Somewhere in Menlo Park, there's a name that still makes recruiters wince: Ruoming Pang. Not because of what he built. Because of what he walked away from.
In July 2025, Meta handed this one AI researcher a pay package worth more than $200 million to leave Apple and join Mark Zuckerberg's new Superintelligence Labs. Seven months later, he was gone — poached again, this time by OpenAI. Two of the richest companies on Earth had just spent nine figures fighting over one engineer, and neither of them got to keep him for long.
This isn't a story about a lucky hire. It's a window into how unhinged the AI talent war has become, and what happens when money stops being the thing that buys loyalty. Here's the real story, fact by fact.
Fact #1 - The $200 Million Number Everyone Slightly Misquotes
Every headline says "$200 million." Almost none of them explain what that actually means.
- ✅ The figure is a multi-year total, not a single paycheck or signing bonus
- ✅ It reportedly combines base salary, stock grants, and performance-tied bonuses stretched across several years
- ✅ Some of Meta's other super-intelligence offers reportedly ran even higher, with top researchers offered packages of up to $300 million over four years
- ❌ It was not, as some early reports implied, $200 million handed over on day one
For context, Meta's own senior research scientists typically earn between roughly $305,000 and $581,000 a year in base pay before stock and bonuses. Pang's deal wasn't a raise. It was an entirely different category of compensation, built specifically to win a bidding war Meta could not afford to lose.
Fact #2 - He Wasn't a Rookie, He Ran Apple's Entire AI Brain Trust
The internet loves to frame these mega-hires as mystery men who appeared out of nowhere. Pang was the opposite of a mystery hire.
Before Meta came calling, he led Apple's Foundation Models team, a group of roughly 100 engineers responsible for the AI systems behind:
- ✅ Apple Intelligence's writing and summarization tools
- ✅ Genmoji generation
- ✅ Email and notification summaries
- ✅ Core on-device large language model research
He joined Apple in 2021 after working at Google, meaning by the time Meta approached him, he had already spent years at two of the most guarded AI research organizations in the world. Apple reportedly didn't even counter Meta's offer — insiders said the number was so far outside normal compensation bands that matching it wasn't realistic for anyone outside CEO-level pay.
Fact #3 - Zuckerberg Personally Recruited Him at His Own Homes
This wasn't a recruiter cold-emailing a LinkedIn profile. Zuckerberg reportedly ran this recruiting push himself.
According to Bloomberg reporting, Zuckerberg:
- ✅ Personally invited top AI targets to his private homes in Palo Alto and Lake Tahoe to pitch them directly
- ✅ Rearranged physical office seating at Meta so the new superintelligence hires would sit closer to him
- ✅ Built a secretive internal shortlist — reportedly nicknamed "The List" — of the world's most sought-after AI researchers to personally chase
This level of hands-on CEO involvement is rare for any hire below the executive suite. For Meta, landing figures like Pang wasn't a staffing decision. It was treated as a strategic, company-defining move toward Zuckerberg's AGI ambitions.
Fact #4 - The Money Was Never "Free Cash" - It Was Chained to Milestones
Here's the detail most viral posts leave out, and it matters more than the headline number itself.
- ✅ Bloomberg reported Pang's package was tied to performance milestones and multi-year vesting schedules
- ✅ Leaving before those milestones hit meant walking away from a large chunk of unvested stock
- ❌ The "$200 million man" did not simply pocket $200 million and disappear
This structure is common across Meta's superintelligence offers — it's designed specifically to punish early exits and lock in retention. Which makes what happened next even more remarkable: Pang left anyway, giving up a meaningful slice of that fortune to do it.
Fact #5 - He Told Colleagues He Was "Very Happy" at Meta - Then Left Anyway
This is the detail that made the whole story blow up in tech circles, not just business pages.
- ✅ According to reporting from The Information, Pang told colleagues at Meta he was genuinely happy with his role there
- ✅ He was overseeing AI infrastructure for the Superintelligence Labs, not a side project — a core, high-visibility position
- ✅ He still chose to leave for OpenAI after months of quiet recruitment
That combination is what makes this story stick. Normally when a high-profile hire jumps ship early, the narrative writes itself: bad culture fit, internal politics, broken promises. Here, by his own account, none of that seemed to be the driver. He apparently liked the job. He left anyway.
For an industry that likes to assume money and mission are the only two levers that matter in retention, that's an uncomfortable data point.
Fact #6 - Meta Actually Restructured Its Offices Around People Like Him
Zuckerberg's involvement in Superintelligence Labs didn't stop once these researchers signed. It shaped how the entire building operated.
