Skip to content
Amazon Wants to Be an AI Infrastructure Company, Not Retail
Amazon
Amazon shares were down about 7% following a 1% September revenue miss and a December guide that was in line with expectations. AWS grew at 39% y/y in September vs. the Street at 38% and a 40% whisper. The big news was management did the unthinkable and guided AWS to grow at least 40% y/y in both 2027 and 2028. To put the magnitude into perspective, they typically only give one quarter of AWS guidance. Separately, they guided for AWS revenue to reach $1T annually in 5–10 years, which suggests the growth will be as high as 42% in each of the next five years or as low as 22% in each of the next ten years. Either way, the magnitude and duration of that growth are off the charts. Today, AWS is about 21% of revenue. In five years it will be around 50%. The fact that management led the earnings call with these two guidance points, along with comments that they will be selling their custom GPUs (Trainium) starting late in 2027, underscores they want investors to think of them as an AI infrastructure company and to focus less on the retail and ad business, which today accounts for about 75% of revenue. In the end, the shift in the investment thesis comes at a good time when the core retail business is poised to come under attack from personalized bots and the core AI infrastructure is set to take off.

Key Takeaways

AWS guidance for at least 40% y/y growth in both 2027 and 2028 signals conviction in demand well beyond what they've been willing to provide in the past.
Management’s belief that AWS can become a $1T business in 5–10 years, compared to $165B in CY26, sends a message to investors to rethink Amazon’s investment case toward AI infrastructure and away from retail.
Building out AWS growth of 22–42% over the next five to ten years will have a meaningful cost when it comes to capex, which is not yet reflected in Street numbers. I believe investors will be OK with that spend.
1

Two Years of AWS Guidance Is the Real Signal

AWS is on fire and will continue to be for the next several years, possibly the next decade. The segment grew 39% y/y in the September quarter, and the company’s outlook calls for at least 40% y/y growth in both calendar 2027 and 2028. What sticks out to me is both the length of that outlook and its magnitude. Amazon typically gives one quarter of guidance, so providing two years is a powerful statement of management’s confidence in how demand will play out over the next two years.

This begs the question: Why go that far out on guidance? I believe the company wants to send a message to investors that, despite all of the hype, we are just starting and the AI movement will prove to be a “once-in-a-lifetime opportunity.”

2

A $1T AWS Revenue Target Changes How Investors Should Think About Amazon

The second key takeaway from earnings was that management believes AWS can become a $1T business over the next five to ten years. As a point of reference, before the updated guidance, the Street expected AWS revenue of $165B in CY26 and growth of around 30% in CY27 and CY28, ending CY28 with about $322B in revenue.

In other words, management is saying that its two-year outlook of 40%-plus AWS growth is not the end point, but the starting point. If you assume they hit $1T in revenue in five years, that implies AWS will grow, on average, 42% in each of the next five years. That’s a growth outlook that is hard for me to wrap my head around. If you take the approach that it takes them ten years to reach that milestone, that suggests they will grow at 40% in each of the next two years and an average of 17% in each of the next eight years, averaging 22% annual growth, which is also an impressive outlook.

Separately, the company plans to begin selling its Trainium GPUs toward the end of 2027. While I don’t expect chip sales to become a measurable business (less than 5% of revenue in 10 years), they are incremental to revenue and the narrative that Amazon is now an AI-first company.

3

Higher Capex Is Part of Capturing the Opportunity

Amazon’s bullish AWS revenue growth outlook comes with a cost: capex. On the call, the question came up around how much it will cost to build the infrastructure to support the growth. The company’s response was that capex spending long term will be “meaningful,” and that’s OK because this is “a once-in-a-lifetime opportunity,” and they’ll invest to capitalize on it.

After the earnings, analysts’ updated estimates call for capex to grow 62% in CY26, 31% in CY27, and 10% in CY28. My reaction is that those estimates are still meaningfully low. I suspect CY27 capex will increase by 45% and 25% in CY28. All of that is to say, this is good news for the AI hardware trade.

Disclaimer

Back To Top