Karpathy Why Suddenly Join Anthropic, Only to be Dario's 「-2」?
- Key Takeaway: OpenAI co-founder and former Tesla AI Director Andrej Karpathy announced he is joining Anthropic to build a team focused on accelerating pre-training research using the Claude model. This move marks a significant talent acquisition victory for Anthropic and signals that the evolutionary flywheel of "AI self-improvement" may be speeding up.
- Key Points:
- Karpathy joins Anthropic's pre-training team led by Nick Joseph, tasked with leveraging Claude itself to accelerate pre-training research, aiming to train better models with less compute.
- Karpathy's stature stems from his ability to define industry-conceptual paradigms (e.g., "Software 2.0", "Vibe Coding"), rather than sheer technical skill alone, acting as a bridge between theory and practice.
- This marks the third core figure moving from OpenAI to Anthropic within two years, following Jan Leike and John Schulman, intensifying the unidirectional talent flow and highlighting the appeal of Anthropic's "research quality first" approach.
- Anthropic's Mythos model has demonstrated "emergent" capabilities, autonomously discovering deep system vulnerabilities without specific training, proving that pre-training enhancements can yield capabilities beyond expectations.
- Karpathy's addition could turn Anthropic's strongest model into a tool for improving its own training, creating a "AI improving AI" flywheel. If successful, this could rewrite the current logic of the compute and data arms race.
Original Editor: Marco
Original Source: 新智元
At 11 PM on May 19th, Andrej Karpathy officially announced his join to Anthropic.
The weight of this name needs no explanation.
OpenAI co-founder, former Tesla AI Director, father of "Vibe Coding," one of the most influential AI educators globally.
His standing in the AI field is roughly equivalent to LeBron James in basketball – a headline-maker for any team he joins.
He posted only three sentences on X.
https://x.com/karpathy/status/2056753169888334312
The first sentence said the next few years at the frontier of LLMs would be "particularly formative." The third mentioned he still loves education. The middle sentence was the most crucial, just five words: "Return to R&D."
This marks the third core figure from the OpenAI camp to join Anthropic within two years.
It is also a person about to turn 40, accomplished, and financially free, actively choosing to become someone else's subordinate.
Why leave? Why Anthropic? And why did Anthropic want him so badly?
Behind each question lies a layer worth unpacking.
What He Will Do
Karpathy started working this week, joining Anthropic's pre-training team.
This team is led by Nick Joseph and is responsible for all large-scale training runs of Claude.
An Anthropic spokesperson confirmed to TechCrunch that Karpathy will form a new sub-team focused on using Claude itself to accelerate pre-training research.
Nick Joseph also provided context on X, stating: "He will build a team focused on using Claude to accelerate pre-training research itself."
https://x.com/nickevanjoseph/status/2056760504949842219
TechCrunch commented, "Karpathy is one of the few researchers capable of bridging the gap between LLM theory and large-scale training practice."
Axios characterized this move as a "major talent war victory for Anthropic."
Cybersecurity expert Chris Rohlf also announced joining Anthropic on the same day, following xAI founding member Ross Nordeen, who joined earlier in the month. The direction of the talent flow is becoming increasingly clear.
https://x.com/chrisrohlf/status/2056744653165092983
Data from Polymarket serves as corroborating market sentiment — traders price Anthropic's probability of having the best AI model by the end of June at 65%, compared to 4% for OpenAI.
https://polymarket.com/event/which-company-has-best-ai-model-end-of-june
Karpathy's joining further reinforces this assessment.
Karpathy, the Definitor
To grasp the significance of this addition, one must understand Karpathy's rarity.
His rarity does not lie in technical skill alone; there is a cohort of top-tier researchers.
His rarity lies in his ability to change the entire industry's understanding of something with a single word.
Born in Slovakia in 1986, he emigrated to Toronto, Canada at age 15.
During his undergraduate studies at the University of Toronto, he took Geoffrey Hinton's course and attended his reading group.
Hinton is the spiritual leader of the deep learning renaissance, a 2018 Turing Award winner, and a 2024 Nobel Prize in Physics laureate.
Karpathy was one of the first young minds ignited by this flame.
Afterwards, he studied under another legendary figure, Fei-Fei Li, at Stanford, creating the CS231n course during his PhD.
This course grew from 150 students in 2015 to 750 in 2017. All lecture materials were made publicly available online, becoming the first stop for countless engineers worldwide to self-study deep learning, and undeniably the premier course in computer vision.
In 2015, he became a founding research scientist at OpenAI.
In 2017, Elon Musk recruited him to Tesla as Senior Director of AI, driving autonomous driving towards a pure vision-based approach.
Musk faced immense pressure during this recruitment process.
https://www.cnbc.com/2026/05/19/anthropic-hires-openai-cofounder-andrej-karpathy-former-tesla-ai-lead.html
That same year, Karpathy published an article on Medium proposing the "Software 2.0" concept, arguing that neural network weights were the new code, datasets the new source code, and gradient descent the new compiler.