- ✅ Meta reportedly rearranged internal seating so newly hired superintelligence researchers sat physically closer to Zuckerberg himself
- ✅ Meta CTO Andrew Bosworth told employees that a small number of leadership-level AI roles "command a premium," acknowledging the eye-watering pay gap internally
- ✅ The lab was positioned as Meta's most strategically important division, ahead of many established product teams
The point of all this proximity and access wasn't comfort. It was retention by immersion — the theory being that researchers who feel like they're inside the center of gravity at a company are less likely to leave it. Pang's exit is the clearest evidence yet that proximity to the CEO isn't the retention tool Meta hoped it would be.
Fact #7 - He Wasn't Alone: Two More Apple Engineers Followed Him to Meta
Pang's move wasn't an isolated transaction. It triggered a chain reaction inside Apple's AI team.
- ✅ Apple engineers Mark Lee and Tom Gunter, who had previously worked closely with Pang on Apple's Foundation Models team, also left for Meta
- ✅ Lee joined shortly after Pang; Gunter followed within weeks
- ✅ Insiders told Bloomberg that additional AFM engineers were privately telling colleagues they were considering leaving too
This is the part of the story that should worry Apple more than any single departure. When a team lead leaves, a company can usually absorb it. When the team lead's closest collaborators start following him out the door one by one, that's not a hiring loss — that's a talent pipeline collapsing in real time, right as Apple was already viewed as playing catch-up on AI.
Fact #8 - His Real Job Wasn't Chatbots - It Was AI Infrastructure
A lot of coverage lazily lumped Pang in with "chatbot researchers." That's not actually what made him valuable.
- ✅ At Meta, Pang oversaw AI infrastructure for the Superintelligence Labs — the unglamorous but critical systems that let large models actually train and run at scale
- ✅ At Apple, his team built the foundation model architecture powering on-device features, not customer-facing chat products
- ✅ Infrastructure and foundation-model expertise is currently rarer — and more expensive — than general machine learning talent, because there are only a handful of engineers who have done it at the scale Big Tech now needs
That distinction matters because it explains why OpenAI wanted him specifically, rather than any of the dozens of other AI researchers currently on the market. Companies aren't just buying brand names anymore. They're buying the small number of people who have actually shipped frontier-scale AI systems before, infrastructure and all — and that list is short enough that losing even one name to a competitor is a real strategic blow.
Fact #9 - OpenAI Chased Him for Months Before He Said Yes
The OpenAI hire wasn't a quick opportunistic grab. It was a slow, deliberate campaign.
- ✅ According to The Information's reporting, cited by Reuters and Yahoo Finance, OpenAI ran a months-long recruitment effort specifically targeting Pang
- ✅ The approach reportedly started well before he had even settled fully into his Meta role
- ✅ He officially left Meta for OpenAI in late February 2026 — roughly seven months after joining
That timeline matters. This wasn't a case of a researcher getting a cold offer he couldn't refuse on a random Tuesday. It was a sustained, patient recruiting push from a direct competitor, happening quietly while Pang was still actively working inside one of Meta's most guarded divisions. For an industry that treats non-competes and quiet periods as normal, that's a notably aggressive move.
Fact #10 - Leaving Early Meant Walking Away From Most of the Money
This is the part that should reframe how you read every "$200 million hire" headline going forward.
- ✅ Because Pang's package was tied to multi-year vesting and performance milestones, leaving after seven months meant forfeiting the large majority of the total value
- ✅ Reports estimate he likely realized only a small fraction of the headline $200 million figure before departing
- ❌ The idea that he "banked $200 million and walked" is simply not how these deals are structured
As one industry commentary put it bluntly: $200 million couldn't buy seven months of loyalty. That's not a knock on Pang specifically — it's a signal to every company currently throwing nine-figure offers at AI researchers that vesting schedules and paper valuations don't guarantee retention, no matter how large the number looks in a press release.
Fact #11 - He's Not the Only High-Profile Departure From Meta's Dream Team
Pang's exit is part of a pattern, not an isolated incident, and that's the detail most headlines buried.
- ✅ Mat Velloso, product lead for Meta Superintelligence Labs' developer platform, announced his own departure the same week — after joining from Google DeepMind less than eight months earlier
- ✅ Both exits came just months after Meta's roster launch, which had been publicly announced with fanfare in July 2025
- ✅ Multiple engineers who followed Pang from Apple to Meta were, according to insiders, already being courted by other labs shortly after arriving
A team that was assembled specifically to be Meta's best shot at artificial general intelligence is already losing pieces less than a year in. That's a retention problem money alone clearly hasn't solved, and it raises real questions about whether Meta's roster can stay intact long enough to deliver on Zuckerberg's AGI timeline.
Fact #12 - The $200 Million Question: What This Means for the Next Hire
Zoom out, and this story stops being about one engineer and starts being about the entire AI labor market.
