Disclaimer: I don't have any insider information about any of the companies, I just connect the dots according to my interpretation.

Reminder about Anthropic: This is one of Open AI and ChatGPT's biggest competitors, some would say it is number 2, and the gap between them is not large. A two-year-old company, founded by OpenAI veterans, has so far raised hundreds of millions from investors, including about 400 million from Google six months ago. Anthropic's chatbot is called Claude, and among Claude's great advantages: it works great in Hebrew and has the ability to read documents (without premium plans), you can add it for free to Slack (recommended!) and it has a particularly large "memory" or "context window" of about 75,000 words 3 times that of ChatGPT.

Amazon does not need to be presented, and yet, in the context of the deal, two data points should be paid attention to:

1. Although Amazon is a retail giant, 70% of Amazon's profits come from the cloud business, AWS (!)

2. Amazon does not currently have an applicative offer for its own GenAI, and this puts it in a very uncomfortable position vis-à-vis competing cloud giants Microsoft (ChatGPT) and Google (Bard)

These facts do not go unnoticed by investors and influence, among other things, Amazon's coverage and stock price. Amazon has deep pockets and a clear need, so how come it gets crumbs ("minority holdings" in nice language) for $4 billion, which was Anthropic's value more or less in its last financing round six months ago? What's going on?! The answer is that Anthropic knows Amazon desperately needs it and is playing hard-to-get. I assume that Anthropic would have had no trouble raising additional funding, but for $4 billion it would have had to give a lot of equity, even if it doubled or tripled its value in the last six months (I doubt it), it still has to give more than a "minority holding" to get such a sum. And why does Anthropic need money? Mainly to fund computing resources for training and using its models, something Amazon has in abundance... And now we can begin to understand what really happened here.

So what does the deal really look like and what does each side gain?

Anthropic is getting what it needs - access to Amazon's endless processing resources on its AWS infrastructure - how much is infinity? I guess it's... Up to a valuation of $4 billion, let's say. That should be enough for Anthropic for a long time. And the price? Equity crumbs, the availability of its model to Amazon customers and a lot of PR. She didn't even commit to exclusivity. Excellent deal for her.Amazon mostly gets bragging rights, and tries to make the most of it. Essentially, Anthropic will install its language model on Amazon's Badrock platform and it will be available to developers through Amazon's infrastructure - this is the model Amazon needs to present an alternative to Google and Microsoft on AWS. In addition, Amazon is making a PR spin on its artificial intelligence processors (built on technology acquired from a former Israeli company, Annapurna Labs, in 2015) whose performance is inferior to Nvidia, and Amazon is trying to market them through the deal with some tailwind from Anthropic.

At the end of the day, it looks less like a huge investment and more like a Barter deal, certainly less binding than the deep connection between Microsoft and OpenAI.