According to Bloomberg, Google is set to invest as much as $40 billion in Anthropic, addressing the AI company’s expanding computing requirements. The subsidiary of Alphabet is pledging an immediate investment of $10 billion, valuing Anthropic at $350 billion, with an additional $30 billion contingent on the company’s achievement of specific performance metrics, as stated by Anthropic. 

This investment commitment follows the release of Anthropic’s newest model, Mythos, to a select group of partners this month. Anthropic claims that Mythos represents the firm’s most advanced model to date and possesses considerable applications in cybersecurity. Due to the risk of improper use, Anthropic has limited wider access while collaborating with certain organizations to assess and mitigate these dangers — although the model has already been acquired by unauthorized parties. It is also expected to be costly to operate at a larger scale. 

The competition in AI is increasingly characterized by access to the computational power necessary for training and deploying these systems. OpenAI has aggressively pursued this capacity through a network of multi-hundred-billion-dollar agreements with cloud providers, chip manufacturers, and energy sectors, including a recently expanded agreement with chip producer Cerebras this month. 

Anthropic has been navigating its own challenges. The company has encountered significant criticism regarding limits on the use of its Claude model in recent weeks and has reacted with a series of infrastructure partnerships. Earlier this month, Anthropic forged an agreement with cloud service provider CoreWeave for data center resources. Additionally, this week, it secured a further $5 billion investment from Amazon as part of a broader understanding in which Anthropic aims to invest up to $100 billion for approximately 5 gigawatts of computational power over time. 

While Google serves as a direct competitor in AI models, it is also a crucial supplier of infrastructure for Anthropic. Anthropic depends significantly on Google Cloud for chips and infrastructure, including access to Google’s tensor processing units (TPUs), which are specialized chips intended for AI tasks and regarded as top alternatives to Nvidia’s highly sought-after processors.

Anthropic’s connection with Google existed before this week’s announcement. Earlier this month, Anthropic disclosed a partnership with Google and chip manufacturer Broadcom, which engineers custom AI chips for Google, to access several gigawatts of TPU-based computing capacity starting in 2027; a following Broadcom securities filing stated that figure as 3.5 gigawatts.

The new investment from Google enhances that collaboration, with Google Cloud now set to provide an additional 5 gigawatts of capacity over the next five years, allowing for further expansion.

As of February, Anthropic’s valuation was reported at $350 billion; since then, investors have shown strong interest in valuing the company at $800 billion or more, according to Bloomberg. The firm is also said to be contemplating an IPO as early as October.