The race among major cloud providers to seal their place in the AI race continues with Amazon announcing it will invest up to $4 billion into Anthropic, the AI startup based in San Francisco.

Integral to the deal is early access for Amazon’s cloud employees and customers to Anthropic’s AI technology, which the cloud provider hopes to integrate into their business and offerings.

For their part, Anthropic agreed to rely Amazon’s cloud services (AWS) as their primary cloud provider to develop and train their models, specifically relying on their AWS Trainium and Inferentia chips, with the potential to collaborate on these chips further.

The deal will also lock in Anthropic to provide AWS customers with access to future generations of its foundation models via Amazon Bedrock, AWS’s manged service providing access to foundation models.

The $4bn investment means that Amazon will have a “minority” ownership position in the company, though the actual percentage has not been disclosed.

As reported by Reuters, the initial investment will amount to $1.25bn, with either company having the ability to initiate the other $2.75bn in funding.

“By significantly expanding our partnership, we can unlock new possibilities for organizations of all sizes, as they deploy Anthropic’s safe, state-of-the-art AI systems together with AWS’s leading cloud technology,” Dario Amodei, co-founder and CEO of Anthropic said.

AWS is currently topping the cloud market, capturing just shy of a third (32%), with Microsoft Azure (23%) coming in second, and Google’s Cloud trailing behind at 10%.

Ties to Google? 

The news comes months after Anthropic announced it had selected Google Cloud as its cloud provider back in February, stating that the companies will co-develop AI computing systems” in the announcement.

Further, Google invested about $300m in Anthropic, which earned it a stake of 10% in the company.

It would appear that this deal is continuing.

Anthropic’s work with Google was initially compared to the relationship between Microsoft and OpenAI. However, it seems that Google has not worked to integrate Anthropic’s AI models – its most famous one being Claude, a LLM comparable to ChatGPT – into its systems as it has its own AI models for this, such as Bard.

The partnerships seems to be leveraging Anthropic’s models to instead further Google’s cloud capabilities.

Ultimately, this deal could see Anthropic working on two different competing cloud services, but the deal also hinges on Amazon’s AI foundation model platform, Amazon Bedrock.

Amazon Bedrock is a service that allows companies to use a range of foundational AI models to create different applications tailored to their business. The platform has already attracted thousands of users, including Sony, Ryanair, and LonelyPlanet.

Now, users of Amazon Bedrock will gain access to Anthropic’s models, which will allow them to customise their AI with Claude 2, a model which can analyse and respond to long prompts like business and legal documents.

AI and the Cloud

Amazon may be winning the cloud game, but it is somewhat late to the AI arena. While is announced Amazon Bedrock in April, it has so far not made any major dealings with AI companies or developed its own AI foundation model.

Recommended reading

• Amazon, Microsoft and Google Face UK Cloud Dominance Probe
• Amazon Enters Generative AI Market With ‘Bedrock’
• Microsoft and AWS Push Back Against Cloud Market Investigation

Microsoft, which holds about a fifth of the global cloud market, currently owns the biggest name in generative AI, OpenAI, the company responsible for ChatGPT.

Google has also released its own AI model, Bard, onto the market, as it plans to integrate this LLM into its search engine and other products.

All companies, however, will need to ensure their cloud services are AI capable in order to maintain their market dominance.

Amazon’s and Google’s approach to Anthropic, which has yet to achieve the fame of OpenAI, by using it to leverage their cloud capabilities and develop new AI-capable chips, may prove to be a beneficial background move.

Companies like Broadcom, Nvidia, and Arm are already trying to leverage their place in the AI-chip race, and cloud companies appear to be doing the same.