June 10 – when the initial Meta/Scale AI deal was going down – which pre-dates Altman's own comments by a whole week! And while it doesn't specifically cite the $100M number, it's in the range of their reporting:
Meta has tapped Alexandr Wang, 28, the founder and chief executive of the A.I. start-up Scale AI, to join the new lab, the people said, and has been in talks to invest billions of dollars in his company as part of a deal that would also bring other Scale AI employees to the company. Meta has offered seven- to nine-figure compensation packages to dozens of researchers from leading A.I. companies such as OpenAI and Google, with some agreeing to join, according to the people.
Nine-figures crosses into $100M territory, of course. But this also says "compensation packages" not "signing bonuses". Another report by Metz and Isaac yesterday does cite the $100M number in their own reporting, as follows:
And Mr. Zuckerberg and his colleagues have embarked on a hiring binge, including reaching out this month to more than 45 A.I. researchers at rival OpenAI alone. Some received formal offers, with at least one as high as $100 million, two people with knowledge of the matter said. At least four OpenAI researchers have accepted Meta’s offers.
Again, that "as high as $100 million" is the "offer" not the "signing bonus".
Also yesterday, Erin Woo of The Information cited the $100M signing bonus number (and Boz refuting it) but noted that it was something Altman said. Her actual reporting said the following:
News of the biggest offers tend to ping-pong around the entangled world of AI researchers, where everyone seems to have worked with everyone else or graduated from the same doctorate programs. One researcher reported that multiple friends called to express disbelief at the offers they were getting from Meta. On top of eight figures in equity per year, the social media company was dangling eight-figure cash signing bonuses. Such cash compensation tends to be rare at big technology companies, where stock compensation typically accounts for a significant portion of overall pay and helps with retention.
"Eight-figure", of course, suggests not $100M, but anything between $10M and $99M. And given that it's those eight-figures in equity (per year) and eight-figures in a (cash) signing bonus, that could very easily go past nine-figures – the mythical $100M threshold.
Anyway, all of the above pretty definitively points to a world where Meta is offering $100M+ packages to certain select AI talent, but to get to those numbers, it goes far beyond the signing bonus and likely includes some combination of yearly salary and equity grants on top of a still undoubtedly very healthy – eight-figure in some cases – signing bonus.
With all that in mind, it's fairly easy to see how a game of Telephone has resulted in the "$100M signing bonus" headlines, fueled by Altman himself, perhaps intentionally, perhaps not.
But again, it really doesn't matter because Meta pushing back against this notion is fairly silly if they are, in fact, offering compensation packages to some people worth $100M+. The form it takes isn't really the issue here, even if Meta wants to make it the issue to misdirect internal strife. And it's not just Altman saying that number. It's WSJ, NYT, and The Information. All of whom have great, credible reporters on such topics.
And while yes, it's a lot of money, if it gets Meta back into the AI race, it undoubtedly will be well worth it. It's also, as Julie Bort and Marina Temkin point out at TechCrunch, a compensation package that's likely in the ballpark of what executives at Meta make. Like, say, Andrew Bosworth. Because he's an officer at a public company, his compensation is public. For the past few years, his total yearly compensation package has been just under $24M. Almost all of that is in RSUs, which vest over a four-year span. Add it up over that span and it's very close to a $100M package.
That is not, like, an exaggeration.
Update July 1, 2025: Mark Zuckerberg has now formally announced those that have jumped over to Meta's ship...
M.G. Siegler
Update July 2, 2025: A couple new reports by Zoë Schiffer for Wired going into more detail on the offers that Meta is making is worth a follow-up here...
M.G. Siegler
1 I mean, the person technically receiving such money presumably would, but I digress...
2 By the way, just to note it, the compensation that Scale AI founder Alexandr Wang is likely getting thanks to the "hackquisition" of his company by Meta is undoubtedly far higher than $100M. By at least an order of magnitude, one imagines. Yes, they're technically buying his shares in that case, but still.
3 I would also just note that this is the same report that states (in a caption) that Apple "considered buying Anthropic" -- something which the piece doesn't actually say. So unless they're now breaking news in captions, this is just sloppy (they clearly meant Perplexity, which is cited and reported in the piece).