Once synonymous with cloud computing, Amazon Web Services (AWS) reigned supreme as the undisputed leader of the industry. It not only pioneered the cloud computing business model but also revolutionized enterprise IT infrastructure by introducing the groundbreaking “pay-as-you-go, unlimited scalability” model. This innovation enabled companies to move away from costly, inflexible on-premises data centers to a more agile and scalable cloud environment. AWS’s success turned it into Amazon’s primary profit engine, granting the company unmatched pricing leverage and influence across the tech industry.Yet, as tides inevitably shift, the rapid rise of Generative AI is now disrupting AWS’s long-held dominance. Internal documents marked “Amazon Confidential,” along with external market data, paint a clear picture: AWS is undergoing a profound transformation. The startup ecosystem and early-stage IT budgets that once formed the bedrock of its success are steadily being eroded by a new wave of AI-centric technologies and stacks.Once the undisputed pioneer of the cloud, AWS now finds itself in the unfamiliar role of a follower. Rivals it once left in the dust are surging ahead in AI computing power, model ecosystems, and market reputation—steadily dimming AWS’s once-bright star.

I. Budget Realignment: The Fundamental Shift in Startup Spending Behavior

AWS’s dominance was originally built on capturing the lion’s share of early-stage startup IT budgets. For entrepreneurs, AWS was the default choice—an easy on-ramp to launching digital businesses without the need for massive upfront capital investments in physical infrastructure. This value proposition created strong customer loyalty and a resilient ecosystem.

1. Alarming Internal Signals and a Clear Change in Capital Allocation

However, internal AWS documents obtained by international media reveal that this long-standing dynamic is breaking down. Reports marked “Amazon Confidential,” authored in March and July 2024, highlight a “fundamental shift” in how startups allocate their budgets.Traditionally, startups prioritized spending on foundational AWS services such as EC2 compute instances, S3 storage, and relational databases—core components of Infrastructure-as-a-Service (IaaS). In contrast, the new AI-driven investment paradigm is fundamentally different. Today, startups are directing their “First Dollar” not to traditional cloud infrastructure, but to AI models, inference infrastructure, and AI developer tools.One particularly stark warning from the documents reads: “Founders tell us they want to wait until their companies are more mature before choosing AWS.” This signals a critical loss for AWS—not just of market share, but of the invaluable opportunity to secure customer loyalty at the very beginning of a startup’s journey.

2. The Rise of “Cloud 2.0”: A New Compute Paradigm

The Generative AI revolution has given birth to what insiders are calling “Cloud 2.0”—a fundamentally new technology stack. In this era, the focal point of computing has shifted from traditional Central Processing Units (CPUs) to specialized Graphics Processing Units (GPUs) and other accelerators optimized for AI training and inference. Likewise, spending priorities have moved away from generic server rentals toward model APIs, inference optimization, and AI-as-a-Service (AIaaS).Startups today often begin by subscribing to leading AI models from providers like OpenAI or Anthropic, or by leveraging modern developer platforms such as Vercel. They postpone engaging with AWS until they require advanced security, global deployment capabilities, or large-scale legacy infrastructure—pushing AWS further down the priority list.Internal metrics underscore this trend. According to AWS’s own data, only 59% of startups in the 2024 Y Combinator cohort utilized more than three AWS services—down more than 4 percentage points from 2022. Over the same period, 88% of those startups adopted OpenAI models, 72% used Anthropic models, while a mere 4.3% engaged with AWS’s AI developer platform, Bedrock. These numbers reflect a decisive pivot in early-stage technology focus and capital allocation.

3. Superficial Loyalty, Substantial Erosion

Even more concerning for AWS is the behavior of startups that appear to remain “loyal” on the surface. Internal case studies, such as that of Cursor—an AI-powered coding platform—reveal that while the company is often categorized as a heavy AWS user, less than 10% of its total AI-related spending goes toward traditional AWS infrastructure. The bulk of its expenditures are directed toward external model APIs and specialized GPU rental providers like CoreWeave, Lambda Labs, and Crusoe.This exposes a critical vulnerability: while customers may maintain a nominal presence on AWS, their most dynamic, high-margin, and sticky AI-related workloads are increasingly migrating to third-party providers. More worryingly, these new spending categories exhibit far less customer stickiness—allowing businesses to switch providers with ease. AWS’s internal documents caution that these emerging AI-focused expenditures “may soon constitute the majority of startups’ overall cloud consumption.”

II. Strategic Lag: Caught Off Guard by the AI Tsunami

The root cause of AWS’s current challenges lies in its delayed and inadequate response to the AI revolution—especially when compared to rivals like Microsoft and Google.

1. A Missed Moment: The re:Invent 2022 Turning Point

AWS’s current struggles are not accidental. A pivotal moment of strategic misalignment occurred in late 2022, the consequences of which continue to ripple through the industry.On November 30, 2022, AWS CEO Adam Selipsky delivered a two-hour keynote at the re:Invent conference in Las Vegas. The speech was emblematic of AWS’s then-current mindset: traditional, infrastructure-focused, and risk-averse. Selipsky emphasized the company’s longstanding value proposition—reliability, scalability, and the ability to “never worry about over- or under-provisioning.” Throughout the address, there was virtually no mention of Generative AI. When AI was referenced, it was limited to traditional machine learning services, with no discussion of large language models or transformative intelligence.Just hours after Selipsky’s presentation, OpenAI CEO Sam Altman unveiled ChatGPT to the world. Almost overnight, the industry landscape was transformed. ChatGPT’s astonishing conversational abilities propelled Generative AI into the mainstream, not only validating the commercial potential of AI but also signaling a fundamental shift in the cloud industry’s center of gravity—from IaaS to AI-as-a-Service.While the rest of the industry pivoted toward GPU infrastructure, model training, and AI application development, AWS remained tethered to its legacy focus. It wasn’t until 2023 that the company introduced Bedrock, its AI-native platform, by which time competitors like OpenAI and Microsoft had already secured key partnerships and customer commitments. AWS attempted to course-correct at the November 2023 re:Invent conference, mentioning AI nearly 100 times in a two-hour address and rushing out services like Amazon Q. Yet, to many observers, this felt more like a reactive scramble than a coherent strategy.

