Anthropic makes biggest case as AI alpha with latest hire

Andrej Karpathy, who was part of the founding members of OpenAI and most recently, was the director of artificial intelligence at Tesla, announced on Tuesday, May 19, that he had joined Anthropic, telling followers on X that “the next few years at the frontier of LLMs will be especially formative.”

For many, the move has brought back to the fore the ongoing debate about which AI firm is currently leading in terms of adoption, especially when a person of Karpathy’s caliber moves to a platform that is a rival to a company he cofounded. 

Who is the new big-name hire at Anthropic?

Karpathy is among the most recognizable researchers in the field of artificial intelligence. He co-founded OpenAI before leaving in 2017 to lead computer vision for Tesla’s Autopilot and Full Self-Driving programs. 

In 2022, he departed Tesla, returning briefly to OpenAI, and then founded the AI education start-up Eureka Labs.

Karpathy will be building a team focused on using Claude to accelerate pretraining research, the work that gives a model its core knowledge and capabilities. It is reported that he will be working with pre-training lead Nick Joseph. 

He wrote on X, “I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.”

Has Anthropic jumped to the front of the AI ace?

Researchers capable of advancing the frontier are seen as a far scarcer resource than capital or chips, and the contest for that small pool is as decisive as any funding round. 

Currently, Anthropic seems to be poaching heavyweights across various departments, with cybersecurity veteran Chris Rohlf, who used to work at Meta, recently joining its frontier red team.

Karpathy also follows Ross Nordeen, a founding member of Elon Musk’s xAI, who joined Anthropic earlier this month, the same day the company struck a deal with SpaceX to rent compute infrastructure at xAI’s Colossus data center in Memphis. The talent is moving one way.

So are the business metrics. The latest Ramp AI Index found that Anthropic overtook OpenAI in corporate adoption for the first time, reaching 34.4% of businesses in April as OpenAI slipped to 32.3%. 

Over the past year, Anthropic has quadrupled its business adoption while OpenAI’s grew by 0.3%. Anthropic is also reportedly raising fresh capital at a valuation approaching $950 billion, ahead of OpenAI’s most recent $852 billion, on the strength of growing revenue, the Claude Code developer tool and the cybersecurity-focused Mythos model that has drawn attention in Washington.

Is the alpha case airtight, or is there a catch?

Ramp, whose own data may be used to reference adoption claims, also pointed out that Anthropic’s incentives are misaligned with customers, since the company earns more as businesses consume more tokens. 

Claude has faced complaints over outages and rate limits in recent weeks, and that a recent model change roughly triples token costs for prompts containing an image.

Some of the fastest-growing vendors on Ramp’s platform last month were inference services offering cheap, open-source models, and OpenAI’s Codex remains a low-cost, low-switching-cost alternative.

The hire, then, does not end the race. What it does is shift the burden of proof. 

A year ago, OpenAI was the unquestioned reference point against which every rival was measured. Today, the question is no longer whether anyone can catch OpenAI but whether OpenAI can hold its lead.