Can Amazon Web Services (AWS), the former leader in the cloud market, catch up with Microsoft Cloud and Google Cloud in AI and reclaim its throne following Jeff Bezos' return?

01

AWS Tackles Its AI Shortcomings

Three and a half years after stepping down as Amazon's CEO, Bezos announced his return this month.

In a recent interview, the reclusive founder of Amazon, Jeff Bezos, revealed that 95% of his focus is on artificial intelligence. "Modern AI is a horizontal enabling layer that can be used to improve everything and will appear in everything," he said.

However, an interesting exchange occurred when the host mentioned that AWS does not have its own large language model. Bezos quickly replied, "You've been too busy," as AWS had just announced the release of its foundational large language model, Nova.

This conversation underscores an awkward reality: as the global cloud market leader, AWS's progress in AI lags significantly behind its old rivals, Microsoft Cloud, and Google Cloud.

Therefore, at the recent 2024 Re: Invent conference, AWS's most important annual event, the company primarily focused on AI announcements and deployments.

For the first time in his three years as CEO, Amazon CEO Andy Jassy took the stage at the Re: Invent conference and announced the launch of six new foundational models. Named "Amazon Nova," these products include text generation models, image generation models, and video generation models, with costs 75% lower than competitors to stimulate customer purchase intent. Moreover, Amazon expects to launch generative large models capable of accepting input in any modality and outputting in any modality by mid-2025.

In addition to the new foundational models, Amazon also announced the launch of the new generation AI chip Trainium 3, which is four times faster than its predecessor, and a partnership with Anthropic to develop the world's largest AI supercomputer.

According to Peter DeSantis, Senior Vice President of AWS Utility Computing, AWS will focus on increasing the capacity of individual systems rather than adopting a "horizontal scaling" approach by building multiple identical systems and distributing workloads across multiple machines. "Our approach to building larger models is to build more powerful servers," he said.

AWS has also reduced the usage cost of Amazon Q, with a starting price of $20, which is lower than Microsoft's Microsoft 365 Copilot and Google Workspace's Duet AI, priced at $30 per month, attracting customer attention.

On its path to re-establishing its competitiveness in AI strategy, AWS has also experienced frequent leadership changes.

In June of this year, Adam Selipsky, who had served as CEO of AWS for three years, officially announced his resignation, and was succeeded by Senior Vice President Matt Garman.

Over the past three years, AWS's business has faced numerous challenges, including a noticeable slowdown in revenue growth. In 2023, AWS's revenue growth rate fell to around 12%, its slowest growth rate on record. Moreover, during Adam Selipsky's tenure, AWS also experienced multiple rounds of large-scale layoffs, affecting over 27,000 employees.

AWS's Greater China business has also been affected, with frequent personnel adjustments. In 2023, AWS Greater China changed its leadership, with Chu Ruisong, former Vice President of Baidu Group, succeeding Zhang Wenyi. Prior to this, AWS Greater China Vice Presidents Tian Zhuo and Duan Weizhi also left for Microsoft Cloud.

In June of this year, Liang Yan, head of AWS Greater China's Top Acct department responsible for major customers, was also recruited by Microsoft Cloud. The exodus of a large number of senior executives to rival companies reveals AWS's embarrassing situation in the AI era.

02

'Amazon is Six Months Behind Microsoft'

Over the past two years, Amazon, which had been a leader in AI services, suddenly found itself falling behind overnight.

Months after OpenAI launched ChatGPT and it became a sensation, Amazon did release its own large language model, Titan, in April 2023, but it did not spawn a groundbreaking product like the ChatGPT chatbot.

In the same month, Amazon AWS also launched the generative AI service Bedrock, which allows businesses to integrate their data into AI algorithms to build products.

"For Amazon, we're actually talking about being six months behind Microsoft," said Chris Mack, an information technology analyst and portfolio manager at Harding Loevner.

Under competitive pressure from Microsoft and Google in AI, AWS invested in Anthropic in September 2023, a rising star in the generative AI field founded by a former OpenAI executive in 2021 and considered a competitor to OpenAI. AWS hopes to establish a model similar to the one between Microsoft and OpenAI. Following the investment, Anthropic designated AWS as its "primary cloud provider" and uses AWS's AI machine learning accelerator chip "Trainium" and inference accelerator chip "Inferentia." The two parties will also collaborate on AI chip development. AWS's investment is divided into two tranches totaling $4 billion.

About a year after ChatGPT's launch, in November 2023, AWS announced its AI strategy at its annual customer conference re:invent in Las Vegas: introducing an AI chatbot tool for enterprises named Amazon Q, highlighting its Bedrock service, launching a new version of the Trainium compute chip optimized for AI, and announcing an expanded partnership with AI chip giant NVIDIA.

Adam Selipsky, then CEO of AWS, stated that generative AI "will reshape every application we encounter in work and life" and that "we are ready to help customers transform through generative AI." However, these statements lagged significantly behind those of American competitors and even major Chinese cloud computing companies.

