Google has announced plans to invest up to AUD 40 billion in the AI research company Anthropic, a strategic move expected to bolster the start-up’s computing capabilities. The investment will commence with an immediate AUD 10 billion, valuing Anthropic at AUD 350 billion, followed by an additional AUD 30 billion contingent on the company meeting specific performance objectives.
This investment follows the recent release of Anthropic’s latest AI model, Mythos, to a select group of partners. Anthropic claims Mythos is their most potent model yet, featuring significant applications in cybersecurity. Due to its potential for misuse, Anthropic has limited its access while collaborating with a few organizations to navigate associated risks; however, reports have indicated that the model has already been accessed by unauthorised users. Additionally, running Mythos at scale may incur substantial costs.
The ongoing AI competition is increasingly driven by the need for robust computing resources essential for training and deployment. OpenAI has been proactive in securing such capabilities through extensive multi-billion dollar agreements with various cloud providers, chip manufacturers, and energy suppliers, including an expanded partnership with Cerebras for chip technology.
Anthropic, meanwhile, has encountered challenges of its own, receiving significant user feedback regarding limitations on its Claude AI system. In response, the company has forged new infrastructure agreements, including a recent partnership with CoreWeave for expanded data centre capacity and a USD 5 billion investment from Amazon. This arrangement is part of a broader plan, potentially amounting to USD 100 billion, aimed at securing around 5 gigawatts of computing capacity over time.
While Google is a competitor in the AI landscape, it also serves as a critical infrastructure provider for Anthropic. The latter heavily depends on Google Cloud for computing resources, particularly its proprietary tensor processing units (TPUs), which excel at handling AI workloads and are considered a prime alternative to Nvidia’s sought-after processors.
The partnership between Anthropic and Google is not a new development; earlier this month, they announced a collaboration with Broadcom, another chipmaker, to access additional TPU-based computing capacity starting in 2027, reportedly up to 3.5 gigawatts. Google’s investment will enhance this arrangement, offering Anthropic an extra 5 gigawatts of capacity over the next five years, with flexibility for further scaling.
Anthropic’s valuation, once pegged at AUD 350 billion a few months ago, has attracted significant interest from investors, with speculation that it could reach AUD 800 billion or more. Furthermore, the company is actively considering a potential initial public offering (IPO) as early as October.
Fanpage: TechArena.au
Watch more about AI – Artificial Intelligence
