Market Position and Enterprise Centre of Gravity
Understanding where each cloud provider leads in enterprise adoption is the starting point for making informed platform decisions. The market share figures — AWS at approximately 30%, Azure at 23%, GCP at 13% — tell only part of the story. More important for enterprise decision-making is understanding where each platform's capabilities are strongest and which enterprise buying motions each vendor has optimised for.
AWS: Breadth, Maturity, and Independence
AWS launched commercially in 2006 and has the broadest service portfolio of the three major platforms. With over 200 managed services spanning compute, storage, database, networking, security, AI, analytics, and developer tooling, AWS serves enterprises that value breadth of capability and flexibility in architecture. AWS does not have a stake in your other enterprise software relationships — it does not sell productivity software, ERP, or desktop operating systems. This independence is a genuine advantage for organisations that want a cloud provider without vendor integration conflicts, and it explains why AWS has historically been the cloud platform of choice for technology-forward organisations building cloud-native infrastructure from a clean slate.
The challenge with AWS is complexity. The breadth of the service catalogue, combined with AWS's pricing architecture — which involves separate pricing for compute, storage, data transfer, API calls, and other consumption dimensions — creates a billing complexity that consistently produces cost surprises for enterprises that do not invest in robust cloud financial management (FinOps) capabilities. AWS has the highest volume of monthly price changes of the three platforms, averaging approximately 197 distinct price modifications per month for GPU and non-GPU instances, reflecting its large and continuously evolving service portfolio.
Azure: Microsoft Ecosystem Integration and Hybrid Advantage
Microsoft Azure launched in 2010 and has grown to become the enterprise cloud platform most closely integrated with the existing Microsoft software estate. For organisations that run Microsoft 365, Active Directory, SQL Server, Windows Server, Visual Studio, and Dynamics 365, Azure provides a native integration advantage that reduces both the technical complexity and the cost of cloud integration. The Azure Hybrid Benefit — which allows organisations to bring existing Windows Server and SQL Server licences to Azure at reduced cost — is particularly valuable for enterprises with significant Microsoft licence investment, as it can reduce Azure compute costs by 40% or more for eligible workloads.
Azure's enterprise buying motion is closely tied to the Microsoft account relationship. Organisations with large Microsoft Enterprise Agreements often find that Azure spend can be managed within the same commercial relationship, consolidated under a Microsoft Azure Consumption Commitment (MACC) that provides commercial and procurement simplicity alongside potential bundling benefits. For organisations that are managing Microsoft EA renewals concurrently with cloud infrastructure decisions, the ability to roll Azure commitments into a broader Microsoft commercial conversation provides negotiating leverage that is not available with standalone AWS or GCP agreements.
GCP: Analytics, AI/ML, and Pricing Predictability
Google Cloud Platform, while the smallest of the three in market share, has built a differentiated position in specific enterprise capabilities. GCP leads in data analytics (BigQuery), managed Kubernetes (GKE is widely regarded as the most mature managed Kubernetes service), and AI/ML infrastructure — Google's TPU-accelerated infrastructure provides performance advantages for large-scale machine learning training workloads. Google's broader AI platform (Vertex AI) and the integration of Gemini models into enterprise applications represent a significant investment that positions GCP as a competitive alternative to Azure OpenAI and AWS Bedrock for organisations building AI-intensive applications.
GCP's pricing architecture has one significant and underappreciated advantage: Sustained Use Discounts (SUDs), which provide an automatic discount of up to 30% for workloads that run consistently throughout a billing month, without requiring any reservation or upfront commitment. This pricing mechanism rewards consistent utilisation automatically, reducing the optimisation overhead compared to AWS Reserved Instances or Azure Reserved VMs. For organisations with steady-state, predictable workloads, GCP's pricing predictability — fewer price changes than AWS, automatic discounting for sustained usage — can result in a lower effective cost even when headline rates appear comparable.
Pricing Architecture: Understanding What You Are Actually Paying For
Direct pricing comparison between the three platforms requires understanding how each bills for comparable workloads. For standard compute (a 2 vCPU, 8 GB RAM configuration), on-demand list prices are broadly comparable across the three platforms, with minor differences between regions and instance families. The real differentiation emerges in the commitment discount structures, which represent the primary lever for managing cloud spend at enterprise scale.
AWS: Reserved Instances vs Savings Plans
AWS offers two primary commitment-based discount mechanisms, and understanding the difference between them is essential for cost optimisation.
