The Cloud Report Card

Programming note: ARPU will be off the rest of the week, back next Monday.

Amazon's 20% Problem

For the first time in a long time, Wall Street was genuinely relieved that Amazon Web Services had a good quarter. The company's stock soared more than 11% after it reported that its cloud unit grew a very respectable 20%. Analysts cheered the "comeback" and the "turning point."

Which is, you know, a funny way to talk about the undisputed king of an industry. Relief is what you feel when you fear the worst, not what you expect from the market leader. And it's the perfect window into how the AI boom has completely scrambled the old, settled hierarchy of the cloud wars.

The long-standing playbook for the cloud market was simple: AWS invented the business, dominated it, and printed money. Microsoft's Azure was the fast-growing challenger, and Google was a distant third. For years, Azure and Google have been steadily chipping away at AWS's market share, but AI has thrown gasoline on that fire.

Last week's earnings reports put this dynamic in sharp relief. The key metric is now relative speed:

• Amazon Web Services grew 20%, hitting $33 billion in revenue.
• Google Cloud grew 34%, hitting $15 billion in revenue.
• Microsoft Azure grew 40% (the Intelligent Cloud segment, which includes Azure, reported $30.9 billion in revenue)

So how did the company that invented the cloud category manage to get caught so flat-footed? One way to think about it is that AWS became a victim of its own success. As a recent Bloomberg report chronicled, the very things that made AWS the undisputed champion of the first cloud era have become liabilities in the second. Its legendary decentralized structure of small, independent teams was perfectly suited for building discrete services like databases and storage, but completely unsuited for the kind of massive, centralized R&D effort required to build a foundational AI model. Its famous frugality and a cultural reluctance to pay for technology it believed it could just build itself led it to pass on an early investment in Anthropic, only to write a massive, seemingly desperate check years later. And its sheer size created a "bloat" and bureaucracy that slowed everything down at the very moment speed became the only thing that mattered.

In absolute terms, AWS is of course a behemoth; its cloud revenue is more than double Google's. But in the AI-obsessed market of 2025, the only thing that matters is the growth rate. The market is in the middle of a great re-rating. Here's Reuters:

Wall Street cheered AWS's comeback, with analysts noting the earnings marked a potential turning point for Amazon.

"There was definitely concern about AWS losing market share to Microsoft Azure and Google Cloud ... But now AWS is aboard the train as well and they're seeing a big revenue increase," said Jed Ellerbroek, portfolio manager at Argent Capital.

Ellerbroek said investors were expecting an AWS boost in the fourth quarter or early next year. "But it's already come this quarter," he said.

"Aboard the train as well" is the key phrase. It implies that for a while, the market believed the AI train had left the station without its biggest passenger. The cheer for Amazon wasn't a celebration of its dominance, but a sigh of relief that it hadn't been left behind entirely.

Of course, this new reality is a double-edged sword. Microsoft, despite its blowout 40% growth, saw its stock dip modestly after announcing a record-breaking capital expenditure of nearly $35 billion for the quarter. The market loves the narrative of AI-fueled growth, but it's getting very nervous about the astronomical cost of the shovels required to dig for all that gold.

And that is the new, chaotic state of the cloud business. The market has spent the last few years rewarding Microsoft and Google for their successful challenge to AWS's dominance. Now, that challenge has been supercharged by AI, creating a frantic, quarter-to-quarter battle for momentum. AWS may still occupy the throne, but for the first time in a decade, the succession is an open question.

More on Hyperscalers:

• Can Google Keep Up With the AI Buildout? (ARPU)
• The great AI buildout shows no sign of slowing (Reuters)

On Our Radar

Our Intelligence Desk connects the dots across functions—from GTM to Operations—and delivers intelligence tailored for specific roles. Learn more about our bespoke streams.

The AI Data Deluge

• The Headline: Data Storage Firms Western Digital and Seagate See Unprecedented Demand and Multi-Year Orders Driven by AI Infrastructure Buildout (Reuters)
• ARPU's Take: The great AI infrastructure build-out has triggered a massive downstream effect in the unglamorous but essential world of data storage. As millions of users adopt AI services, the resulting explosion of user-generated data is forcing hyperscalers to secure the vast 'digital real estate' needed to store it all for years to come.
• The Operations Implication: Legacy hardware like HDDs is becoming a mission-critical component of the AI infrastructure stack. For Western Digital and Seagate, this transforms their core product from a low-margin commodity into a strategic, capacity-constrained asset. This also creates a new operational challenge for hyperscalers, who must now treat storage procurement with the same long-term planning and supply chain security as they do for high-end AI accelerators.

Sovereign AI Franchises

• The Headline: Nvidia Partners with South Korean Government and Tech Giants to Build National AI Infrastructure (Nikkei Asia)
• ARPU's Take: This isn't just a sales deal; it's the blueprint for Nvidia's new business model: selling "Sovereign AI Franchises." Nvidia is positioning itself as a foundational utility for nations, providing the entire stack—from chips to software—that a country needs to build its own AI-powered economy.
• The Go-to-Market Implication: This is a prime example of Nvidia's "national infrastructure" go-to-market strategy, moving beyond enterprise sales to becoming a foundational partner for countries building sovereign AI. By securing a top-down commitment from the South Korean government and its key industrial players, Nvidia locks in a massive, multi-year deployment and deeply embeds its entire stack (CUDA, Blackwell, AI models) into the country's economic engine. This strategy effectively marginalizes competitors like AMD and Intel, relegating them to niche or secondary supplier roles within these national ecosystems.

P.S. We're building a handful of bespoke intelligence streams for members. If you're facing a specific intelligence challenge, drop us a line here.