OpenAI rival Anthropic PBC is seeking to raise up to $2 billion in new funding, multiple publications reported today.

CNBC and the Wall Street Journal cited sources as saying that investment could value the company at $60 billion. That’s more than three times the valuation Anthropic received last year in a funding round led by Menlo Ventures. Lightspeed Venture Partners is said to be leading the company’s latest raise. 

Anthropic develops a family of large language models called Claude. It released the most advanced model in the series, Claude 3.5 Sonnet, in June. Two months later, the company rolled out an upgraded version of the LLM that can perform actions in applications on the user’s behalf.

The latest Claude 3.5 Sonnet release scored 49% on SWE-bench Verified, a benchmark that measures neural networks’ programming skills. That’s a nearly 16% improvement over the original version that debuted in June. Claude 3.5 Sonnet is available alongside a less capable but more cost-efficient model, Claude 3.5 Haiku, that scored 40.6% on SWE-bench Verified.

Anthropic offers its LLMs through an application programming interface. The company also provides a ChatGPT-like chatbot called Claude Enterprise. The service can generate code, create sales forecasts and perform other business tasks based on datasets provided by the user. 

CNBC cited a source as saying that Anthropic currently generates $875 million in annualized recurring revenue, mainly from deals with enterprises. This past September, Reuters reported that OpenAI was on track to to end 2024 with $3.7 billion in revenue and expects to reach $11.6 billion this year. CNBC’s source didn’t specify how fast Anthropic is growing.

When it released the latest version of Claude 3.5 Sonnet, Anthropic revealed plans to launch an even more capable model called Claude 3.5 Opus in the future. The new funding that it’s raising could potentially help support the development effort. 

OpenAI’s newest reasoning models are based on a machine learning approach known as test-time compute. It boosts an LLM’s response quality by increasing the amount of hardware used to process prompts, which in turn raises inference costs. If Anthropic decides to implement test-time compute in its future LLMs, the $2 billion it’s raising could make it easier to shoulder the associated costs.

The reported funding round follows three earlier multibillion-dollar investments.

In 2023, Anthropic secured an investment worth up to $2 billion from Google LLC. Later that year, Amazon.com Inc. bought a stake in the LLM developer for $4 billion. AWS doubled its investment this past November as part of an agreement that also saw it became Anthropic’s primary partner of AI training infrastructure.

OpenAI closed a $6.6 billion funding round of its own last year at a $157 million valuation. It also secured a $4 billion line of credit from a group of banks. Cohere Inc., another major player in the general-purpose LLM segment, earlier raised $500 million from Advanced Micro Devices Inc. and other investors. 

Image: Anthropic