Amazon AI doubles AWS revenue projection to 600bn by 2036

Amazon’s AI-driven AWS revenue projection doubles to $600bn by 2036, accelerating cloud consolidation and forcing competitors into unsustainable AI spending races.

Within 10 years, Amazon Web Services could achieve a $600 billion annual revenue run rate—double its prior $300bn target—driven by enterprise AI workloads consuming compute, storage, and specialized AI services. This trajectory will widen AWS’s lead over Microsoft Azure and Google Cloud, compelling them to match AI infrastructure investments at scale or cede dominance in the enterprise AI stack. Enterprises standardizing on AWS for AI will deepen lock-in, reducing multi-cloud flexibility and increasing Amazon’s control over the AI lifecycle from training to inference.

Amazon CEO Andy Jassy revealed in an internal all-hands meeting that AWS’s 10-year revenue potential has shifted from $300bn to at least $600bn annually due to AI demand, according to a Reuters review of his comments. AWS recorded $128.7bn in sales for 2025, up 19% year-over-year, implying a baseline growth rate of ~17% annually to reach $300bn by 2036; AI accelerates this to $600bn. Jassy noted that AWS’s AI opportunity includes demand for GPU instances, custom chips like Trainium, managed services such as SageMaker and Bedrock, and associated data storage and networking. Amazon’s stock rose ~1% to $213.87 following the disclosure.

The projection reflects a structural shift: AI is not an incremental cloud workload but a primary revenue driver that redefines cloud economics. Enterprises deploying large-scale AI models require massive, GPU-dense compute clusters, high-throughput storage for training data, and low-latency networking for distributed inference—services where AWS holds early advantages through its Nvidia partnerships, in-house chip development, and global infrastructure footprint. As AI moves from experimentation to production, the cloud provider that offers the most performant, cost-effective AI stack will capture disproportionate share of the growing AI budget, estimated to reach trillions globally by 2030.

timeline
    title AWS AI Revenue Growth Trajectory
    2025 : $128.7B actual sales
    2026 : $150B+ baseline (17% CAGR)
    2028 : $250B+ with AI acceleration
    2030 : $400B+ AI-driven run rate
    2036 : $600B AI-enabled revenue target

flowchart LR
    A[Enterprise AI Workload] --> B[GPU-Intensive Compute]
    A --> C[AI Training Data Storage]
    A --> D[Low-Latency Inference Networking]
    A --> E[Managed AI Services (SageMaker/Bedrock)]
    B --> F[AWS EC2 P5/P4 Instances]
    B --> G[AWS Trainium/Inferentia Chips]
    C --> H[AWS S3/EBS Storage]
    D --> I[AWS Elastic Fabric Adapter]
    E --> J[AWS SageMaker JumpStart]
    F & G & H & I & J --> K[AWS Revenue Acceleration]
    K --> L[$600B Annual Run Rate by 2036]

pie
    title AWS Revenue Composition Shift (2025 vs 2036 Projected)
    "2025 Core Infrastructure" : 70
    "2025 AI Services" : 5
    "2025 Other" : 25
    "2036 Core Infrastructure" : 40
    "2036 AI Services" : 35
    "2036 Other" : 25

The financial scale is staggering: an incremental $300bn annually by 2036 represents more than double AWS’s current total revenue. For enterprises, this translates to a potential shift of tens of billions in AI cloud spending toward a single vendor if competitors fail to match AWS’s AI-optimized infrastructure. At a hypothetical 70% gross margin, the additional $300bn generates $210bn in incremental annual profit—enough to fund massive R&D dividends or aggressive price wars.

Control shifts decisively toward Amazon as enterprises increasingly rely on AWS’s AI-native services for model training, customization, and deployment. Companies adopting AWS Bedrock for foundation model fine-tuning or SageMaker for MLOps lock into Amazon’s tooling, data formats, and compliance frameworks, raising switching costs. This enables Amazon to influence AI stack standards, negotiate preferential terms with chip suppliers, and bundle AI services with its broader cloud offerings—a virtuous cycle that marginalizes multi-cloud strategies.

quadrantChart
    title Cloud Provider AI Competitive Positioning
    x-axis Low AI Differentiation --> High AI Differentiation
    y-axis Low Enterprise Lock-in --> High Enterprise Lock-in
    "AWS (High AI, High Lock-in)" : [0.8, 0.9]
    "Azure (Medium AI, Medium Lock-in)" : [0.5, 0.6]
    "GCP (Low AI, Low Lock-in)" : [0.3, 0.4]
    "Niche AI Clouds (High AI, Low Lock-in)" : [0.7, 0.2]

Winners and losers emerge with surgical precision:

Winners:

• Amazon Web Services — captures disproportionate AI cloud spend through integrated hardware (Trainium), software (SageMaker), and global scale
• Nvidia — supplies GPUs for AWS AI instances while co-developing optimized software stacks, reinforcing its dominance in AI compute
• Enterprises with AI-centric strategies — gain access to pre-built, compliant AI services that accelerate time-to-market for generative AI applications

Losers:

• Microsoft Azure and Google Cloud Platform — face revenue compression if they cannot match AWS’s AI performance-per-dollar, forcing unsustainable capex increases
• Traditional cloud cost management tools — become obsolete as AI workloads introduce unpredictable, bursty consumption patterns that defeat reservation-based optimization
• Multi-cloud management vendors — see reduced relevance as AI-driven workloads concentrate on single cloud providers offering superior AI-native integration

Executives should act now to navigate this shift:

1. Audit current cloud AI spending patterns against projected 2030–2036 growth curves to identify potential over-reliance on single-vendor AI services — complete within 60 days.
2. Benchmark AWS AI services (Bedrock, SageMaker, Trainium) against Azure AI and GCP Vertex AI for specific workloads (LLM training, vector search, real-time inference) using standardized TCO models — pilot within 90 days.
3. Negotiate enterprise agreements with AWS that include AI-specific price caps and exit clauses to mitigate lock-in risks while leveraging volume discounts — initiate within Q2 2026.
4. Deploy workload portability layers (e.g., Kubernetes with cloud-agnostic operators) for AI inference services to retain flexibility to shift providers if competitive parity emerges — implement by Q4 2026.
5. Monitor Amazon’s AI chip roadmap (Trainium2, Inferentia3) and performance claims to anticipate future cost advantages and adjust long-term cloud commitments — ongoing with quarterly reviews.

The Infomly Close: For enterprises seeking to optimize AI cloud spend while avoiding vendor lock-in, Infomly’s cloud cost optimization advisory provides vendor-neutral frameworks, benchmarking tools, and negotiation playbooks tailored to AI workloads. Contact admin@infomly.com.