Low prices. A wide selection. It’s hard to beat the convenience.
These have been important keys to Amazon’s dominance in e-commerce over the past two decades.
Now, the tech titan looks to be doubling down on that winning playbook, borrowing parts of it to push a new AI strategy from it A division of Amazon Web Services which is focused partly on low prices and wide selection.
To be clear, it will be a long time before the business world can appreciate the effectiveness and financial sustainability of Amazon’s approach. But if successful, the game plan will go a long way toward assuaging critics who say Amazon is playing catch-up in the AI wars, and cement the company’s future position as one of the world’s most powerful and influential tech corporations. the following decades.
At its flagship AWS Re:Invent conference in Las Vegas this week, Amazon executives unveiled important parts of its AI strategy. One key element is a new set of proprietary base models known as FM or LLM, dubbed Nova, which can handle text, image and video requests respectively.
The introduction of a new class of Amazon’s own artificial intelligence models may at first glance be confusing, since the company already invested $8 billion in Anthropiccreator of the popular Claude family of AI generation models. But like my colleague Sharon Goldman recently notedAmazon is betting that “there will never be one tool” or AI model to rule them all. In short, Amazon believes businesses will want to choose models, whether it’s from Amazon, Anthropic, or other tech giants like Meta.
Amazon actually started pushing this idea of offering a choice of AI models through a single API to business customers when its AWS division introduced a service called Amazon Bedrock last year. Through Bedrock, business customers could choose from a relatively limited selection of AI models, but nonetheless a choice to train for their own needs and serve as the basis for their own AI applications.
On Wednesday, Amazon doubled down on the strategy, announcing the Bedrock Marketplace, which offers a total of 100 AI models. LLMs on the market come from many different companies, some of which are designed for specialized purposes.
“Finding and evaluating these models can be difficult and expensive,” Amazon said in a blog post announcing the market. “You need to expose them to different services, build abstractions to use in your applications, and create complex layers of security and management. The Amazon Bedrock Marketplace addresses these issues by providing a single interface to access both specialized and generic (baseline) models.”
Looking back at Amazon’s e-commerce business, the core part is Amazon Marketplace, where hundreds of thousands foreign merchants list products for sale, which account for 60% of all items sold through Amazon. Amazon supplements the selection of these marketplaces by also selling its own inventory, including under its own brands, when a certain product category or price point is not filled by Amazon’s marketplace sellers or brand partners.
Similarly, Amazon offers businesses an enterprise version of an AI-powered marketplace, with choices that can only be expected to expand in the future. (It’s also worth noting that Amazon’s core business, AWS, offers a marketplace of more than 10,000 software tools spanning categories from cybersecurity to data analytics.)
Low prices have also been another hallmark of Amazon’s retail business. Amazon is aggressively price-matching other retailers and running two giant discount promotions that attract big spenders and new Prime customers. (The FTC, in its antitrust suit against Amazon, alleged that the e-commerce giant was artificially inflating consumer prices online by penalizing merchants who undercut other retailers, but that’s a topic for another day.)
and of course Amazon CEO Andy JassyThe presentation of the new Nova AI models started with the price: “75% more cost-effective” was the first feature mentioned.
Simon Willison, an independent AI researcher, ran a quick test and agreed, writing on social networking app Bluesky that “Amazon’s price and performance (are) competitive Google Gemini family, which means they are _really_ inexpensive.’
“I think with this release, Amazon has taken its place among the top model suppliers,” Willison added. “Maybe we need a new FAANG acronym that covers OpenAI, Anthropic, Google, Meta and Amazon. I like GAMOA.”
Amazon executives won’t care what the new acronym is called, as long as they earn a place in it. If they do, the company’s longstanding hallmarks of low prices and selection will likely be the main reasons.
Are you a current or former Amazon or AWS employee with thoughts on this topic or advice to share? Contact Jason Del Rey at jason.delrey@fortune.com, jasondelrey@protonmail.comor through the messaging apps Signal and WhatsApp at 917-655-4267. You can also send him a message is LinkedIn or at @delrey on X.
