Multi-cloud cost optimization: a practical playbook for AWS, Azure and Google Cloud
Country Manager, Sweden
AI, DevOps, Security, and Cloud Solutioning. 12+ years leading enterprise cloud transformation across Scandinavia
Multi-cloud cost optimization means running AWS, Azure and Google Cloud as one budget: normalise every provider's billing into one dataset, tag spend to owners, right-size and remove idle resources, buy commitments per provider only for steady load, and control data transfer between clouds. FinOps practices and a FOCUS-format cost export make this manageable.
Key takeaways
- Waste is real and growing: respondents to Flexera's 2026 State of the Cloud survey estimate 29% of their cloud spend is wasted, the first increase in five years.
- Start with one cost view. AWS, Microsoft and Google Cloud all export billing in the FinOps Foundation's FOCUS format, so you can compare like with like.
- Commitments save the most but lock you in. AWS Savings Plans and Azure reservations offer up to 72% off on-demand prices; size them per provider on steady baseline usage only.
- Data transfer is the hidden multi-cloud cost. Design to keep chatty workloads in one cloud.
- In the EU, the Data Act caps egress fees for simultaneous multi-cloud use at cost from 12 September 2025 and bans switching charges from 12 January 2027.
Why is cost harder to control in a multi-cloud setup?
Each provider bills differently, names services differently and discounts differently. Without a common view, nobody can say what a product or team costs across clouds.
- Different billing data. AWS, Azure and Google Cloud use their own fields, units and discount models, so totals cannot be added up directly.
- Split ownership. One team runs Azure for Microsoft workloads, another runs AWS for the product, and data science works in Google Cloud. Each sees only its own bill.
- Duplicated services. Monitoring, backup, security and networking are often paid for twice.
- Fragmented discounts. Volume spread over three providers earns smaller discounts than the same volume on one.
- Data transfer. Traffic leaving one cloud for another is billed as egress, and it grows quietly as integrations multiply.
Waste is not falling on its own. Flexera's 2026 State of the Cloud Report found that 85% of organisations name managing cloud spend as a top challenge, and that self-estimated waste rose to 29%, driven by AI workloads.
How to build one cost view across AWS, Azure and Google Cloud
The FinOps Foundation's FOCUS specification (FinOps Open Cost and Usage Specification) defines one schema for billing data. AWS, Microsoft and Google Cloud all publish FOCUS exports, so you can load all three into one table and query them together.
- Turn on the exports. On AWS, create a FOCUS export in Data Exports. On Azure, use the FOCUS cost and usage dataset in Cost Management. On Google Cloud, export billing to BigQuery and use the FOCUS view.
- Land them in one place, for example a data warehouse or a FinOps platform.
- Apply one tagging standard. Agree mandatory tags or labels such as cost centre, product, environment and owner, and enforce them with policy (AWS tag policies, Azure Policy, Google Cloud organisation policies).
- Allocate shared costs such as networking, support plans and shared clusters by a documented rule.
- Publish dashboards per team, with budgets and anomaly alerts, so owners see their own spend weekly.
Which optimisation levers work in every cloud?
| Lever | AWS | Azure | Google Cloud |
|---|---|---|---|
| Commitment discounts | Savings Plans (up to 72% vs on-demand), Reserved Instances | Reservations (up to 72% vs pay-as-you-go), Azure savings plan for compute | Committed use discounts, spend-based or resource-based, 1 or 3 years |
| Spare capacity | EC2 Spot Instances | Spot Virtual Machines | Spot VMs |
| Right-sizing advice | AWS Compute Optimizer | Azure Advisor | Recommender (Active Assist) |
| Budgets and alerts | AWS Budgets, Cost Anomaly Detection | Cost Management budgets and anomaly alerts | Cloud Billing budgets and alerts |
| Licence reuse | Bring your own licence where terms allow | Azure Hybrid Benefit for Windows Server and SQL Server | Bring your own licence on sole-tenant nodes where terms allow |
Use the levers in this order:
- Remove waste. Delete idle instances, unattached disks, old snapshots and unused IP addresses.
- Schedule non-production. Development and test environments rarely need to run nights and weekends.
- Right-size. Move to smaller or newer instance families based on measured CPU and memory, not requested size.
- Modernise where it pays. Autoscaling, serverless and managed databases reduce idle capacity.
- Commit last. Buy Savings Plans, reservations or committed use discounts for the baseline that remains after steps 1–4. Commit too early and you lock in waste.
How do you control data transfer costs between clouds?
Place workloads that talk to each other constantly in the same cloud and region. Keep cross-cloud traffic to batch data flows, backups or user-facing failover, where the volume is predictable.