Reserved Instances (RIs) provide discounts of up to 75% compared to on-demand rates in exchange for a one-year or three-year commitment to a specific instance type, operating system, and region. Standard Reserved Instances provide the deepest discounts but offer no flexibility — you are committed to exactly the instance type, size, and location you reserved. Convertible Reserved Instances offer slightly lower discounts (up to 66%) but allow you to exchange the reservation for a different instance type during the term. Reserved Instances are powerful cost optimisation tools for stable, predictable workloads, but they require accurate capacity forecasting. Unused Reserved Instance capacity is still charged at the Reserved Instance rate regardless of whether it is used.
Savings Plans are the more flexible alternative. Compute Savings Plans provide discounts of up to 66% in exchange for a commitment to a specific dollar amount of compute spend per hour, but allow that spend to apply across any EC2 instance family, size, region, operating system, or tenancy, as well as Lambda and Fargate. EC2 Instance Savings Plans provide higher discounts (up to 72%) but are tied to a specific instance family and region. Savings Plans have replaced Reserved Instances as the preferred commitment mechanism for most organisations because the flexibility to apply savings across changing workloads is more valuable than the marginally higher discount available from Standard RIs.
The key distinction between RIs and Savings Plans for enterprise buyers is this: Reserved Instances provide the deepest discounts for workloads you are highly confident will remain stable in their resource profile. Savings Plans provide nearly equivalent discounts with much greater flexibility for workloads that may change in instance type, size, or location over the commitment period. Most mature AWS cost management programmes use a combination of both, reserving the most stable workloads with Standard RIs and covering the remaining compute commitment with Savings Plans.
Azure: Reserved VMs, Savings Plans, and Hybrid Benefit
Azure's commitment discount architecture parallels AWS's in structure. Azure Reserved VM Instances (RVMIs) provide discounts of up to 72% in exchange for one-year or three-year commitments to specific VM sizes and regions. Azure Savings Plans for Compute — launched in 2022 — provide flexibility comparable to AWS Compute Savings Plans, applying discounts across any VM family, region, and operating system within a committed hourly spend threshold.
Azure's distinctive advantage is the Hybrid Benefit. Organisations with existing Windows Server licences with Software Assurance can apply those licences to Azure VMs, eliminating the Windows Server licensing component of the VM price — typically a 40% reduction in VM cost for Windows workloads. An equivalent benefit applies to SQL Server licences. For enterprises renewing their Microsoft licence agreements concurrently with managing Azure spend, maximising Hybrid Benefit coverage is one of the highest-value cost optimisation actions available. Hybrid Benefit savings compound with Reserved VM discounts, making the effective cost of Windows workloads on Azure materially lower than headline on-demand pricing suggests.
GCP: Committed Use Discounts and Sustained Use
Google Cloud offers Committed Use Discounts (CUDs) that provide up to 57% off on-demand rates in exchange for one-year or three-year commitments to specific resource amounts (vCPUs and memory). As noted above, GCP's Sustained Use Discounts provide an automatic 30% discount for workloads running more than 25% of the month without any reservation requirement. The combination of automatic sustained use discounting and committed use discounts means that GCP's effective cost for steady-state workloads is often lower than on-demand pricing implies, and the optimisation overhead required to achieve those savings is lower than the equivalent effort on AWS.
GCP also offers a negotiated discount tier — the Cloud Committed Use Agreement — for large enterprise customers, similar to the AWS EDP and Azure MACC. These agreements provide additional percentage discounts in exchange for spend commitments, and are available to organisations with significant GCP footprint.
Data Egress: The Cost That Consistently Surprises Enterprises
Data egress is the single most common unexpected cost category across all three major cloud platforms, and it deserves specific attention in any enterprise cloud cost analysis. Every time data leaves a cloud platform — to the internet, to another cloud provider, to an on-premises data centre, or to users in another geography — the platform charges for that outbound transfer at rates that accumulate quickly for data-intensive workloads.
Standard internet egress rates for the three platforms are broadly comparable: AWS charges approximately $0.09 per GB for the first 10TB monthly from US East (us-east-1), with similar pricing in other regions. Azure charges $0.087 per GB for the first 10TB monthly from most regions. GCP charges $0.08–$0.12 per GB depending on destination and region. Data transfer between availability zones within the same region incurs lower charges (typically $0.01–$0.02 per GB), while cross-region data transfer is charged at intermediate rates.
