Sarah is a freelance web developer based in Birmingham. She signed up for AWS to learn cloud skills — used the “free tier,” spun up an EC2 instance, ran it for a weekend, and then shut it down. Three weeks later, her bank sends a notification: Amazon Web Services has charged her £67.

She contacts support. After waiting four days and going through three different chat agents, she’s told the charges came from a NAT Gateway she accidentally left running ($0.045/hour × 720 hours = $32 before data processing fees) plus EBS snapshots she didn’t know existed. The support agent is polite. But the money doesn’t come back.

This is not a rare edge case. It’s a documented, recurring pattern — replicated across thousands of Trustpilot reviews, Reddit threads, and developer forums worldwide. AWS is the most powerful cloud infrastructure platform ever built. It runs 30% of the global cloud market. Netflix, NASA, and the UK’s National Health Service trust it. And yet the same platform that keeps mission-critical enterprise workloads alive for 99.999% of the year can silently drain a student’s account overnight without a single warning notification.

That contradiction — world-class infrastructure, chaotic billing experience — is the core tension this review is built to examine.

The cloud computing market crossed $400 billion in annual revenue in 2025 — its biggest milestone yet. AWS contributed the largest single slice at 29–30%, but the growth story is more complex than the headline number suggests.

AWS’s market share has been in slow decline for five consecutive years — from 33% in 2021 to 29% by Q3 2025. Meanwhile, Azure grew from leveraging its Microsoft enterprise ecosystem, and Google Cloud accelerated on the back of AI and Gemini infrastructure. For the first time in AWS’s history, the question “should we use Azure instead?” has a credible answer for many organisations.

On the product side, 2025 brought major support plan restructuring — the old Developer, Business, and Enterprise On-Ramp plans were all discontinued in December 2025. The new model: Business+ at $29/month minimum, Enterprise at $5,000/month minimum, and Unified Operations at $50,000/month minimum. The rebranding lowered entry prices on paper but quietly eliminated the affordable mid-tier plan most SMBs relied on.

Amazon Web Services is a subsidiary of Amazon.com, launched in 2006 when Amazon decided to commercialise the internal cloud infrastructure it had built to power its own e-commerce operations. Founded in Seattle, AWS is led by CEO Matt Garman (as of 2024) and operates as Amazon’s most profitable business unit — generating $29 billion in Q1 2025 alone, representing 60% of Amazon’s total operating profit despite being only 15% of total revenue.

AWS is not a hosting platform in the traditional sense. It’s an infrastructure-as-a-service (IaaS) and platform-as-a-service (PaaS) ecosystem with over 200 products across compute, storage, databases, networking, machine learning, analytics, security, IoT, and developer tools. It operates 117 Availability Zones across 37 global regions, with 3 more regions planned. Its flagship services include EC2 (virtual machines), S3 (object storage), RDS (managed databases), Lambda (serverless functions), and CloudFront (CDN).

The positioning AWS uses — “the infrastructure for everything” — is not marketing hyperbole. Companies as different as Netflix, Airbnb, Unilever, the NHS, Deliveroo, and countless government agencies all run critical workloads on AWS. The question for most potential users isn’t whether AWS is capable. It’s whether AWS is appropriate for their stage, budget, and technical capacity.

Pattern 1: Free Tier — The Gap Between Promise and Reality

The free tier exists. It works. But “free” requires surgical precision to maintain. The moment you spin up a resource and forget to configure a billing alarm, forget to delete a snapshot, or accidentally enable a service in a different region — you’re billed. This pattern repeats across Trustpilot and Reddit with startling consistency. Students experimenting for certifications, developers testing personal projects, and early-stage founders all report the same experience: charges that arrive without warning.

The structural issue isn’t dishonesty — it’s architecture. AWS was built for enterprise-scale users who have billing alerts set up, cost dashboards reviewed weekly, and cloud engineers on payroll. The same architecture applied to a student’s first account creates a billing minefield.

Pattern 2: Infrastructure Reliability — Genuinely Excellent

Here’s where AWS earns every star: the core infrastructure performance is outstanding. EC2 uptime, S3 durability (eleven nines — 99.999999999%), and the global network architecture are benchmarks the entire industry measures itself against. Enterprise users on G2 and TrustRadius consistently report years of operation with minimal downtime. One user with 17 EC2 instances over two years reported zero single infrastructure-related failure. AWS’s Load Balancer handles 400 billion weekly requests. When AWS infrastructure fails, it makes news precisely because it’s so uncommon.

Pattern 3: The Learning Curve Is a Real Business Risk

Multiple G2 and TrustRadius reviewers describe the AWS console as “overwhelming,” “fragmented,” and “not beginner-friendly.” Each service has a separate console. There’s no universal search across regions by default. IAM policy management in JSON syntax is genuinely complex. A 2-person startup using AWS without prior cloud experience isn’t just learning a tool — they’re taking on infrastructure risk. When something breaks at 2am, AWS’s free Basic Support provides documentation links and community forums. That’s all.

Pattern 4: Account Management Issues — A Recurring Crisis

Account lockouts, MFA recovery failures, and suspended accounts are among the highest-frequency complaints on Trustpilot. Several UK users report their accounts being frozen mid-operation with no clear explanation and support responses that take days. One business reported their domain remained down for over a week while AWS support failed to respond with a solution. Another user documented being locked out of their account for 30+ days due to an MFA issue, with email-only support offering no resolution. For businesses running production workloads, account access issues at this severity represent an existential risk.

Pattern 5: AI Services — Powerful but Expensive at Scale

Amazon Bedrock and SageMaker are real, capable tools. Enterprises using AWS for AI/ML report genuinely strong results — GPU scaling, integration with S3 and RDS, and managed Kubernetes via EKS. But AI workloads are where the cost explosion risk is highest. Token-based pricing, GPU instance costs, data ingress/egress between AI pipelines, and training compute can turn a $500/month architecture into a $5,000/month one inside a single sprint. Product Hunt reviewers specifically called out console complexity as a barrier to AI adoption.

What this data shows: AWS’s infrastructure reliability claim is the one that holds up strongest — enterprise users broadly agree the core platform is excellent. The “pay only for what you use” claim is partially true but materially misleading: users pay for what they use, but the definition of “use” is much broader than most people expect (idle NAT Gateways, forgotten snapshots, and background data transfers all count). The free tier claim has the largest gap — nearly 6 in 10 users in negative reviews mention unexpected charges during what they believed was a free-tier engagement. Support satisfaction is split sharply by tier: paid Enterprise users are generally positive, Basic and free-tier users are consistently dissatisfied.

🔴 Billing Complaints

🟡 Product & UX Issues

🟢 Genuine Praise

If you’re going to use AWS — and for many workloads, you should — here is the only rational approach to avoid the patterns that destroy people’s bills and trust.