- Map the biggest data flows between providers and on-premises before choosing where each workload runs.
- Compress and batch transfers, and cache at the edge where you can.
- Use private interconnects (AWS Direct Connect, Azure ExpressRoute, Google Cloud Interconnect) when volumes justify the fixed cost.
- Check the transfer line in your FOCUS data monthly; it is often the line that grows fastest.
EU law now helps. Under the Data Act, providers may not charge egress fees above their actual transfer costs when services are used simultaneously, since 12 September 2025, and switching charges are banned from 12 January 2027. France's telecoms regulator Arcep summarises the Article 29 and Article 34 rules in its February 2026 consultation. Ongoing data transfer is still billed, so architecture still matters.
How do FinOps practices apply to multi-cloud?
The FinOps Framework organises the work in three phases: Inform (visibility and allocation), Optimize (rates and usage) and Operate (governance and continuous improvement). In multi-cloud the phases are the same; the difference is that one central team must own the common data model, tagging rules and commitment strategy, while engineering teams own usage in their own cloud.
- Inform: FOCUS data, tagging, showback or chargeback per team.
- Optimize: right-sizing and scheduling by teams; commitment purchases centrally, per provider.
- Operate: monthly cost reviews, budgets in the planning cycle, unit costs such as cost per customer or per transaction.
Should you use a native tool or a third-party platform?
| Option | Fits when | Trade-off |
|---|---|---|
| Native tools per cloud (Cost Explorer, Azure Cost Management, Cloud Billing reports) | One dominant cloud and a small second one | No single view; three sets of reports |
| FOCUS exports into your own data warehouse and BI | You have data engineering capacity and want full control | You build and maintain the allocation logic |
| Third-party FinOps platform | Several clouds, many teams, need for recommendations and automation | Licence cost, often a share of spend; check FOCUS support and data residency |
Whichever you choose, the tool only shows the savings. People still have to act on them, so assign owners and track realised savings, not recommendations.
A real example
Branäsgruppen, one of Sweden's largest ski resort groups, moved from a manually scaled VMware estate to AWS with Opsio. Pairing right-sized auto-scaling with disciplined FinOps practices delivered a 40% reduction in total cost of ownership, alongside 40% faster average page response and 99.9%+ application uptime. The Branäsgruppen case study describes the setup. The same practices apply in each cloud of a multi-cloud estate.
How Opsio helps
Opsio runs FinOps and cost optimisation for companies on AWS, Azure and Google Cloud: one cost view, tagging and allocation, right-sizing, commitment planning and monthly reviews. See our FinOps service, cloud cost optimisation services and Azure cost management. For AWS-specific steps, read our AWS Cost Explorer guide.
Frequently asked questions
What is multi-cloud cost optimization?
It is the practice of managing spend across several cloud providers as one budget: a common cost dataset, consistent tagging and allocation, removal of waste, right-sizing, provider-specific commitment discounts and control of data transfer between clouds.
Is multi-cloud more expensive than a single cloud?
Often, yes, at least at first. You lose volume discounts, duplicate tooling and pay for data transfer between providers. Multi-cloud pays off when it gives you something specific, such as a service only one provider offers, regulatory separation or bargaining power, and you manage it with one cost view.
What is FOCUS in FinOps?
FOCUS is the FinOps Open Cost and Usage Specification, an open schema for billing data maintained by the FinOps Foundation. AWS, Microsoft, Google Cloud and Oracle publish FOCUS-format exports, so costs from different providers can be analysed in one table.
Which tools help with multi-cloud cost management?
Native tools such as AWS Cost Explorer, Azure Cost Management and Google Cloud Billing reports work per provider. For one view across clouds, load FOCUS exports into a data warehouse or use a third-party FinOps platform that supports all your providers.
Does the EU Data Act remove egress fees?
Not entirely. Since 12 September 2025, egress fees for simultaneous multi-cloud use may not exceed the provider's actual transfer costs, and from 12 January 2027 switching charges are banned. Normal data transfer during operation can still be billed.
Related reading
Related Services
Related Articles
About the Author
Country Manager, Sweden
Johan leads Opsio's Sweden operations, driving AI adoption, DevOps transformation, security strategy, and cloud solutioning for Nordic enterprises. With 12+ years in enterprise cloud infrastructure, he has delivered 200+ projects across AWS, Azure, and GCP — specialising in Well-Architected reviews, landing zone design, and multi-cloud strategy.
Editorial standards: This article was written by a certified practitioner, named in the byline above. We revise it when the technology or regulation it covers changes; the dates above show the latest update. Opsio maintains editorial independence — we recommend solutions on technical merit, not commercial relationships.