The organisations that most frequently experience egress cost surprises are those that have not explicitly mapped their data flows during architectural design. Common high-egress scenarios include: applications that retrieve large data payloads from cloud storage and return them to on-premises systems or end users, analytics pipelines that move data between cloud regions or between cloud platforms (inter-cloud transfer), AI and machine learning workloads that retrieve large training datasets or return large inference outputs across network boundaries, and content delivery architectures that serve assets directly from cloud storage to internet-facing users at scale rather than through CDN.
For enterprises managing multi-cloud environments — running workloads on both AWS and Azure, for example — cross-platform data transfer costs are particularly significant because each platform charges for outbound transfer, creating a double-billing effect for every workload that moves data between them. Designing data flows to minimise cross-platform transfer, or consolidating workloads that interact heavily with each other on the same platform, is the most effective mitigation.
Multi-cloud spend getting hard to manage?
We review AWS, Azure, and GCP commercial agreements independently — buyer-side only.Enterprise Negotiation: EDP, MACC, and Cloud Committed Use Agreements
All three major cloud platforms offer negotiated enterprise discount programmes that provide committed spend discounts beyond what is available through standard Reserved Instance or Savings Plan mechanisms. Understanding how each works — and what it takes to achieve meaningful discounts — is essential for enterprise cloud buyers.
AWS Enterprise Discount Program (EDP)
The AWS EDP is a multi-year spend commitment agreement that provides a blanket discount percentage across eligible AWS services in exchange for a defined annual spend commitment, typically ranging from one to three years. Unlike Savings Plans and Reserved Instances, which discount specific compute services, the EDP applies a percentage discount across the entire eligible AWS bill — compute, storage, database, AI services, data transfer, and more. Bedrock API calls, S3, RDS, and most other managed services are eligible for EDP discounts.
The practical threshold for meaningful EDP discounts is approximately $2 million or more in annual AWS spend. Below this level, EDP discounts are typically in the 6–9% range, which may not justify the contract complexity and commitment risk. At $2 million and above, discounts in the 10–15% range are achievable on one-year agreements; three-year agreements at $2M+ can secure 15–20% discounts. Above $5 million in annual commitment, discounts of 20–25% or more are available with appropriate negotiation and willingness to commit to a longer term.
The EDP has important constraints that buyers must understand. AWS requires that annual commitments do not decrease year-over-year — if you commit to $2 million in year one, you cannot reduce to $1.75 million in year two of the agreement. This commitment ratchet means that organisations should model their AWS spend trajectory conservatively when sizing their EDP commitment. Overcommitting to the EDP and then failing to meet the spend threshold results in payment for unused committed spend. Additionally, EDP participation requires that all accounts under the payer account be enrolled in AWS Enterprise Support, which carries a minimum cost of $15,000 per month (3% of monthly AWS spend, whichever is higher). The Enterprise Support cost is a real incremental expense that partially offsets the EDP discount — factor it into your net savings calculation.
AWS Marketplace purchases of ISV products and services can contribute up to 25% of your annual EDP commitment, which is valuable for organisations that procure significant ISV software through the Marketplace. Confirming this treatment with your AWS account team when structuring your EDP is recommended.
Azure MACC (Microsoft Azure Consumption Commitment)
The Azure MACC is Microsoft's enterprise cloud commitment programme, and it differs from the AWS EDP in one important way: MACC commitments can be integrated with the broader Microsoft EA commercial relationship. This means that organisations negotiating an EA renewal can structure Azure consumption commitments as part of the same commercial negotiation, potentially trading Azure commitment volume for better EA pricing on Microsoft 365, Dynamics, or other Microsoft products, or receiving concessions on Azure pricing in exchange for expanded Microsoft on-premises licence coverage.
This integration with the broader Microsoft commercial relationship is both the primary advantage and the primary risk of MACC versus standalone cloud commitment programmes. The advantage is the commercial flexibility to use multi-dimensional Microsoft spend leverage in a single negotiation. The risk is that bundling Azure commitments into an EA negotiation can obscure the individual economics of each component, making it difficult to assess whether the Azure commitment terms are independently competitive.
MACC commitments include Azure services specifically and do not cover Microsoft 365, Dynamics, or other Microsoft SaaS products. Minimum MACC amounts vary by agreement structure, but meaningful discount discussions typically begin at annual Azure spend in the $1 million or above range. Unlike the AWS EDP, MACC commitments are typically expressed as a dollar amount to be consumed from a prepaid balance rather than as a percentage discount — Microsoft draws down the MACC balance against your Azure consumption at standard rates (with any applicable Reserved VM and Savings Plan discounts already applied).
