M.G. Siegler •

Zuck's Eleven

7 from OpenAI, 2 from Google, 1 from Anthropic, 1 from Sesame... But just as interesting, what's *not* mentioned.
Zuck's Eleven

Step aside, Traitorous Eight, there's a new group of ship jumpers in Silicon Valley. In his memo announcing his new AI organization, the Meta Superintelligence Labs (MSL), Mark Zuckerberg specifically lists out eleven new Meta employees who have come on board for this effort. Well, technically the list is thirteen if you include the new leads of the group, Alexandr Wang and Nat Friedman. But I'm sorry, I need to reference Ocean's 11 with my headline and image here – also, the original – as it's superior to Ocean's 13.

Ruthlessly, Zuck lists out not just the names of the eleven, but where they're coming from and what they did at those old places of employment. Here they are, per the memo obtained by Jonathan Vanian for CNBC:

  • Trapit Bansal -- pioneered RL on chain of thought and co-creator of o-series models at OpenAI.
  • Shuchao Bi -- co-creator of GPT-4o voice mode and o4-mini. Previously led multimodal post-training at OpenAI.
  • Huiwen Chang -- co-creator of GPT-4o’s image generation, and previously invented MaskGIT and Muse text-to-image architectures at Google Research
  • Ji Lin -- helped build o3/o4-mini, GPT-4o, GPT-4.1, GPT-4.5, 4o-imagegen, and Operator reasoning stack.
  • Joel Pobar – inference at Anthropic. Previously at Meta for 11 years on HHVM, Hack, Flow, Redex, performance tooling, and machine learning.
  • Jack Rae -- pre-training tech lead for Gemini and reasoning for Gemini 2.5. Led Gopher and Chinchilla early LLM efforts at DeepMind.
  • Hongyu Ren -- co-creator of GPT-4o, 4o-mini, o1-mini, o3-mini, o3 and o4-mini. Previously leading a group for post-training at OpenAI.
  • Johan Schalkwyk – former Google Fellow, early contributor to Sesame, and technical lead for Maya.
  • Pei Sun – post-training, coding, and reasoning for Gemini at Google Deepmind. Previously created the last two generations of Waymo’s perception models.
  • Jiahui Yu – co-creator of o3, o4-mini, GPT-4.1 and GPT-4o. Previously led the perception team at OpenAI, and co-led multimodal at Gemini.
  • Shengjia Zhao – co-creator of ChatGPT, GPT-4, all mini models, 4.1 and o3. Previously led synthetic data at OpenAI.

By my count, that's:

  • 7 from OpenAI
  • 2 from Google
  • 1 from Anthropic
  • 1 from Sesame

That's... a lot from OpenAI.1 Now you see why they were sending around internal notes that sound quite shaken and stirred – "someone has broken into our home".2 I can't speak directly to the quality of this team, that's above my pay grade (clearly), but it certainly seems more impactful than the downplaying of the poaching a couple weeks back by Sam Altman. It may or may not be their "best people" – but it's a lot, no question. Money wins, and all that.

Beyond that, the most interesting aspects of the memo are what's not mentioned.

Most notably, Friedman's investing partner, Daniel Gross. This immediately jumped out to me because he would be a massive get by Meta, given not just his AI work and investing through the NFDG fund, but also his current role as um, CEO of Safe Superintelligence. You know, the other "superintelligence" effort, started by Ilya Sutskever as his post-OpenAI project. And the one currently valued at $32B despite having just been started and not having any products in market – in fact, they may never have any products in market, but I digress.

It was wild that Gross was apparently going to jump over to Meta. Especially since reports stated that Zuckerberg had tried to hire Sutskever himself (and presumably tried to acquired SSI), but was rebuffed. Divide and conquer, a common tactic used by the Romans in Gaul! Again, ruthless. But the fact that Gross isn't listed here may suggest one of two things. First, either his deal isn't finalized yet.3 Second, that someone, maybe Sutskever, maybe the investors in SSI, got to him and pulled him back from the brink. I suspect we'll hear more about this very soon.

Update: we have now heard more soon, as expected, see: update below.

Also not mentioned in the memo: Yann LeCun. Never one to be quiet about the current state of AI, he's been awfully restrained in the past couple of weeks as all of this has swirled around. On the other hand, while he's been against all the talk about achieving AGI through LLMs, he's apparently on board with ASI. Still, it seems weird not to even mention your Chief AI Scientist when you just hired a Chief AI Officer...

Meta's previous model work with Llama does get one paragraph – seven paragraphs into the memo:

I’m excited about the progress we have planned for Llama 4.1 and 4.2. These models power Meta AI, which is used by more than 1 billion monthly actives across our apps and an increasing number of agents across Meta that help improve our products and technology. We’re committed to continuing to build out these models.

