The cloud pricing shift

In early 2025 AWS, Microsoft Azure, and GCP all rolled out major changes to how they price their services. It didn’t happen all at once, or in the same way, but there is a distinct pattern.

Each provider announced pricing changes framed as customer-friendly. Each presented new savings opportunities and flexibility. And each introduced price increases in places that matter more than the headlines admit.

The messaging, as you would expected, focused on discounts. AWS announced up to 85% price reductions for S3 Express One Zone, including lower storage costs, cheaper PUT and GET requests, and reduced data transfer rates. GCP reduced pricing for object storage and persistent disk in select regions. Microsoft quietly adjusted regional pricing for some SaaS products, cutting prices by 5-6% in places like the UK. But, as it often is - the real story is in the details.

Where prices are going up

At the same time AWS announced its storage discounts, it increased the hourly cost of running EKS clusters on older Kubernetes versions from 10 cents to 60 cents per hour. That’s a sixfold increase for users who haven’t upgraded to the latest versions.

Microsoft announced that starting in April 2025, all monthly billing options for annual or three-year license subscriptions will cost 5% more than the same plan billed annually. That means companies that cannot afford to pay upfront face a baked-in premium.

At the product level, Microsoft also raised prices on several core offerings. “Teams Phone Standard”, for example, increased from 8 to 10 dollars per user per month. Power BI Pro rose from 120 to 168 dollars per user per year. On-premises server products including SharePoint, Exchange, and Skype for Business saw price increases of 10%. CAL Suites went up between 15-20%.

GCP also made changes. While it cut prices in a few storage tiers, it raised them in others. Dual-region and Coldline storage became more expensive. Premium GPU instances and advanced AI services moved into new, higher-cost pricing tiers. In parallel, AWS increased outbound data transfer prices in select regions and introduced premium pricing for high-performance AI instances such as Trainium and Inferentia.

A pattern emerges

The pricing changes are not uniform, but they share a strategy: keep core compute and base storage services affordable, and make a show of discounting new or niche offerings. Then quietly raise prices on the services that matter most, such as managed Kubernetes, scalable storage, AI compute, cross-region networking, or support for older platforms.

At the same time, providers are reshaping the way discounts work. AWS updated its Reserved Instances and Savings Plans policies to restrict usage to a single end customer, effectively removing the ability for resellers or managed service providers to pool purchases on behalf of multiple clients. This change, which took effect on June 1, 2025, effectively prevents resellers and managed service providers from pooling purchases across multiple clients. As a result, smaller organizations may lose access to discounts that previously helped them stay competitive.

This is a structural change in how providers manage margins, enforce standardization, and segment their customers.

The myth of multi-cloud as leverage

For smaller businesses, the impact is already visible. If you can’t pay annually, you now pay more. If you rely on resellers to access volume discounts, you no longer can. If you’re still running older versions of core services like Kubernetes, MySQL, or PostgreSQL, expect to face extended support fees or enforced upgrades.

Need the flexibility to scale workloads up and down? You’re likely paying premium on-demand rates. Want to integrate with AI services or use high-performance GPUs? That’s no longer part of the standard pricing model. Everywhere you turn, flexibility is getting monetized.

For years, multi-cloud was positioned as a solution to lock-in. The idea was that if you can run workloads across providers, you can force them to compete on price. But in 2025, that idea is harder to justify.

The latest Canalys report shows global cloud spending is projected to grow 19% in 2025. But that growth is not purely organic - though some of it is driven by expanded usage,parts of it is a result of rising per-unit pricing across premium services and new fees layered into existing consumption models.

When AWS, Azure, and GCP all adjust pricing in the same directions, (although differently), multi-cloud stops being about leverage, and becomes about choosing which provider overcharges you in which category.

The end of the competitive cloud era

Though it might feel like that - what we see is not collusion, but market maturity. The land grab is over. The major cloud providers have carved out their territory, and they are monetizing it accordingly. 

Compute and object storage remain relatively cheap. These are the services that draw customers in and maintain the illusion of price competition. But the real money flows through premium services: AI training, global-scale inference, real-time data pipelines, high-throughput analytics, or versioned storage. All of it is becoming more expensive, and often, not optional.

The providers have learned that it no longer needs to compete on price in these areas. Switching costs are high. Optimization is difficult. And all three have moved toward monetizing the complexity they created.

The role of FinOps

In this environment, FinOps becomes more than just a cost optimization practice. It becomes a survival tool. Cloud cost management today requires understanding dynamic pricing models, anticipating platform-level changes, and proactively managing usage. But when pricing structures shift in parallel across providers, optimization alone can’t keep up. You can’t out-optimize structural cost pressure.

Some organizations are deploying FinOps tools provided by the same cloud providers raising their prices. This creates an odd dynamic, the same vendor that introduces complexity offers a tool to manage it. Useful, yes. But ironic nonetheless.

Where we stand

In 2025, cloud pricing is no longer simple, or predictable. Every provider is rewriting the rules, every discount comes with conditions, and every flexibility feature has a catch.

For customers, that means adapting, for smaller teams, that might mean stricter usage policies or delayed adoption of advanced workloads. For enterprises, it could mean renegotiating contracts, shifting architectural decisions, or accelerating FinOps maturity.

What’s clear is that the age of infinite cloud flexibility is ending, and the next phase will reward those who understand how to build in the cloud, and how it’s priced, governed, and shaped to protect the provider’s margins, first.