AWS Optimistic About Growth Despite Supply Chain, Power Constraints

‘AWS is a reasonably large business by most folks’ standards, and though we expect growth will be lumpy over the next few years as enterprise adoption cycles, capacity considerations, and technology advancements impact timing, it's hard to overstate how optimistic we are about what lies ahead for AWS customers and business,’ Amazon CEO Andy Jassy told analysts during his company’s fiscal 2024 conference call.

Amazon Web Services’ business is growing quickly as the IT world continues to embrace AI, but it could be growing faster except for some significant constraints, Amazon CEO Andy Jassy told financial analysts.

Jassy, in his prepared remarks Thursday during Amazon’s fiscal 2024 fourth quarter conference call, said that AWS saw its fourth quarter revenue grow 19 percent year-over-year and now has a $115 billion annualized revenue run rate.

“AWS is a reasonably large business by most folks’ standards, and though we expect growth will be lumpy over the next few years as enterprise adoption cycles, capacity considerations, and technology advancements impact timing, it's hard to overstate how optimistic we are about what lies ahead for AWS customers and business,” he said.

That optimism stems from what Jassy said he sees when looking at the future.

“While it may be hard for some to fathom a world where virtually every app has generative AI infused in it, with inference being a core building block just like compute, storage and database, and most companies having their own agents that accomplish various tasks and interact with one another, this is the world we're thinking about all the time,” he said. “And we continue to believe that this world will mostly be built on top of the cloud, with the largest portion of it on AWS.”

To realize this future requires several powerful capabilities, Jassy said.

At the bottom layer are compelling chips, the key compute ingredient driving training and inference, Jassy said.

“Most AI compute has been driven by Nvidia chips and we obviously have a deep partnership with Nvidia and will for as long as we can see into the future,” he said. “However, there aren't that many generative AI applications of large scale yet, and when you get there, as we have with apps like Alexa and Rufus, cost can get steep quickly. Customers want better price-performance, which is why we built our own custom AI silicon.”

Amazon’s latest silicon is Trainium2, launched in December at its AWS Re:Invent conference. Jassy said AWS EC2 instances on Trainium2 typically offer 30 percent to 40 percent better price performance than other current GPU-powered instances available, which is very compelling at scale, he said.

“We're collaborating with Anthropic to build Project Rainier, a cluster of Trainium2 ultra servers containing hundreds of thousands of Trainium2 chips,” he said. “This cluster is going to be five times the number of Exaflops as a cluster that Anthropic used to train their current leading set of cloud models. We're already hard at work on Trainium3, which we expect to preview late in ’25, and defining Trainium4 thereafter.”

Building top-performing chips with leading price performance has become a core strength of AWS, Jassy said. “It’s something unique to AWS relative to other competing cloud providers,” he said.

Another key AWS component is services that make it easier for model builders to construct their models. He cited as an example Amazon SageMaker AI, which he said is the go-to service for AI model builders to manage their AI data, build models, experiment, and deploy those models. Amazon at re:Invent unveiled SageMaker’s enhanced HyperPod technology, which automatically splits training workloads across many AI accelerators, with new capabilities including the ability to manage costs at a cluster level and prioritize which workloads should receive capacity when budgets are reached.

For developers looking to leverage frontier models to build GenAI apps, Amazon Bedrock offers features that make it easy to build a high-quality generative AI application, Jassy said. Amazon at re:Invent said over 100 popular emerging models now work on Bedrock and Sagemaker.

“We also just added DeepSeek’s R1 models to Bedrock and SageMaker,” he said. “And additionally, we delivered several compelling new Bedrock features at re:Invent, including prompt caching, intelligent prompt routing, and model distillation, all of which help customers achieve lower cost and latency in their inference.”

At the top layer of the stack, Amazon Q has become the most capable GenAI-powered assistant for software development leveraging a company’s own data, Jassy said. “Early customer testing indicates that Q can turn what was going to be a multi-year effort to do a mainframe migration into a multi-quarter effort, cutting by more than 50 percent time to migrate mainframes,” he said. “This is a big deal and these transformations are good examples of practical AI.”

Jassy, when asked by a financial analyst whether AWS growth is being slowed by supply chain constraints, said that for a company with a multi-billion-dollar annualized revenue run rate business and AI revenue that’s growing by triple-digit percentages year over year, it's hard to complain.

“However, it is true that we could be growing faster if not for some of the constraints on capacity,” he said. “I would say chips from our third-party partners come a little bit slower than before, with a lot of midstream changes that take a little bit of time to get the hardware actually yielding the percentage [of] healthy and high-quality servers we expect. It comes with our own big, new launch of our own hardware on our own chips with Trainium2, which just went to general availability at re:Invent. But the majority of the volume is coming in, really, over the next couple quarters, next few months.”

Other headwinds come from worldwide constrained power availability as well as shortages in certain components such as motherboards, Jassy said.

“I think the team has done a really good job scrapping and providing capacity for our customers so they can grow,” he said. “We're still growing at a pretty reasonable clip, as I mentioned earlier, but I do think we could be growing faster if we were unconstrained. I predict those constraints really start to relax in the second half of ’25.”

DeepSeek

When asked by another analyst about the impact of opensource AI on AWS, Jassy said that many in the industry were impressed with what DeepSeek has done.

“In part, impressed with some of the training techniques, primarily in flipping the sequencing of reinforcement learning being earlier and without the human in the loop,” he said. “We thought that was interesting ahead of the supervised fine tuning. We also thought some of the inference optimizations they did were also quite interesting. For those of us who are building frontier models, we're all working on the same types of things. We're all learning from one another. I think you have seen and will continue to see a lot of leapfrogging between us. There is a lot of innovation to come.”

Jassy said that rather than fear DeepSeek, it is important to embrace it along with as many AI models as possible.

“If you run a business like AWS, and you have a core belief like we do that virtually all the big generative AI apps are going to use multiple model types and [that] different customers are going to use different models for different types of workloads, you're going to provide as many leading frontier models as possible for customers to choose from,” he said. “That's what we've done with services like Amazon Bedrock. And it's why we moved so quickly to make sure that DeepSeek was available both in Bedrock and in SageMaker faster than you saw from others. We already have customers starting to experiment with that.”

Amazon By The Numbers

For its fiscal 2024 fourth quarter, ended December 31, Amazon reported revenue of $187.79 billion, up 10.5 percent over the $169.96 billion the company reported for its fourth fiscal quarter 2023.

That beat analyst expectations by $560 million, according to Seeking Alpha.

Total revenue included product sales of $82.23 billion, up from $76.70 billion, and services sales of $105.57 billion, up from $93.26 billion.

Of the total sales in the quarter, Amazon Web Services accounted for $28.79 billion, up from last year’s $24.20 billion.

Amazon also reported GAAP net income of $20.00 billion or $1.86 per share, nearly double last year’s $10.62 billion or $1.00 per share. That beat analyst expectations by 38 cents per share, according to Seeking Alpha.

For its full fiscal year 2024, Amazon reported revenue of $637.96 billion, up 11.0 percent from the $574.79 billion the company reported for fiscal 2023.

That included product sales of $272.31 billion, up from last year’s $255.89 billion, and services sales of $365.65 billion, up from $318.90 billion.

Of the full year’s sales, $107.56 billion came via AWS, up from last year’s $90.76 billion.