Just two months ago Amazon announced a astronomical investment of $50 billion in OpenAI. Today he made a movement very similar to the announce which will invest $5 billion in Anthropic and could invest an additional $20 billion “tied to certain commercial milestones) in the future. There are counterparts and some circular financing, of course, but also a clear pattern: Amazon has no winning horse in the AI race, so it is betting on its competitors.
More circular financing. Amazon now has alliances in the form of active investment with the two leading AI companies in the world. In return, both OpenAI and Anthropic commit to huge spending on their services on AWS. There is a lot of circular financing here: me I lend you the money so that you spend it on me. Those houses of cards that OpenAI and Anthropic are building have clear risks, but the industry is totally immersed in that maelstrom.
Analysts warn. There are concerned analysts here and others who defend this type of agreement. M. Mohan asked in X why regulators are not on top of these types of financially dangerous agreements: the domino effect if OpenAI or Anthropic fall could be terrible. For others like the well-known Jim Cramer this is not circular financing. According to him, circular agreements are designed to inflate profits, and here no one’s profits are being inflated. Their argument is that Amazon has real computing, Anthropic needs real computing, and the value of the investment is genuine.
History repeats itself. The same debate occurred in January with OpenAI, and the conclusion was the same then: the image of circular financing is there but it does not necessarily imply fraud, it implies that Amazon has found a way to monetize the AI craze without betting on any particular model. Or for the two who seem to be winning the race. But everyone is doing it.
The numbers of the agreement with Anthropic. Amazon puts up $5 billion immediately, taking advantage of the company’s current valuation of $380 billion. It is also committed to investing up to an additional $20 billion linked to “certain commercial milestones” that have not been specified. In exchange, Anthropic commits to using Amazon technology, and specifically its Trainium and Graviton chips, for the next decade. No less than 5 GW of computing capacity is secured, which is more or less the capacity consumed by New York City.
This is perfect for Anthropic. He Anthropic statement about the agreement contains an interesting paragraph. In it, the company admits that the demand for AI by companies, developers and users is generating “inevitable tension” in its infrastructure. Or what is the same: they can’t do everything, so they are resorting to measures that “penalize” the excessive use of their AI models. They restrict session limits during peak hours, change the pricing model in companies to a “pay as you go”, or change the level of effort of their models and they sign up for token inflation. The agreement with Amazon makes it possible to mitigate the problem of computing shortages.
The race for gigawatts. The truth is that Anthropic has been moving for months to try to avoid more and more problems with the computing capacity they can access. In a few weeks we have seen how Amazon’s 5 GW have been secured and also “multiple gigawatts” computing teams contracted with Google and Broadcom.
What Amazon is actually building. Viewed as a whole, Amazon’s strategy is simple and elegant. You don’t need to win the AI modeling race, which is unpredictable and extraordinarily expensive. It only needs that whoever wins it depends on it and its infrastructure. By investing at the same time in two rivals like Anthropic and OpenAI and securing massive spending contracts from both, it achieves something striking. Turn uncertainty into an asset: it doesn’t matter who wins, because she will end up getting paid. This also reinforces the relevance of its Trainium and Graviton chips, something that validates its commitment to its own chips.
{“videoId”:”xa4n2g8″,”autoplay”:false,”title”:”An initiative to secure the world’s software ｜ Project Glasswing”, “tag”:””, “duration”:”349″}
Win-Win. The agreement seems perfect for both parties. Amazon ensures, as we say, consumption in its infrastructure for the next ten years, and Anthropic achieves an investment that increases its market value again. The same happens with OpenAI, and in both cases these agreements and financial support only reinforce expectations about their imminent IPOs.
Image | Fortune Brainstorm TECH
In Xataka | OpenAI and Anthropic have proposed the impossible: lose $85 billion in one year and survive
(function() { window._JS_MODULES = window._JS_MODULES || {}; var headElement = document.getElementsByTagName(‘head’)(0); if (_JS_MODULES.instagram) { var instagramScript = document.createElement(‘script’); instagramScript.src=”https://platform.instagram.com/en_US/embeds.js”; instagramScript.async = true; instagramScript.defer = true; headElement.appendChild(instagramScript);
–
The news
Amazon is clear about its strategy for the AI war: if you can’t beat your enemy, invest in them
was originally published in
Xataka
by
Javier Pastor
.
GIPHY App Key not set. Please check settings