GCP Cloud Committed Use Agreements
Google Cloud's enterprise discount programme — variously referred to as Cloud Committed Use Agreements, Private Pricing Agreements, or simply negotiated contracts — provides discount structures similar to AWS and Azure for large-spend customers. Negotiated GCP discounts of 6–15% on spend commitments in the $500,000 to $5 million range are available, with larger discounts accessible at higher commit levels. GCP's commercial approach has historically been more aggressive in pursuing enterprise customers from AWS and Azure, meaning that competitive displacement scenarios — where an organisation is evaluating migrating workloads from AWS or Azure to GCP — often produce more favourable starting discount offers than incumbent renewal negotiations.
GCP also offers sustained use discounts and committed use discounts that stack with negotiated contract discounts, which can create effective savings rates that meaningfully exceed what is visible in the headline committed use pricing.
Workload Allocation: Which Platform for Which Workload
Most large enterprises run workloads on more than one cloud platform, and workload allocation decisions significantly affect both cost and operational efficiency. The framework below summarises where each platform typically provides the strongest fit, based on our experience supporting enterprise cloud commercial decisions.
AWS Strengths
AWS is the strongest platform for workloads that require the broadest service selection, organisations that are building cloud-native infrastructure from scratch without significant legacy Microsoft investment, workloads that benefit from AWS's global availability zone density and edge network, and technology-forward organisations that prioritise access to leading-edge managed services at scale. AWS is also typically the platform of choice for independent software vendors (ISVs) and technology companies that sell to other enterprises, because AWS's market share makes it the largest addressable deployment environment.
Azure Strengths
Azure is strongest for organisations with significant existing Microsoft licence investment — particularly those running Windows Server, SQL Server, Active Directory, and Microsoft 365 at scale — because Hybrid Benefit and native integration reduce effective cost and integration complexity for those workloads. Azure is also the natural home for Microsoft-centric application development (Visual Studio, Azure DevOps, .NET applications) and for organisations where the Copilot platform and Microsoft AI services are part of the enterprise roadmap. Azure Government and sovereign cloud options are well-developed, making Azure a strong fit for government and regulated industries in jurisdictions where Microsoft has established data residency commitments.
GCP Strengths
GCP is strongest for analytics-heavy workloads (BigQuery has a genuine performance and ease-of-use advantage for large-scale data analytics), AI and machine learning workloads that benefit from Google's TPU infrastructure and Vertex AI platform, organisations building Kubernetes-native applications (GKE remains the leading managed Kubernetes service), and organisations where Google Workspace is the primary productivity platform and native integration with GCP reduces operational overhead. GCP is also increasingly competitive for organisations building generative AI applications, particularly those evaluating Gemini models as an alternative to OpenAI/Azure OpenAI or Anthropic/Bedrock.
The Hidden Cost Categories That Affect Every Platform
Data Transfer and Egress (Revisited)
We have addressed data egress above, but it bears repeating as the single highest-impact cost category that enterprise organisations consistently underestimate. Every major cloud cost review we conduct identifies data egress as a top-five cost driver that was not adequately modelled in the original cloud business case. Build egress cost modelling into every cloud architecture review and specifically quantify the monthly egress cost for every data flow that crosses the cloud network boundary.
Support Tiers
Enterprise-grade support from all three platforms carries significant cost. AWS Enterprise Support is mandatory for EDP customers, at 3% of monthly spend or $15,000 minimum monthly. Azure's Unified Support has moved to a consumption-based model that can represent 10% or more of Azure spend for large customers. GCP's Enhanced or Premium Support is similarly priced as a percentage of spend. When evaluating cloud platform total cost of ownership, the support tier cost should be explicitly modelled — particularly for EDP/MACC customers where it may be mandatory.
API Call and Request Charges
Many AWS managed services charge separately for API calls, requests, and operations in addition to the underlying storage or compute consumption. S3 charges per PUT, GET, and COPY request. DynamoDB charges for read and write capacity units. Lambda charges per invocation and per GB-second of execution time. Azure and GCP have equivalent per-operation charges for many managed services. These granular operational charges accumulate in ways that are difficult to forecast from first principles, particularly for applications with high transaction rates. Investing in application-level instrumentation that captures API call volumes before deployment to production allows for more accurate cost modelling.
