You may already know that Amazon had a massive Q2. The e-commerce company beat guidance on both net sales and operating income.
Its overall revenue was up 11% YoY and its ad revenue was up 22% YoY, even in a relatively quiet quarter. (Amazon’s Q2 notably does not include Prime Day.)
Similarly, AWS made up 68% of Amazon’s entire operating income, despite it only accounting for 16% of the company’s net sales.
In other words, AWS is by far the most profitable segment of the business. The road forward looks good for anyone utilizing cloud services.
But those top-line numbers only tell one piece of the story. Our team pored over the earnings report and read the transcript.
What we found: Amazon had some fascinating things to say about AI.
Here’s what you need to know:
“Every” Amazon team is working on generative AI
Amazon is not shy about the massive amount of investment it is pouring into Large Language Models (LLMs).
In the earnings call, Amazon CEO Andy Jassy broke down his approach to integrating LLMs—the group of AI tools, like ChatGPT, that create and analyze content—into the Amazon stack.
Jassy said that its AI strategy had three main parts:
1) Simplified training of LLMs
2) Offering LLMs as a service
3) Running applications on top of LLMs.
Let’s take these one by one.
Training. Essentially, Amazon wants to lower the cost of training an AI model. It created a line of chips, called Trainium, that it hopes will make it easier to refine a custom AI model.
LLMs as a service. What if you’re a big company, and you want to create your own AI model? Amazon plans to provide big companies with the tools to make their own LLMs, using pre-developed infrastructure. It already has partnerships with 3M, Ryanair, Travelers, and more.
Running applications on top of LLMs. This is the part of the story that brands and agencies are thinking about most. Running applications on top of LLMs means integrating AI into the functionality of the Amazon platform itself.
In AWS, for instance, it’s already happening: Amazon has created a tool called Amazon CodeWhisperer that recommends code as developers build.
It isn’t hard to imagine a similar tool coming to Amazon in the near future. As Jassy noted, “Every one of our teams is working on building generative AI applications.”
What could all this mean for brands and agencies?
We don’t have the full story yet, but we’ve already seen some glimpses of what’s in the works. Namely, Amazon has already tested summarizing product reviews with AI and generative images and videos for ads with AI.
We are probably in store for much more AI on the marketplace in the near future.
In a recent article, we speculated that perhaps LLMs could soon be used to tell you why your ad revenue changed. Soon, a LLM might be able to come up with a narrative description explaining what happened or what changed in your advertising account.