This framework reshaped the entire industry's understanding of "what programming is."
After leaving Tesla in 2022, he created the "Neural Networks: Zero to Hero" course series on YouTube, garnering over a million subscribers.
Concurrently, his open-source projects like micrograd, nanoGPT, and nanochat, with minimal code, precisely hit core concepts, earning the moniker "runnable textbooks."
In February 2025, he coined the term "Vibe Coding," which was selected as Word of the Year by Collins Dictionary.
https://x.com/karpathy/status/1886192184808149383
In June, during a speech at YC AI Startup School, he proposed the "Software 3.0" and "Decade of Agents" frameworks, becoming one of the most widely discussed AI talks of the year.
TIME named him to its "100 Most Influential People in AI" in 2024.
From Hinton to Fei-Fei Li, from Altman to Musk, he has been at the forefront at every juncture.
Yet his most enduring contributions are not any single product or paper, but the conceptual frameworks he left behind.
Software 2.0, Vibe Coding, LLM OS. These words changed how people think about AI.
Why Willing to be "-2"?
Karpathy's career has a clear trajectory; he has never chased titles.
He has been a student of Hinton and Fei-Fei Li, a colleague of Altman, and a direct report of Elon Musk.
In each of these experiences, his position in the organizational structure was high-level.
Now he joins Anthropic, reporting directly to Nick Joseph, Head of Pre-training.
Nick Joseph reports to Dario Amodei.
Karpathy is now positioned on the third tier of the organizational chart.
Nick Joseph is one of the 11 co-founders of Anthropic, previously working at Vicarious and OpenAI.
During his time at OpenAI, he worked on code models in the safety team. Realizing that fine-tuned GPT-3 could write code, he saw the potential for AI self-improvement, and left with the head of the safety team to found Anthropic.
His team trained the entire Claude model series, including Mythos.
The reason Karpathy is willing to work under Nick Joseph is simple: this position is closest to what he wants to do.
Looking back at his every career move, the driving force is the same: "Where is the biggest experiment right now?"
Going to Tesla in 2017 because autonomous driving was the biggest experimental ground for Software 2.0.
Leaving in 2022 because the architecture was set, leaving only engineering optimization.
Returning to OpenAI in 2023 because the explosion period following GPT-4's launch with ChatGPT was the most exciting frontier.
Founding Eureka Labs in 2024 to test the hypothesis of AI-native education.
Joining Anthropic in 2026 because the pre-training revolution of "using AI to study AI" is happening here.
Each departure was not due to dissatisfaction, but because the current position was no longer the site of the biggest experiment.
Why not return to OpenAI? The talent flow provides the answer.
Jan Leike, former head of alignment at OpenAI, joined Anthropic in May 2024.
OpenAI co-founder John Schulman followed in August of the same year.
Now it's Karpathy's turn.
Three people in two years, all moving in one direction, with no comparable reverse case.
OpenAI's strategic focus has shifted from pure research to platformization and acquisitions. The intervals between acquiring Chat.com, io Products, Windsurf, TBPN are getting shorter, and the amounts larger.
It is a company transforming into an "AI-era consumer giant."
For a researcher wanting to "return to R&D," Anthropic's path of "winning through research quality" holds greater appeal.
Why Anthropic Wanted Him So Much
Anthropic's recruitment motive can be broken down into several layers.
The most superficial layer is technical need.
No matter how large Anthropic's compute budget is, it cannot compare to OpenAI backed by Microsoft or Google with its TPUs.
In a pure compute arms race, Anthropic cannot win.
It must find a way to train better models with less compute.
"Using Claude to accelerate pre-training research" is this path, and Karpathy possesses the rare combination of deep pre-training theory, large-scale engineering experience, and intuition for AI-assisted research. Very few people have all three.
The next layer is the talent signal.
With three core OpenAI figures moving unidirectionally to Anthropic in two years, the narrative of "frontline researchers voting with their feet" is established.
Every addition at Karpathy's level lowers the psychological barrier for the next top-tier talent to join. Talent attracts talent; the flywheel spins itself.
Then there's the brand polish before an IPO.
Anthropic is reportedly in talks for a $30 billion funding round at a $900 billion valuation, with IPO preparations underway.
Karpathy is one of the most publicly recognizable technical figures in AI: a million-subscriber YouTube channel, a word-of-the-year coiner, the CLAUDE.md repository with 220k GitHub stars.
His name on Anthropic's employee roster provides an instant line for investment banks to include in the prospectus.
But the most intriguing layer might be something Anthropic didn't explicitly state as a recruitment motive, yet is destined to yield the greatest returns: Karpathy's ability to define paradigms.
Any technical exploration he undertakes at Anthropic will likely be discussed publicly by him — tweets, blog posts, YouTube videos.
When he names what is happening in his unique way, Anthropic naturally becomes the origin of that paradigm.
Hiring a top pre-training researcher comes with the industry's most influential technical narrator.
The Flywheel's Tipping Point
Placing this personnel change in a broader context, it marks a technological inflection point.
In April 2026, Anthrop