Leading with "I'm excited" probably says all you need to know there. But just in case, the billions now being spent to play catch up after the billions spent on Llama drills home the point: Llama, at least as we previously knew it, is yesterday's AI news within Meta.

In parallel, we’re going to start research on our next generation of models to get to the frontier in the next year or so. I’ve spent the past few months meeting top folks across Meta, other AI labs, and promising startups to put together the founding group for this small talent-dense effort. We’re still forming this group and we’ll ask several people across the AI org to join this lab as well.

"It's better to be a pirate than join the navy." We're gonna need a picture of this group together, for good measure. Ideally in suits. But one well-placed bow-tie and a pirate flag will work too. Fine, I'll accept one in front of the Bellagio fountain.

One more thing: one phrase never mentioned once in the entire memo: "open source". Interesting.


Update July 2, 2025: Some further thoughts and analysis on the comp offers...

Meta’s ‘Godfather’ Offer
Which, amazingly, everyone has seemingly refused – so far…
Can Meta Buy the Future of AI?
No.

Update July 3, 2025: And sure enough, the Daniel Gross shoe has finally dropped. As it turns out, he is joining Meta, as Shirin Ghaffary and Kurt Wagner report for Bloomberg. And yes, it seems likely that his deal wasn't finalized yet when the others were announced per above (or else Zuckerberg would have announced it, obviously). Though interestingly, Ilya Sutskever, in commenting on the news, noted that Gross technically stopped working for SSI on June 29 – the day before Zuck's memo. Perhaps that's just a vesting/leaving issue and he hadn't yet finalized his Meta deal. No word on Gross' deal terms, though he'll apparently be working on "AI products". Imagine that.

Yuliya Chernova of WSJ also has more details about how Meta is structuring the "hackquisition" of Gross and Nat Friedman's NFDG fund, confirming a lot of what was previously reported. One new key detail: Meta would apparently be paying current book value for the "up to 49%" stake they buy from LPs, meaning a nice, quick and easy 4x return on those positions (while the rest presumably can ride...).

With those numbers in mind, we can do some quick and dirty math here. If the $1.1B fund is "about half" deployed – let's call it $500M to make it easy – and the book value of those holdings is up 4x, that's $2B. If Meta is acquiring "up to 49%" – again, let's call it 50% to keep it simple – that's $1B that they're likely paying for the stakes. And it's also stated that the fund won't make new investments, so they're presumably returning that remaining $500M to LPs. So that's $1.5B returned on the $500M invested, with $500M (currently marked up 4x) still riding. A nice, quick and clean return for investors with good upside potential left!

Also, assuming Gross and Friedman got a slight premium on the "2 & 20" model, let's guess they have 2.5/25 terms on the fund. That means they would likely get 25% of $500M (the profit of the fund after $500M deployed is fully paid back to LPs, and the unused $500M is returned). That's $62.5M each (minus any carry they gave to others helping with the fund) for a couple years worth of work, not bad. And they still have the upside on the money left riding too – which is now in carry and again, marked up 4x on paper... So that's another $250M worth of carry as it stands right now — each.

And presumably Meta added some sweeteners on top of the above to get them to walk away from their fund. A "Godfather" offer perhaps?...

And how's this for awkward: that fund's crown jewel? Safe Superintelligence.


👇
Previously, on Spyglass...
$100M AI Signing Bonuses
Exaggeration, conflation, a game of Telephone, directional accuracy, brinksmanship, or all of the above?
Meta’s $10B+ AI Reset
Zuckerberg recruits a band of pirates to shake up and wake up their AI efforts – including a new don’t-call-it-a-deal for Scale AI talent…
Hackquisitions & Hackquihires
A look at Meta’s Scale deal in relation to other such deals…
Meta Hackquires a VC Fund
“Team Superintelligence” continues to assemble at Meta, but it’s getting even more awkward…
Meta’s Open Source AI Mistake
The writing isn’t just on the wall for Llama, it’s on the new paychecks…

1 With a special shout-out to Google as 5 worked there at some point, which continues the narrative that modern AI was very much born at Google but they let aspects slip away.

2 And it doesn't include the three OpenAI employees moving over that were previously announced/reported. So that's... 10 from OpenAI alone.

3 The fact that Zuck states that Friedman will "define his role going forward" – not to mention the sloppiness of the whole "lead" and "partner to lead" element for Wang and Friedman around MSL suggests this was all a bit rushed. And that may lend credence to Gross' deal simply not being done. We'll see.