Google to invest up to $40B in Anthropic in cash and compute

In a bold move, Google is set to invest as much as $40 billion in Anthropic, a rising player in the artificial intelligence sector, as reported by Bloomberg. The Alphabet subsidiary will kick off this initiative with an immediate $10 billion investment, valuing Anthropic at a staggering $350 billion. An additional $30 billion will be available contingent upon Anthropic meeting specific performance benchmarks. This significant investment follows the recent launch of Anthropic's latest AI model, Mythos, which has been made available to a select group of partners. Mythos is touted as the company's most advanced model to date, boasting considerable applications in cybersecurity. However, Anthropic is exercising caution by limiting wider access due to concerns over potential misuse, while it collaborates with chosen organizations to mitigate associated risks. Notably, the model has already been reported to have fallen into unauthorized hands. As the competition in the AI landscape intensifies, access to the necessary computing power for training and deploying these sophisticated systems has become crucial. OpenAI has been proactive in securing substantial compute capacity through extensive agreements with cloud providers and chip manufacturers, including a recent collaboration with Cerebras. Anthropic is also in the midst of its own urgent efforts to bolster its infrastructure. Recently, the company has faced backlash over usage limits of its AI model, Claude, prompting it to forge new partnerships, including a deal with CoreWeave for enhanced data center capabilities. Additionally, Anthropic has secured a $5 billion investment from Amazon, as part of a comprehensive agreement that could see the firm investing up to $100 billion for around 5 gigawatts of computing power in the future. Despite being a direct competitor in AI model development, Google plays a vital role as an infrastructure provider for Anthropic. The latter relies extensively on Google Cloud for essential computing resources, including access to advanced tensor processing units (TPUs), which are highly sought after for AI workloads. Anthropic’s partnership with Google dates back further, having recently announced a collaboration with Google and Broadcom to access multiple gigawatts of TPU-based computing capacity starting in 2027, with projections indicating a capacity of 3.5 gigawatts. This latest investment from Google will enhance that partnership, with an additional 5 gigawatts of capacity promised over the next five years, and potential for further scalability. Anthropic's valuation has seen significant interest, climbing to $350 billion as of February, with investors now eyeing a figure upwards of $800 billion. The company is also reportedly contemplating an initial public offering (IPO) as early as October.