Why Anthropic Is the Only Top AI Lab That Keeps Every Cofounder

OpenAI has lost 8 out of 11 cofounders. Ilya Sutskever left to start SSI. John Schulman moved to Anthropic. But Anthropic still has all 7 cofounders on board. Stability might be the strongest weapon in the AI race.

How high can employee retention go at top AI labs?

Eighty percent.

This number came out last year.

And it does not belong to Google DeepMind. It does not belong to Meta. It does not belong to OpenAI. It belongs to Anthropic.

ai video generator porn

According to research from venture capital firm SignalFire, among employees who joined between 2021 and early 2023, Anthropic kept 80 percent still working there.

Google DeepMind sits at 78 percent. OpenAI is at 67 percent. Meta trails at only 64 percent.

Almost every top AI lab in America has lost at least one cofounder.

The only exception is Anthropic.

Anthropic cuck chat has 7 cofounders and none of them have left. That alone makes the company worth studying.

Why Anthropic Is the Only AI Lab That Keeps Every Cofounder

Why have all 7 Anthropic cofounders stayed? One reason stands out. Earlier this year, Anthropic locked in a 60 billion dollar valuation. All 7 cofounders became billionaires overnight.

But regular employees also stay. Anthropic holds an 80 percent retention rate across the board.

AI startup founder Yuchen Jin was curious about this. He ran a poll on X and asked people why Anthropic keeps its people. Here are the top answers he got:

1. Claude Code. 2. Long stock vesting schedules. 3. A strong safety mission. 4. Better food in the office. 5. Fast growing revenue. 6. Long posts by Dario Amodei on internal Slack. 7. The belief that coding is the path to AGI.

Every option got votes.

The food answer got a lot of support. YouTuber and angel investor Matthew Berman said Anthropic has such good desserts that he forgets about AGI when he visits.

But Yuchen Jin could not believe Anthropic serves better food than xAI.

He thinks Anthropic has the rarest trait among top AI labs. He started betting on Anthropic to win.

He says he now believes Anthropic might win simply because there is no drama.

And the numbers back him up. Among the big four AI labs, Anthropic has the highest retention rate.

On top of that, engineers leave OpenAI for Anthropic at 8 times the reverse rate. Engineers leave DeepMind for Anthropic at nearly 11 times the reverse rate.

SignalFire said this trend is partly expected. Anthropic is the hot new player. DeepMind has older and larger teams so some people want a change.

But the gap is still shocking.

When it comes to hiring, Anthropic is crushing OpenAI and DeepMind in a one sided battle.

Still, Anthropic is not untouched. Meta used its cash power to pull away engineers Joel Pobar and Anton Bakhtin last year. They have not confirmed the move on social media yet.

But some Anthropic staff would not even take a Meta interview. They said they do not want to talk to Zuckerberg at all.

OpenAI Lost 8 of 11 Cofounders

In 2015, the non profit OpenAI launched with Sam Altman and Elon Musk as co chairs. The founding team included Ilya Sutskever and other top AI researchers plus research engineers and scientists.

At launch, Ilya Sutskever was research director. Greg Brockman was chief technology officer. The other founding members were Trevor Blackwell, Vicki Cheung, Andrej Karpathy, Durk Kingma, John Schulman, Pamela Vagata, and Wojciech Zaremba. Sam Altman and Elon Musk served as co chairs.

Of the original 11 cofounders, only 3 remain at OpenAI today.

Sam Altman is now CEO. Greg Brockman is now chairman. Wojciech Zaremba leads the Codex and language teams.

Altman and Brockman both have AI backgrounds. Zaremba has an interesting story. He started a PhD under Yann LeCun but dropped out to become an OpenAI cofounder.

The rest scattered across tech. Some even sued OpenAI.

Elon Musk, the co chair and early investor, left OpenAI long ago and is now suing the company.

Trevor Blackwell is a robotics expert and was also a Y Combinator cofounder. He met Altman there. He worked at OpenAI until 2017 and stayed as a Y Combinator partner until 2020.

Vicki Cheung was a founding engineer at Duolingo. She got in touch with OpenAI when it was still in stealth mode. In a 2021 podcast about OpenAI early days, she said it was just her and Greg Brockman working from his apartment.

After two years leading infrastructure, she left OpenAI and joined Lyft as a software engineer in 2018.

In 2020 she left Lyft and started Gantry with another early OpenAI employee Josh Tobin.

In 2025 she became CTO at Matter Intelligence, a geospatial AI startup.

Pamela Vagata was the other woman on the OpenAI founding team. She started her tech career as a software engineer at Microsoft. Then she spent six years at Facebook, two of them as an AI research engineer.

Her exact role at OpenAI is not clear. Her LinkedIn does not list it.

Media reports say she later helped launch OpenAI core products and played a key role in early releases.

After OpenAI she joined Stripe in 2016 as one of its first machine learning engineers.