Third-Party Licensing on Cloud Infrastructure
Running commercial software on cloud infrastructure — Oracle Database, SQL Server with Bring Your Own Licence, SAP HANA, or other licensed applications — introduces licensing cost considerations that are specific to the cloud deployment model. Oracle's licensing rules for cloud environments are particularly complex and have generated significant audit risk for organisations that moved Oracle workloads to cloud without appropriate licence analysis. Understanding which commercial software products have cloud-specific licensing terms, and how those terms interact with your existing licence entitlements, is an important part of cloud total cost of ownership modelling.
Need an independent review of your cloud commercial agreements?
We negotiate AWS, Azure, and GCP agreements for enterprise buyers globally. Buyer-side only, no cloud provider affiliation.Building Your Cloud Commercial Strategy
Effective cloud commercial management requires a governance framework that addresses platform selection, commitment sizing, cost visibility, and ongoing optimisation. The following principles apply regardless of which platform or combination of platforms you are managing.
Establish Baseline Cost Visibility Before Committing
The single most common cloud commercial mistake is signing a multi-year EDP, MACC, or CUD agreement before establishing accurate baseline cost visibility. Committing to $3 million in annual AWS spend before you have instrumented your current consumption accurately creates the risk of over-committing (paying for unused commitment) or under-committing (missing discount thresholds that would justify the agreement). Spend at least 90 days establishing tagged, attributed cloud cost reporting before sizing a multi-year commitment.
Use Competitive Process to Establish Pricing Tension
All three major cloud platforms are responsive to competitive pressure. Organisations that obtain competitive proposals from two or three platforms — even if they have no genuine intention of migrating workloads — consistently achieve better terms in their primary platform negotiation. The credibility of the competitive alternative does not need to be complete: a scoping exercise for a Greenfield workload deployment on an alternative platform, or a proof-of-concept for a specific workload, is sufficient to establish that your organisation is a buyer with genuine platform optionality. This shifts the dynamic of every subsequent pricing conversation.
Align Cloud Commitments with Procurement Cycles
Cloud platform commercial terms interact with other major IT procurement cycles. For Azure, the optimal negotiating moment is typically concurrent with Microsoft EA renewal, because both conversations involve the same Microsoft account team and the same commercial decision-makers. For AWS, negotiating the EDP in advance of a major infrastructure expansion (a new application deployment, an acquired company's infrastructure migration) provides commitment sizing leverage that is not available during steady-state operations. Timing cloud commercial negotiations to coincide with workload growth inflection points consistently produces better outcomes than renewing agreements simply because they expire.
Build FinOps Capability Before Scaling Cloud Spend
Cloud financial management — FinOps — is the organisational discipline of managing cloud spend with the same rigour applied to other operating cost categories. Enterprises that scale cloud spend without corresponding FinOps capability consistently discover that waste, untagged resources, unused reservations, and unmanaged egress accumulate in ways that erode the commercial advantage of even well-negotiated EDP or MACC agreements. The investment in FinOps tooling and capability — whether through internal hire, training, or external advisory — is justified at any cloud spend level above approximately $1 million annually and becomes essential above $5 million.
For more on multi-cloud commercial strategy, visit our GenAI Knowledge Hub and speak with our AWS contract negotiation specialists and Google Cloud advisory team.
Conclusion: The Framework for Informed Cloud Platform Decisions
The AWS, Azure, and GCP comparison is not a question with a single correct answer. Each platform has genuine strengths, and the right allocation of workloads depends on your existing technology stack, your organisational Microsoft investment, your AI and data strategy, and your appetite for commitment versus flexibility. What is clear from our engagement with enterprise cloud buyers is that the organisations achieving the best commercial outcomes are those that approach cloud platform decisions with the same commercial discipline applied to other major enterprise software investments: structured competitive evaluation, accurate baseline cost modelling, informed commitment sizing, and proactive negotiation engagement that does not wait for the platform to propose terms.
Data egress will remain a cost trap that surprises organisations that do not map their data flows explicitly. The Reserved Instance versus Savings Plans distinction will continue to matter for AWS cost optimisation. The EDP meaningful discount threshold — approximately $2 million in annual AWS commit — will continue to define which organisations can access the commercial terms that large-scale AWS users require. And the integration advantage of Azure for Microsoft-centric organisations will continue to be a genuine factor that affects total cost of ownership beyond headline compute pricing. Understanding all of these dimensions, and building a commercial framework that addresses them systematically, is the foundation for managing cloud spend as a strategic asset rather than an uncontrolled operating cost.