After being "raided" by Zuckerberg, OpenAI launches a strong counterattack: poaching four top talents in a row. Was Elon Musk also "collaterally damaged"?

In the past few weeks, Silicon Valley has witnessed a fierce talent war, which can be described as a "spy thriller in the era of the AI arms race."

The main players in this drama are OpenAI, the rising star behind ChatGPT, and Meta, which is aggressively poaching talent in an attempt to "create the next - generation AGI." After having its talent poached by Meta, OpenAI urgently announced a one - week vacation for all employees. Meanwhile, Meta has spent hundreds of millions of dollars to establish a new laboratory and has poached 11 top researchers from leading AI companies all at once.

This week, as OpenAI's vacation ended, the situation of its "passive defense" seems to be quietly reversing. According to foreign media outlet WIRED, OpenAI is launching a counter - attack and has recruited four top engineers from Tesla, xAI, and Meta to join its crucial Scaling team.

Reversal of "Poaching": OpenAI Recruits Four Top Talents with High Salaries

According to the report from WIRED and an internal notice sent by Greg Brockman, the co - founder of OpenAI, on Tuesday in the internal Slack channel, the four key members who have recently joined OpenAI's Scaling team are:

● David Lau: Former vice - president of software engineering at Tesla, who has long been responsible for the implementation of autonomous driving and AI systems.

● Uday Ruddarraju: Former head of infrastructure engineering at xAI and X, who has presided over the construction of a super - large - scale computing power platform.

● Mike Dalton: A senior infrastructure engineer from xAI, who has worked with Ruddarraju for many years.

● Angela Fan: Former researcher at Meta AI, focusing on model training and algorithm optimization.

The most eye - catching among them are Uday Ruddarraju and Mike Dalton. During their time at xAI, they led the development of Colossus - a supercomputing platform composed of 200,000 GPUs, which is one of the world's most powerful AI infrastructures.

In addition, David Lau's joining means that OpenAI has gained a "veteran" polished in the traditional high - intensity engineering system. He once managed the entire software engineering system at Tesla, responsible for the construction and deployment of the core architecture of the autonomous driving system. Angela Fan can inject more diverse perspectives into OpenAI's product capabilities.

Hannah Wong, a spokesperson for OpenAI, said, "We are very pleased to welcome these top talents to the Scaling team. Our goal is to continuously build and integrate world - class infrastructure, research, and product teams to accelerate our mission and bring the benefits of AI to hundreds of millions of users worldwide."

Why "Scaling"? Infrastructure is the Real Battleground for AI Victory

After the emergence of ChatGPT, people generally focused on what it can do and how smart its answers are. But behind these appearances, what really supports its leap - forward growth is the continuously invested underlying infrastructure, including larger data sets, stronger computing power, and more efficient training systems.

OpenAI's Scaling team is responsible for the construction and maintenance of the entire backend hardware and software systems, including data centers and training platforms, which are the main paths for improving the model's intelligence. Therefore, since its inception, the "Scaling" team has been highly anticipated. It is not a product line for users, but the most fundamental and crucial cornerstone for OpenAI's sprint towards AGI.

This is particularly evident in the "Stargate" project - a new AI infrastructure center jointly built by OpenAI and its partners, with a planned investment of $500 billion. The goal is to provide computing power and data support for next - generation AI technologies (such as general artificial intelligence AGI).

As Uday Ruddarraju said in an interview, "Infrastructure is where research turns into reality, and OpenAI has proven this with practical actions. Especially Stargate, which is a real infrastructure - level'moon landing project' that perfectly matches my passion for system - level challenges."

Moreover, David Lau also added, "I am increasingly convinced that promoting the rapid progress of safe and aligned general artificial intelligence is the most worthwhile cause for me in my next career."

Netizens Tease: How Did Musk Get "Inadvertently Hit"?

OpenAI's act of poaching talents from xAI, Tesla, and Meta has made many people sigh again that the AI field is extremely competitive. Not only is the technology iteration fast, but the speed of poaching and counter - poaching is also quite astonishing.

Just last week, Mark Zuckerberg, the CEO of Meta, officially announced the establishment of the long - awaited Meta Superintelligence Labs (MSL), poaching 11 core technical personnel from multiple laboratories such as OpenAI, Anthropic, and Google DeepMind in a "stealing - home" manner - seven of them were from OpenAI. This week, OpenAI immediately announced the introduction of four heavy - weight new members, which is undoubtedly a powerful counter - attack.

Interestingly, some netizens joked, "We thought OpenAI was just having a feud with Zuckerberg, but unexpectedly, Musk got inadvertently hit."

Those familiar with the development history of OpenAI know that Musk was originally a co - founder of OpenAI. However, due to differences in development routes and disputes over management rights, he left in 2018. In 2024, he officially sued OpenAI, accusing it of deviating from its original intention of "benefiting all mankind" and becoming a vassal of Microsoft's capital. Later, OpenAI also counter - sued Musk, accusing him of maliciously interfering in the company's operations and violating contracts.

Now, OpenAI's act of "taking away" talents from companies under his name has further brought the conflict between the two parties to the surface.

Reference link: https://www.wired.com/story/openai-new-hires-scaling/

This article is from the WeChat official account "CSDN". It was compiled by Zheng Liyuan and published by 36Kr with authorization.