Later she started Pebblebed, a venture fund focused on AI investing, in 2021.

In 2015, Musk brought in Ilya Sutskever from Google Brain.

At Stanford, Ilya was Hinton top student. He called AlexNet one of the most important moments in deep learning.

Ilya became OpenAI research director and later chief scientist in 2018.

In May 2024, after the board drama, he left OpenAI to start Safe Superintelligence Inc, or SSI, an AI company focused on safety.

Before Ilya left, OpenAI lost Jan Leike in May 2024. He quit over safety concerns and said OpenAI had lost its safety focus.

In August 2024, cofounder John Schulman left OpenAI and joined Anthropic.

In September 2024, president Greg Brockman took a long leave. He returned three months later.

Even earlier, in 2018, Durk Kingma left OpenAI and joined Anthropic.

Andrej Karpathy has a more complex history with OpenAI.

He joined right after his PhD in 2015.

In 2017 he left for Tesla as AI director. He stayed until 2022.

In 2023 he came back to OpenAI.

In 2024 he left again to start his own company.

After the OpenAI board crisis, many top leaders left.

Several reports from former staff said OpenAI lost nearly half of its long term safety and AGI team in 2024. Names include Jan Hendrik Kirchner, Collin Burns, Jeffrey Wu, Jonathan Uesato, Steven Bills, Yuri Burda, and Todor Markov.

Multiple executives said Altman leadership style was the reason people left.

In 2025, Meta threw money at the problem and OpenAI lost more core staff. Here is a partial list of ChatGPT team departures.

Thinking Machines Lab Fires Its Own CTO

Mira Murati, former OpenAI CTO, started Thinking Machines Lab or TML. She announced that Barret Zoph was leaving.

Luke Metz and Sam Schoenholz also quit.

But just 58 minutes later, OpenAI app CEO Fidji Simo posted online to welcome Barret Zoph, Luke Metz, and Sam Schoenholz back.

Reports say at least 5 people have quit. The steady stream of departures has made TML staff nervous.

TML has about 100 people. Losing 5 is not huge. But Barret Zoph, Luke Metz, and Sam Schoenholz are founding members.

Rumors say 50 percent of the founding tech leaders have already left.

This human drama has hurt the hot AI startup. But the seeds were planted months ago.

SSI Cofounder Jumps to Meta

After leaving OpenAI in 2024, Ilya Sutskever started Safe Superintelligence with Daniel Gross and Daniel Levy.

Daniel Gross, Ilya Sutskever, and Daniel Levy.

But last summer, Zuckerberg hired Daniel Gross away with a massive offer. Gross joined Meta SuperIntelligence Lab.

Meta tried to buy SSI outright but failed. So Meta hired Gross and his longtime partner Nat Friedman instead.

They run an early stage venture fund called NFDG. Media reports say Meta will buy part of NFDG as part of the deal.

DeepMind Cofounder Mustafa Suleyman Leaves

In 2022, DeepMind cofounder Mustafa Suleyman left and joined venture firm Greylock Partners.

This ended his 8 year run at Google.

His last role at Google was VP of AI product and policy.

Google bought DeepMind in 2014 and Suleyman came along. He ran DeepMind applied AI.

Reports say he was put on leave in 2019 after staff accused him of bullying. Later that year he moved to Google headquarters.

Suleyman later said he messed up badly and still feels sorry for the pain he caused.

During his PhD, Demis Hassabis met Shane Legg. Mustafa Suleyman had been friends with Hassabis since childhood.

No one thought Suleyman would leave over workplace issues and break away from Hassabis.

In 2010, Demis Hassabis, Mustafa Suleyman, and Shane Legg started DeepMind in London.

Today Suleyman is CEO of Microsoft AI. Before that he cofounded and ran Inflection AI.

xAI Loses Cofounder to AI Safety Investing

xAI is very secretive. No one even knows the exact size of the founding team. Reports say about 12 people.

Earlier, xAI cofounder Igor Babuschkin left.

Early in his career, he led the DeepMind AlphaStar project for StarCraft. He saw firsthand how powerful reinforcement learning becomes at scale.

As frontier models gain more autonomy over longer time frames, they will grow more capable. This makes AI safety research critical.

He started Babuschkin Ventures to fund AI safety research and invest in startups that push human progress and explore space through AI and agent systems.

Reports say xAI general counsel Robert Keele quit in August 2025. He joined in May 2024 and handled the X merger and OpenAI lawsuits.

Keele called working with Musk the adventure of a lifetime. But he said the work hurt his family life. He loves his two young kids but does not see them enough. He cannot focus on both.

His replacement is Lily Lim, former head of privacy and IP.

xAI CFO Mike Liberatore also left. He started in April 2025 and quit by July. The former Airbnb exec helped raise 60 billion in debt and 50 billion in equity for AI data centers and the Grok platform.

His reason for leaving is unknown. But it came right after cofounder Igor Babuschkin left in August to start his AI safety fund.