Why cloud infrastructure prices resist competition despite multiple providers

Three major cloud providers—Amazon Web Services, Microsoft Azure, and Google Cloud—compete aggressively for enterprise infrastructure business, yet enterprise computing costs have remained remarkably sticky. Despite this competition, cloud infrastructure spending by U.S. businesses has reached levels where enterprises increasingly explore alternatives like colocation data centers specifically to reduce costs. In September 2026, Data Center Knowledge reported that enterprises were strategically choosing colocation facilities to host AI applications and hybrid cloud deployments specifically for cost reduction alongside infrastructure scaling.

The reason prices stick despite competition lies in how cloud providers structure their offers. Rather than competing primarily on per-unit compute or storage rates, they use a tiered discount system that rewards long-term commitments and high volume—mechanisms that inadvertently lock customers into single providers and make switching expensive. This pricing architecture emerged across all three major cloud providers: AWS, Azure, and Google Cloud have converged on similar economic models that favor commitment and consolidation.

How commitment discounts create multi-year lock-in

Each major cloud provider offers the same basic purchasing tiers: on-demand pricing where customers pay per unit consumed without advance commitment, and discounted rates for committed usage over one to three-year terms. AWS calls these Savings Plans; Azure uses Reserved Instances; Google Cloud offers Committed Use Discounts alongside automatic Sustained Use Discounts for resources that run continuously. The savings can be substantial—AWS documentation indicates that customers who commit to Savings Plans receive lower per-unit costs compared to on-demand pricing, offering significant savings over on-demand pricing.

These discounts create a powerful lock-in mechanism by design. An enterprise that commits $500,000 over three years to AWS virtual machines has spent money that becomes partially sunk if the company later wants to switch to Azure or Google Cloud. The committed discount applies only to the provider offering it. Unlike a subscription that can be cancelled, a three-year commitment creates a financial obligation that persists regardless of competitive offerings or business changes. A company midway through a three-year commitment faces the choice of writing off remaining discount value or continuing to use the expensive provider. This dynamic affects not just the direct cost comparison but also the psychology of the purchasing decision—executives who approved a three-year commitment have organizational incentive to see that commitment justified.

Microsoft Azure structures this similarly. Azure offers Reserved Virtual Machine Instances in exchange for reduced rates. The company also offers an Azure Hybrid Benefit allowing customers who own existing Windows Server licenses with Software Assurance to reuse those licenses on Azure virtual machines, creating additional switching friction. Google Cloud's Committed Use Discounts work identically. For storage services like S3, AWS uses tiered pricing where higher monthly usage levels receive lower per-gigabyte rates, creating volume tiers that increase the cost of splitting data across providers.

This structure exists across all three major providers' pricing models, meaning switching incurs both the immediate cost of the unused commitment and the expense of re-architecting applications to work with a different provider's services. For large enterprises, re-architecting can take months or years. A company running containerized applications on AWS requires the team to understand how to convert those workloads to Azure's equivalent services, test them, manage the transition, and support staff through the migration—costs that often exceed the financial discount available from the competing provider.

Volume-based pricing concentrates usage within one vendor

Cloud providers also use volume discounts—lower rates per unit as consumption increases. AWS's storage service S3, for example, uses tiered pricing where higher monthly usage levels receive lower per-gigabyte rates. Higher monthly usage levels receive lower per-gigabyte rates. This creates a visible economic penalty for customers who split their storage across providers. A company running 100 terabytes across multiple providers—say, 50 terabytes each with AWS and Azure—pays more per terabyte than a company running all 100 terabytes with one provider.

This structure economically pressures enterprises to consolidate their infrastructure with a single provider rather than pursuing multi-cloud strategies. The incentive to concentrate usage with one vendor is not arbitrary; it emerges directly from the pricing structure itself. Azure uses similar tiered pricing for its storage services, with rates declining as usage increases within a single account. Google Cloud applies this same principle across its service offerings. Once an enterprise reaches high-volume thresholds with one provider, the discount difference between staying put and splitting usage elsewhere becomes prohibitively expensive.

Multi-cloud strategies, which might offer resilience by avoiding dependence on a single provider or flexibility by using best-of-breed services from each vendor, become economically disadvantageous. Enterprises pay a premium for architectural choices that the market might otherwise reward as beneficial for competition or risk management. A company that wants to use AWS for compute because it has the strongest machine learning services, Azure for databases because of its SQL Server integration, and Google Cloud for data analytics because of BigQuery's price-performance ratio cannot do so cost-efficiently. The volume discounts at each provider are lower for customers splitting usage. The company ends up paying more total infrastructure cost to pursue architectural diversity.

This volume-penalty mechanism means that enterprises face constant economic pressure to consolidate with their largest provider. Growth that might otherwise trigger competitive reevaluation instead drives deeper commitment to the incumbent. A customer whose usage doubles naturally receives better pricing discounts—but only if all the growth stays with the same provider. This creates a compounding advantage for the incumbent cloud provider within each customer account.

