Zuckerberg’s sky-high contract of US$200 million failed to retain this top talent in basic models.February 26,OpenAICompleted a textbook poaching,Recruited Daniel Pang Ruoming, who has only been with Meta for seven months.

Pang Ruoming was once regarded as the core genius of "Apple's AI backbone". He decided to give up the astonishing billion-dollar option incentive before he even took up his position at Meta, and resolutely turned around and joined Ultraman.
People familiar with the matter revealed that OpenAI had been poaching him for several months.Although Pang Ruoming told his colleagues that he was happy working at Meta and that the infrastructure team was in good condition, he ultimately chose to leave.

1. A top-notch brain that cannot be retained even with US$200 million.
Naturally, it is not ordinary people who can make OpenAI so persistent in poaching people.
Pang Ruoming graduated from Shanghai Jiao Tong University and received his PhD in computer science from Princeton University in 2006. His career was equally impressive. After five years of intensive work at Google, he became the leader of Apple's basic model team in 2021, responsible for developing the core system behind Apple Intelligence. At that time, Apple's progress in the field of AI was relatively slow, and it even had to consider introducing Anthropic or OpenAI models to provide support for Siri. This acquiescence to his own limitations reflects Pang Ruoming's key value in Apple's internal breakout operation.

In 2025, Meta forcibly poached Pang Ruoming with a salary package worth more than US$200 million over several years.This contract, which includes a basic salary, a signing bonus and a large amount of stock, is the ceiling of Silicon Valley's top talent market.
Compare this to:
Apple executive salaries:Except for CEO Tim Cook, the annual salary of other executives is less than $28 million.
Rank gap:An engineer one level below Pang Ruoming's original position usually earns an annual salary of around US$778,000.
Now it seems that this contract, which is enough to achieve freedom of wealth, has not locked up the genius's whereabouts. He still chose to give up his huge salary and switch to OpenAI.
2. Meta’s brain drain: money is not a panacea
Pang Ruoming’s lightning departure is another member of Meta’s core AI team that has been lost in the past year.
Last week, Meta Superintelligence Lab’s Mat Velloso left the company after a short tenure. Previously, Yann LeCun, Meta’s long-time chief AI scientist, also bid farewell to the front line of management. There is a lot of discussion in the industry: Is Zuckerberg really unable to pay?

But it turns out that simplifying the problem to "not enough money" obviously underestimates the ambitions of the top players.At Pang Ruoming's level, the numerical difference of an annual salary of over 100 million may be far less important than "who to work with" and "which direction to pursue."The appeal of OpenAI, which comes with its own sense of mission, has never been just based on salary.And the battle for talent is never one-way. In order to fight back, Zuckerberg once tried to offer a $100 million signing bonus when poaching people from OpenAI.This war for talent without gunpowder has long gone beyond the scope of ordinary business logic.
3. Zuckerberg’s big bet on AI: Fatal shortcomings after stacking up computing power
Meta is using violent recruitment to ignite the craziest talent war in the history of Silicon Valley.In order to compete for tickets to lead the next wave of AI, technology giants are not willing to throw out personal solutions worth millions or even tens of millions of dollars.
Pang Ruoming is in charge of the Super Intelligence Laboratory at Meta, a pioneering team aiming to build systems more powerful than humans. He worked under Scale AI co-founder and former CEO Alexandr Wang and former GitHub CEO Nat Friedman.
Other members include Daniel Gross, who previously worked at OpenAI co-founder Ilya Sutskever’s startup Safe Superintelligence.
Zuckerberg is clearly making a big move about the future. Not only did he declare on Threads that he would build the most dense talent team in the industry, he also planned to invest hundreds of billions of dollars in computing resources for the project.
To this end, Meta is building multiple supercomputing networks:
Prometheus:It is expected to be officially launched in 2026.
Hyperion:It is planned to expand to a terrifying scale of 5 gigawatts in the next few years.
Zuckerberg hopes to use the highest per capita computing power to retain the top researchers, but Pang Ruoming's departure has obviously cast a shadow on this grand plan: computing power can be bought, but the hearts of top talents may not be tied to computing power.
4. Apple’s embarrassing situation: the aftershocks after losing its spine
Looking back at this talent flow across Silicon Valley, Apple, which had "bleeded" seven months ago, is still worth examining.
As a person who played a key role in Apple's embedded AI research and development, Pang Ruoming participated in the leadership of the basic model team, which is an important technical force in Apple Intelligence's attempt to achieve a balance between privacy and performance on the device side. This device-side architecture was once regarded as Apple’s differentiated advantage in the AI game.
Pang Ruoming's resignation seven months ago, although it did not collapse Apple's technology building, did interfere with its independent research and development rhythm to a certain extent.
Although Apple has since urgently reorganized its technical architecture, the loss of key talents is often symbolic. This also explains why subsequent versions of Siri will choose to cooperate with external forces such as Google Gemini. This switching of technical paths itself reflects the realistic pressure Apple faces in self-developed core models.
As of press time, Meta, Apple and OpenAI have not officially responded to this personnel change.But what is certain is that the AI talent war in Silicon Valley has reached an almost crazy extreme state: $200 million can get you a job, but whether you can stay depends on more than the zero on the check.