2. Lost Opportunities with Anthropic and Cultural Barriers

AWS was not entirely blind to the potential of AI. Years ago, its engineers had the foresight to construct a powerful AI supercomputing cluster, linking approximately 6,000 NVIDIA GPUs. However, this initiative was not prioritized internally. At a time when enterprise demand for such compute power was limited, the project was dismissed by some executives as an expensive, non-essential research endeavor—even facing the threat of decommissioning.The situation changed with the emergence of Anthropic, an AI startup founded by former OpenAI researchers and quickly gaining recognition for its cutting-edge work in Generative AI. Anthropic demonstrated an insatiable appetite for compute resources, creating an ideal opportunity for AWS to secure a strategic partnership with a rising AI leader.Yet, internal cultural and strategic hurdles prevented AWS from acting decisively. Executives expressed skepticism about Anthropic’s ability to commercialize its technology, underestimating the disruptive potential of Generative AI. More deeply, AWS’s long-standing cultural inertia—its preference for developing technology in-house rather than paying a premium for external innovation—led to the missed opportunity to invest early in Anthropic.The consequences were significant. In search of reliable compute capacity, Anthropic turned to other providers, notably Google, which became a strategic investor in its early 2023 funding round.AWS’s response was delayed and costly. It wasn’t until September 2023 that the company committed billions to invest in Anthropic, aiming to secure cloud exclusivity and promote its own chips. However, to many insiders, this move felt like an act of desperation. Furthermore, Anthropic has since adopted a multi-cloud approach. Google recently announced plans to provide Anthropic with up to 1 million TPU v5p chips—further eroding AWS’s ambitions for exclusivity and exposing the cost of its earlier strategic missteps.

3. Organizational Challenges: Bureaucracy, Talent Drain, and Innovation Stagnation

AWS’s inability to keep pace in the AI race is also a reflection of broader organizational decline. Once celebrated for its entrepreneurial, fast-moving culture, AWS has in recent years succumbed to the same bureaucratic tendencies that plague larger enterprises.The increase in hierarchical layers is one glaring symptom. A former sales engineer recalled that when he joined AWS, there were only six management levels between him and Amazon founder Jeff Bezos—a hallmark of a flat, agile organization. By 2024, that number had ballooned to 15 between him and current CEO Andy Jassy. Although described as an anomaly, Jassy’s subsequent order to reduce management layers is an implicit acknowledgment of the problem.This bureaucratic sprawl has slowed decision-making to a crawl. Employees report that the time required to draft and approve even simple initiatives has become excessive—often rendering ideas obsolete by the time they are implemented. One engineer noted that while Anthropic could deploy a new experiment in a week, AWS’s internal processes required three weeks just to write the necessary documentation. This glacial pace has severely hampered AWS’s competitiveness.Talent attrition has compounded the issue. In the fierce battle for AI expertise, AWS has struggled to retain key personnel. Promotions and compensation have lagged behind industry standards, and recent layoffs have disproportionately impacted AI-focused teams. The departure of Jon Jones, VP of Startups and Venture Capital, in early 2024 further weakened AWS’s leadership in early-stage market engagement.Internal assessments warn that two and a half years after the launch of ChatGPT, AWS is still perceived by many founders, investors, and industry leaders as a laggard in AI. This damaged reputation has made it harder for AWS to attract top talent and secure speaking opportunities at key industry events—further entrenching a vicious cycle of declining reputation, talent loss, and market share erosion.

III. Market Share Erosion: The Rise of Competitors

The impact of AWS’s strategic delays is quantifiable. Market data reveals a clear shift in the cloud computing landscape—from a period of AWS dominance to a more contested, multi-cloud environment.While AWS reported an 18% revenue increase in Q2 2024, its competitors posted even stronger growth: Microsoft Azure and Google Cloud both exceeded 30% growth rates. Although AWS still holds the largest share of the enterprise cloud market—38% according to 2023 Gartner data—this figure is a sharp decline from its peak of nearly 50% in 2018.Emerging “neocloud” providers are also making inroads. Specialized GPU rental firms like CoreWeave, though starting from a smaller base, have achieved year-over-year revenue growth exceeding 200%. Their flexible models and competitive pricing are attracting startups away from AWS. Oracle, too, has re-entered the spotlight by securing massive AI-focused contracts.Market intelligence from CB Insights tracking 1,100 leading AI startups shows AWS’s market share in this critical segment has slipped from 33% in 2022 to 30% in 2024. In contrast, Google Cloud has risen from 34% to 38%, overtaking AWS, while Microsoft Azure has held steady at around 7%.Pricing and ecosystem limitations have further weakened AWS’s position. Internal documents indicate that 90% of startups backed by Radical Ventures choose alternative cloud providers, primarily due to AWS’s higher GPU costs. Despite recent price cuts of 45% on NVIDIA GPU instances, dissatisfaction persists. Neocloud providers, with their more granular billing and flexible GPU offerings, are winning over cost-sensitive customers.

IV. AWS Strikes Back: A Multi-Front Strategy for AI Dominance

Faced with mounting pressure, AWS has begun to regroup. Starting in late 2024, the company has launched a comprehensive counteroffensive across chips, platforms, and ecosystems—what one former executive described as “AWS reinventing itself, just as it once invented the cloud.”