AWS shifts focus to LLMs, generative AI as growth continues to taper

Amazon’s cloud computing division, AWS, is shifting its focus towards large language models (LLMs) and generative AI-based offerings as it continues to see a downward spiral in overall revenue growth.

Amazon Web Services (AWS) has posted 16% year-on-year growth for the first quarter of fiscal year 2023 on the back of revenue of $21.4 billion. However, this revenue growth is slower compared to the 20%, 27.5%, and 33% growth seen in the fourth quarter, third quarter, and second quarter of 2022, respectively. 

The slowdown in growth, according to top executives of the company, can be attributed to enterprises optimizing cloud spend due to uncertain macroeconomic conditions.

“Given the ongoing economic uncertainty, customers of all sizes in all industries continue to look for cost savings across their businesses, similar to what you’ve seen us doing at Amazon. As expected, customers continue to evaluate ways to optimize their cloud spending in response to these tough economic conditions in the first quarter,” Brian Olsavsky, chief financial officer at AWS, said during an earnings call.

“We are seeing these optimizations continue into the second quarter with April revenue growth rates about 500 basis points lower than what we saw in Q1,” Olsavsky added.

In response to the trend, Olsavsky said that AWS’ sales and support teams have continued to spend much of their time helping customers optimize their spending to help them “better weather this uncertain economy.”

However, AWS top executives remained bullish on the growth perspectives of the division, citing the opportunity around the conversion of on-premises workloads.

“The new customer pipeline looks strong. The set of ongoing migrations of workloads to AWS is strong. The product innovation and delivery is rapid and compelling. And people sometimes forget that 90-plus percent of global IT spend is still on-premises,” Amazon CEO Andy Jassy said during the call.

AWS shifts focus to generative AI

In addition, Jassy hinted that the company’s major chunk of cloud business will come from machine learning requirements.

“And in my opinion, few folks appreciate how much new cloud business will happen over the next several years from the pending deluge of machine learning that’s coming,” Jassy said.

The company has already been making capital expenditure adjustments to reroute funds toward the improvement of large language models and generative AI capabilities.

Amazon has been bringing down spending at its fulfillment and transportation divisions year-on-year and has decided to route the savings to AWS to be invested in infrastructure and large language models, Olsavsky said.

AWS’ strategy, according to Jassy, is to target revenue generation by providing compute resources, training capabilities and applications for generative AI and large language models.

“I would say that there’s three macro areas in this space. If you think about maybe the bottom layer here, is that all of the large language models are going to run on compute. And the key to that compute is going to be the chip that’s in that compute,” Jassy said, adding that the company has already launched its Trainium chips and accelerators for memory-intensive tasks, ideal for AI-heavy workloads.

The second layer, according to Jassy, would be to train foundation models, and AWS has just launched its Amazon Bedrock service that provides multiple foundation models designed to allow companies to customize and create their own generative AI applications, including programs for general commercial use.

The third macro area or layer will be offering applications for developers, such as ChatGPT-enabled Copilot from Microsoft-owned GitHub, said Jassy, citing Amazon CodeWhisperer.

“Every single one of our businesses inside Amazon are building on top of large language models to reinvent our customer experiences, and you’ll see it in every single one of our businesses, stores, advertising, devices, entertainment,” the chief executive added.

Other investments in the first quarter by the cloud computing division include a new region in Malaysia and a second region in Australia.