The aftermath of this conference was relatively quiet, as news of AI innovations emerged daily, with competitors constantly exchanging blows. Compared to Microsoft's momentum in generative AI, AWS is clearly behind.

As early as 2021, a year before OpenAI launched ChatGPT, Microsoft integrated OpenAI's models and technologies into the Azure platform to launch Azure OpenAI services. In 2019, Microsoft made its first $1 billion investment in OpenAI. Shortly after ChatGPT's launch, the tech giant integrated ChatGPT into its Azure cloud business. Moreover, its Copilot tool embeds AI functionality into widely used Office applications in enterprises, including Excel, Word, and Teams.

Even Wall Street has formed the argument of "excluding Amazon." As a result, Amazon CEO Andy Jassy had to refute during the earnings call in February 2024 that "generative AI is and will continue to be an area of focus and investment for Amazon company-wide." Andy Jassy succeeded Amazon founder Jeff Bezos in 2021. He stated that generative AI would generate "tens of billions of dollars" in revenue over the next few years.

"We still see momentum among customers who want to work with AWS on long-term generative AI projects," Andy Jassy reiterated, seemingly refuting the notion that Microsoft is the preferred AI partner.

Previously, in December 2023, a survey by UBS analysts of over 100 enterprise software buyers found that Microsoft Cloud Azure was the most popular choice for enterprises currently running AI workloads, with 62% of respondents indicating they were using Azure's services. AWS lagged slightly, with 52% of respondents using it, indicating that enterprises use services from more than one vendor.

However, it is noteworthy that 51% of respondents said they were most likely to use Azure for future generative AI projects, while only 29% would choose AWS. "The survey results suggest that looking ahead, Azure's advantage over AWS may further widen," said UBS analyst Karl Keirstead.

If AWS does not fully change its momentum in AI, it will face tricky problems in the future. There is also an industry saying that AI has become the next battlefield for cloud services. Global cloud computing companies may experience a reshuffle in the landscape of intelligent computing services.

03

Potential Risks in Revenue Growth

The impact of large models and generative AI on the revenue growth of cloud vendors is already evident. Microsoft Cloud and Google Cloud have obviously narrowed the gap with AWS, the leader in cloud computing, during this wave of generative AI development.

Originally, as enterprises cut their IT expenditures, the sales growth of AWS, Microsoft Azure, and Google Cloud had already slowed down. However, the third quarter of 2023 was a clear demarcation point. Microsoft Azure and Google Cloud rebounded first.

In this quarter, Microsoft's revenue from its Intelligent Cloud segment (including Azure, Windows Server, and enterprise services) reached $24.3 billion, ending a streak of declining revenue growth rates for multiple quarters with a growth rate of 19.4%, and maintaining a stable revenue growth rate of around 20% for the subsequent five quarters.

Google Cloud hit the lowest point of revenue growth in the recent eight quarters in this quarter and has since been on an upward trajectory, with year-over-year growth surging from 22.4% in Q3 2023 to 35% in Q3 2024, and revenue increasing from $8.41 billion to the latest figure of $11.35 billion.

In contrast, Amazon AWS, which deployed generative AI later, experienced a more delayed surge in revenue growth. In the first quarter of 2024, AWS generated $25.037 billion in revenue, a year-over-year increase of 17.3%, a direct increase of nearly four percentage points from the previous quarter's growth rate of 13.2%, and maintained revenue growth rates of 18.7% and 19% in the subsequent two quarters, respectively.

During the latest earnings call for the third quarter of 2024, the management teams of Amazon, Microsoft, and Google all mentioned the driving effect of AI business on their respective businesses.

Microsoft's management disclosed during the earnings call that revenue from Azure and other cloud services grew by 33%, higher than the market expectation of 29.4%, with 12 percentage points coming from its AI services, up from 11 percentage points in the previous quarter.

Google CEO Sundar Pichai announced during the earnings call that usage of Google's Gemini API had surged 14-fold in the past six months.

While Amazon did not disclose the specific contribution of generative AI to AWS business growth in its earnings report, CEO Andy Jassy revealed during the call that generative AI has become a business with "billions of dollars in revenue run rate" under AWS and "continues to grow at a triple-digit annual growth rate." Its growth rate is also more than three times that of AWS's current growth rate.

Although in terms of cloud business revenue, Amazon AWS remained the absolute "leader" in the third quarter of 2024, generating $27.452 billion in revenue, higher than Microsoft Intelligent Cloud's $24.09 billion and Google Cloud's $11.35 billion. However, in terms of revenue growth rate, AWS's revenue increased by 19% year-over-year in Q3 2024, still lower than Google's (35% year-over-year growth for Google Cloud) and Microsoft's cloud business (33% year-over-year growth for Azure revenue, disclosed separately for the first time).

However, the industry believes that AWS cannot be excluded from the generative AI era as some people might think. The company is waging a counterattack in AI. Whether AWS can achieve a turnaround remains to be seen.