Integration depth makes migration technically complex

Beyond contractual and pricing mechanisms, cloud customers become locked in through technical integration at the application level. Applications built on AWS Lambda functions for serverless computing, Google Cloud's BigQuery data warehouse for analytics, or Azure's managed services like Cosmos DB use cloud-native APIs and features that are difficult to replicate elsewhere. Migrating these workloads means rewriting application code, not just moving virtual machines.

A company using AWS Lambda to run event-driven functions might have hundreds of Lambda functions integrated with AWS's SNS messaging service, DynamoDB database, and S3 storage—all AWS services designed to work together seamlessly. Moving that same architecture to Azure requires converting those Lambda functions to Azure Functions, SNS to Azure Service Bus, DynamoDB to Cosmos DB, and S3 to Azure Blob Storage. Each conversion involves code changes, testing, and potential performance differences. The application architecture has become AWS-specific rather than cloud-agnostic.

This integration depth means that the effective switching cost includes not just the financial penalty of breaking a commitment, but the engineering effort required to rebuild applications on a competing platform. A business unit with a team of five engineers spending six months on migration faces roughly $750,000 in labor costs alone (assuming fully-loaded cost of $250,000 per engineer annually), before counting any application downtime, testing overhead, or mistakes during the transition. For many enterprises, this technical cost exceeds the financial discount benefit of switching to a cheaper provider—even if such a provider existed and offered significantly lower prices.

How competition functions within pricing constraints

Given these structural constraints, competition among cloud providers functions differently than in markets without switching costs. When a customer is locked in through multi-year commitments, volume tiers, and technical integration, the ability to bid aggressively on price to win new customers must be weighed against the reality that many large customers are unavailable—they're under contract with a competitor or embedded in that competitor's services.

New customers and workloads represent the main competitive battleground. Cloud providers can offer discounts to new customers or for new workloads that aren't yet integrated into a competitor's ecosystem. However, the majority of enterprise cloud spending goes to existing workloads, not new deployments. A company that has been running applications on AWS for five years has most of its infrastructure already committed and integrated. The opportunity to win that customer's infrastructure spending is limited—the provider can potentially win incremental new projects or applications, but converting the installed base is expensive.

Enterprises seeking alternatives to expensive cloud models have increasingly turned to colocation data centers, where they can purchase raw compute capacity and manage infrastructure more directly. Data Center Knowledge reported in September 2026 that enterprises were strategically choosing colocation facilities to host AI applications, deploy hybrid cloud models, add incremental capacity, improve connectivity, and reduce costs. This shift suggests that cloud pricing, despite theoretical competition among three large players, has not reached levels that many enterprises consider optimal for all workloads.

New entrants challenge traditional pricing models

Newer entrants to cloud infrastructure have experimented with different pricing approaches. CoreWeave, an AI cloud provider that emerged specifically to serve machine learning workloads, introduced flexible pricing models moving away from the fixed multi-year terms that dominate AWS, Azure, and Google Cloud. According to Data Center Knowledge reporting in March 2026, CoreWeave's flexible AI cloud pricing model represented a strategic shift in how infrastructure providers approach customer contracts. Rather than requiring three-year commitments, CoreWeave offered terms aligned more closely to the uncertain demand cycles of AI infrastructure.

This experimentation suggests that traditional cloud pricing structures still leave room for competitors to win customers through different financial terms. However, the three incumbent providers' market dominance means most enterprise customers remain within traditional pricing models. AWS, Microsoft, and Google control the vast majority of the enterprise cloud market, and their pricing structures have become the industry standard that most enterprises expect and accept.

For the largest enterprises, direct negotiation with cloud providers offers some relief from standard pricing. AWS documentation indicates that large customers with substantial commitments can negotiate private pricing arrangements across 200 or more eligible services. These private deals, however, further entrench existing relationships—a company that negotiates a custom pricing arrangement with AWS has even more embedded value to protect from switching. For mid-sized and smaller enterprises without negotiating leverage, standard pricing controls remain in place. They face the full force of lock-in through commitments and volume-based incentives to concentrate usage.

Why structural costs discourage provider switching

The economic and technical barriers to switching create a market where price competition remains theoretical rather than practical for most enterprise customers. An enterprise comparing AWS to Azure for a new workload might find Azure offers a lower per-unit rate. But that same company cannot easily switch its existing AWS workloads because of commitments, volume discounts lost through splitting usage, and the cost of re-architecting applications. The competitive analysis becomes not whether Azure is cheaper in absolute terms, but whether Azure is cheaper enough to justify the switching costs—a much higher bar.

This explains why cloud pricing has remained sticky despite competition. The market has three large competitors, but the structural features of how they price create effective monopolies at the customer level. Each customer is locked into its primary provider in ways that go beyond simple habit or switching costs. The lock-in is encoded into the pricing structure itself, creating a situation where competition exists at the margins but not at the core of enterprise cloud spending.