Internal bureaucracy, uninspired AI! What’s going on with Amazon AWS?

As the inventor and long-time absolute leader of the cloud computing business, Amazon is showing unprecedented signs of fatigue. The company's profit engine is now struggling due to growing internal bureaucracy, sluggish responses in key AI areas, and aggressive competition from rivals, putting its market dominance to a severe test.

The latest developments have further heightened market concerns. On Thursday, Google, a subsidiary of Alphabet, announced it would provide AI startup Anthropic PBC with up to one million of its dedicated AI chips. This move not only deepens Google’s partnership with the rapidly growing AI newcomer but also deals a direct blow to Amazon, which had previously invested billions of dollars in Anthropic.

Meanwhile, AWS’s operational stability has raised red flags. Just on Monday, this cloud division experienced one of the most severe outages in its history, crippling one of its most critical data center clusters and interrupting operations for hundreds of companies and consumer apps. The event lasted about 15 hours before full service was restored.

These incidents highlight deeper problems AWS faces. Nearly three years after OpenAI’s ChatGPT sparked an AI boom, the market generally believes AWS has fallen behind its tech peers in the AI field. Although AWS remains the cloud market leader, its growth has slowed, and competitors are fiercely encroaching on its once-stable territory, bringing new uncertainty to investors and the market.

Slow Start in the AI Race, Missed Opportunities

In the decisive AI battleground of the future, AWS’s start was noticeably a step behind. A symbolic moment occurred on November 30, 2022: On that day, then-AWS CEO Adam Selipsky gave a two-hour keynote at the annual re:Invent conference, barely mentioning AI. Yet, that same afternoon, OpenAI launched ChatGPT, completely upending the entire industry.

According to sources, AWS was not oblivious to AI’s potential. It had even supported Anthropic with powerful computing resources early on, but skepticism about new tech’s profitability and an internal culture reluctant to pay for outside technology made Amazon miss the initial chance to invest in Anthropic. This hesitation allowed Google to seize the opportunity.

It wasn’t until September 2023 that Amazon invested its first large sum into Anthropic, planning a total investment of up to $4 billion in exchange for Anthropic's commitment to use AWS’s computing and self-developed chips. For many veterans familiar with Amazon’s aversion to hefty tech premiums, this move seemed like an act of desperation. Google’s deepening partnership with Anthropic on Thursday further underscores Amazon’s passive position in this race.

The "Big Company Disease" of a Vast Empire

Once proud of its "startup" spirit, Amazon is now weighed down by the bureaucracy of its vast empire. According to media interviews with 23 current and former AWS employees, they commonly described an internal bureaucratic system that moves slowly when agility is needed.

One sales engineer recalls, before the pandemic, he was separated by only six management layers from then-CEO Jeff Bezos. Yet after a promotion, reviewing the org chart this year, he discovered he was now 15 layers away from current CEO Andy Jassy (Amazon claims this was an isolated case). In the increasingly bureaucratic culture, decision-making has slowed dramatically. Three employees working on separate AI projects said they were required to repeatedly revise proposals until market trends changed and their ideas were outdated.

Meanwhile, mass recruitment during the pandemic caused management layers to swell, and subsequent layoffs and cost cuts made promotions and raises harder to come by. Amid a heated industry-wide battle for AI talent, AWS instead lost several key executives, including heads of AI, startup sales, chip design, and data center infrastructure.

Changing Competitive Landscape: Startups "Switch Sides"

While AWS struggles with internal "bloat," the external competitive landscape has changed radically. Microsoft's cloud business backlog growth rate now exceeds Amazon's. Oracle, once considered a cloud market also-ran, has signed multibillion-dollar contracts to provide hosting for cutting-edge AI development. Google has become a more threatening rival, especially in attracting AI startups.

“If you don’t win those AI-native startups whose scale will grow five to tenfold in the future, that could be a real problem for the business,” said Raymond James analyst Josh Beck. Pete Schwab, who worked for Amazon for ten years, chose Google when founding his own AI company, noting that Google "does a better job supporting small companies like ours."

Even AWS's existing major clients are “shopping around” in AI. Writing assistant company Grammarly Inc. has long run on AWS, but CTO Mark Schaaf said in an interview that AWS's AI model marketplace Bedrock failed to meet the company’s needs on price and other factors, so they turned to OpenAI and Meta models instead.

New CEO Takes Helm, Striving to Regain the Initiative

Facing internal and external challenges, AWS is trying to adapt. This year, longtime AWS veteran Matt Garman replaced Selipsky as CEO. Colleagues describe him as shrewd and capable, and outside analysts believe he's better suited for AWS’s current “wartime” state.

After taking office, Garman urged employees to focus on delivering promised products. To speed things up, some AI teams have even suspended parts of typical product development practices, sacrificing documentation and routine reviews. Over the past two years, AWS has terminated or halted development of more than thirty services and key features to free up engineers for building AI tools.

AWS is pinning its hopes for a comeback on AI services such as Bedrock and its self-developed Trainium2 chips. Company spokesperson Selena Shen stated that Bedrock now has tens of thousands of customers and is widely regarded as its most successful AI product. Amazon CEO Andy Jassy has repeatedly emphasized the cost efficiency of the Trainium2 chip as a potential competitive advantage, claiming its AI workload costs are 30% to 40% lower than rival hardware.

However, whether it’s spending $11 billion building a data center in Indiana for Anthropic or accelerating internal product line iterations, AWS must face a harsh reality: the competitors it once easily left behind have now dragged it into a fierce struggle. Google’s latest partnership with Anthropic demonstrates once again that, in this crucial AI race for the future, Amazon no longer enjoys an easy path to victory.

Risk Disclaimer and StatementThe market entails risks; investment requires caution. This article does not constitute personal investment advice, nor does it account for individual users’ unique investment objectives, financial status, or needs. Readers should consider whether any opinions, viewpoints, or conclusions herein are appropriate for their specific circumstances. Investors bear their own responsibility for any actions taken based on this article.