Amazon’s overwhelming AI demand is just a bronze medal compared to its rivals
Weak guidance for the current quarter overshadowed a strong second-quarter earnings report. Despite Amazon being the leader in cloud computing, analysts questioned its slower growth compared to competitors.
Amazon has so much demand for AI in its AWS services that it has a $195 billion backlog. Its earnings and revenue for the second quarter beat analysts’ expectations. But investors overlooked that good news to focus on a weaker-than-expected operating income forecast for the current quarter and huge spending on capital expenditures.
Like Microsoft, Amazon’s AWS cloud business benefits from any customer’s AI computing needs, and has invested heavily in meeting those needs.
Amazon is building massive clusters of data centers filled not only with Nvidia GPUs, but also many in-house custom Trainium2 chips, which CEO Andy Jassy called “the backbone for Anthropic’s newest generation cloud models.”
But Jassy was pressed on the company’s earnings call about why AWS — the leader in the market — was growing slower than its competitors. Alphabet’s cloud business grew 31% year on year, and Microsoft’s Azure business grew 39% year on year this quarter. Amazon’s AWS revenue grew 17.5% for the quarter. Jassy’s long nonanswer did not soothe investors.
And the heavy capex spending to keep pace with demand could affect profits, Brian Lisowski, Amazon’s CFO, said:
“We expect AWS operating margins to fluctuate over time, driven in part by the level of investments we are making at any point in time. We will continue to invest more capital in chips, data centers, and power to pursue this unusually large opportunity that we have in generative AI.”
Tariff uncertainty
When asked about the impact of President Trump’s chaotic tariff plans, Jassy said the company hasn’t seen diminished demand or widespread price increases, but:
“We just don’t know what’s going to happen moving forward. It’s hard to know where the tariffs are going to settle, particularly in China. It’s hard to know what will happen when we deplete some of the pre buys that we did on our own first party retail and then some of the forward deploying that we saw of our third-party selling partners. And, you know, that that could change in the second half.”
Project Kuiper vs. Starlink
In response to an analyst question about Project Kuiper, Amazon’s answer to SpaceX’s Starlink satellite internet service, Jassy said he felt the company had a good shot at being second in the space, thanks to what he says is a price and performance edge and the strong relationships the company can leverage. Jassy said:
“If you think about the three key customer segments who want low Earth orbit satellite — consumers, enterprises, and governments — we have very strong relationships with all three customer segments given our consumer businesses and our AWS business.”
Jassy also said that even though the service hadn’t launched yet, Amazon has already signed enterprise and government contracts for the service, which aims to launch a “commercial beta” by the end of the year or beginning of next year.
Jassy: “It’s so early” in AI
On the earnings call, Jassy was asked if there would be surge of growth over the next year, with the explosion of generative AI spreading everywhere.
Jassy explained that all of these AI applications don’t exactly result in steady growth going up all the time:
“If you look at what’s really happening in the space, you have — it’s, it’s very top heavy. So you have a small number of very large frontier models that are being trained that spend a lot on computing.”
Jassy said while the computation required for training is huge, that only happens every so often. Most of the AI computing time is spent for “inference” — running actual AI queries for customers.
“But in at scale, you know, 80% to 90% of the cost will be in inference because you only train periodically, but you’re spitting out predictions and inferences all the time.”
And that is where Amazon believes it will have a long-term advantage with its cheaper and more energy efficient custom chips. But time will tell if that strategy will pay off in the fast-moving world of AI.