- ✅ Compensation at this level is now a normalized recruiting tool across Meta, OpenAI, Google, and Microsoft — not a one-off headline stunt
- ✅ Meta has reportedly made offers exceeding $300 million over four years to other researchers, and even a reported $1.5 billion offer to one candidate who turned it down
- ✅ Google has separately spent billions securing AI talent through deals like its $2.4 billion Windsurf agreement and $2.7 billion Character.AI deal
- ✅ Despite all of that spending, none of it has proven it can guarantee loyalty once a researcher is inside the building
The real lesson of the "$200 Million Man" story isn't that Meta overpaid. It's that in 2026, the ceiling on AI compensation keeps rising while the thing that actually keeps top researchers in place — mission, autonomy, team stability — is still something no amount of money has reliably bought. Pang's seven-month stint at Meta may end up being remembered less as a hiring win and more as the moment the entire industry realized that.
The Full Timeline: How a $200 Million Hire Unraveled in Seven Months?
Sometimes the clearest way to see how wild a story is isn't through analysis — it's through the raw timeline. Here's exactly how the "$200 Million Man" saga played out, start to finish.
- ✅ 2021 — Ruoming Pang joins Apple from Google, eventually rising to lead the company's Foundation Models team
- ✅ Early July 2025 — Meta offers Pang a multi-year compensation package reportedly exceeding $200 million to join its new Superintelligence Labs
- ✅ July 2025 — Pang accepts and departs Apple; two of his closest Apple collaborators, Mark Lee and Tom Gunter, follow him to Meta in the following weeks
- ✅ Mid-to-late 2025 — Zuckerberg personally continues recruiting additional AI talent, reportedly offering one candidate up to $1.5 billion and restructuring Meta's offices around the new hires
- ✅ Late 2025 — OpenAI reportedly begins a quiet, months-long campaign to recruit Pang away from Meta
- ✅ February 2026 — Pang officially leaves Meta for OpenAI after roughly seven months, forfeiting a large share of his unvested compensation in the process
- ✅ Same week — Mat Velloso, another senior Superintelligence Labs hire, also departs Meta after less than eight months on the job
Laid out like this, it's less a story about one expensive researcher and more a case study in just how fast the ground can shift underneath even the most carefully engineered hiring strategy in tech.
What This Means for Apple, Meta, and OpenAI Going Forward?
Each company walks away from this story with a very different lesson.
For Apple, the damage isn't just losing Pang — it's the ripple effect. Losing a team lead and two of his closest engineers in quick succession, while already being viewed as behind in the AI race, adds real pressure to a company that reportedly considered further AI acquisitions to fill the gap left behind.
For Meta, this is a retention warning sign wrapped inside a hiring win. Zuckerberg proved Meta can out-bid almost anyone for talent. What he hasn't yet proven is that Meta can keep that talent once the novelty of a nine-figure offer wears off, especially when a rival lab is willing to play the long game to win someone back.
For OpenAI, the hire is a quiet but meaningful statement: even against Meta's checkbook, patient recruiting and reportedly a stronger sense of mission alignment can still win. It also signals that the AI talent war isn't a one-directional flow of researchers toward whoever pays the most — it's a genuinely fluid, ongoing contest.
Frequently Asked Questions (FAQ)
Who is Ruoming Pang?
Ruoming Pang is an AI researcher who led Apple's Foundation Models team from 2021 until mid-2025, overseeing the systems behind Apple Intelligence features like Genmoji and email summaries, before joining Meta and later OpenAI.
Did Ruoming Pang actually receive $200 million?
The $200 million figure represents the total reported value of a multi-year compensation package tied to performance milestones and vesting schedules, not a lump-sum payment. Because he left Meta after roughly seven months, he reportedly did not realize the full value of that package.
Why did Ruoming Pang leave Meta after only seven months?
According to reporting from The Information, OpenAI ran a months-long recruitment campaign to bring him on board. Pang reportedly told Meta colleagues he had been happy there, suggesting the move was driven by OpenAI's persistent outreach rather than dissatisfaction at Meta.
What is Meta's Superintelligence Labs?
Meta's Superintelligence Labs (MSL) is the division Mark Zuckerberg established in mid-2025 to accelerate the company's push toward artificial general intelligence, led in part by former Scale AI CEO Alexandr Wang and former GitHub CEO Nat Friedman.
Are AI researchers really being paid hundreds of millions of dollars?
Yes. Multiple reports confirm that Meta, OpenAI, Google, and Microsoft have each offered top AI researchers compensation packages ranging from tens of millions to several hundred million dollars, reflecting how scarce elite AI infrastructure and research talent has become.