ExxonMobil achieved one of the largest cloud cost reductions in AWS history by subjecting its sprawling infrastructure to an Amazon Optimization and Licensing Assessment (OLA). The energy giant, which shifted massive workloads to AWS five years ago, had accumulated a complex web of over-provisioned resources, legacy licensing, and storage inefficiencies that were bleeding millions annually. The OLA, conducted over six months in late 2025, uncovered savings of 34% on compute, 45% on database licensing, and 22% on storage—reductions that AWS says redefine what’s possible for cloud economics at scale.

At the core of the engagement was a forensic analysis of nearly 12,000 EC2 instances, 400 Oracle and SQL Server databases, and 18 petabytes of S3 storage across ExxonMobil’s production, analytics, and edge environments. The assessment revealed that 60% of instances were running at less than 15% CPU utilization, while 30% of provisioned IOPS on RDS databases had never exceeded 50% of capacity. More critically, the company was paying for 2,100 Oracle Database Enterprise Edition licenses that could be migrated to AWS-native services, and 850 SQL Server licenses tied to Windows Server hosts that were long retired.

“Enterprise workloads come with a heavy licensing debt that most CFOs don’t see until an OLA uncovers it,” said Phil Magnuson, GM of AWS Optimization, in a briefing on Sunday. “ExxonMobil’s engagement proves that the biggest cloud savings aren’t in compute—they’re in the way you align licensing, architecture, and demand patterns.”

What Exactly Is an AWS Optimization and Licensing Assessment?

The OLA is a no-cost, multi-week engagement where AWS architects, licensing specialists, and FinOps analysts pore over a customer’s entire cloud estate. It was launched in 2021 to address the growing pain of cloud sprawl and the parallel headache of intellectual property (IP) and license compliance as enterprises lift-and-shift on-premise workloads. Unlike generic cloud cost tools that recommend shutting off idle instances, OLA digs deep into the licensing implications of every workload—checking if a Microsoft SQL Server instance could run on a Linux host with license mobility, or if an Oracle RAC cluster could be split into smaller, license-free Aurora PostgreSQL instances.

The assessment produces a ranked list of actions tied to dollar values: right-sizing recommendations, license migration strategies, modernization pathways, and storage tiering plans. AWS shared anonymized data showing that across 2,400 OLAs conducted in 2025, the median savings was 28% of total AWS spend, with the top quartile exceeding 40%.

How ExxonMobil’s OLA Translates to Real Dollars

ExxonMobil’s AWS monthly spend before the OLA was approximately $14.7 million, according to two sources familiar with the engagement. The OLA identified $5.1 million in recurring monthly savings, broken down as:

  • Compute right-sizing and savings plans: $1.8 million/month by moving 3,400 instances from on-demand and reserved instances to Compute Savings Plans and AWS Graviton-based instances.
  • Database modernization and license elimination: $2.2 million/month by migrating 60% of Oracle databases to Amazon Aurora PostgreSQL, reclaiming unused Oracle licenses, and implementing SQL Server license mobility on Amazon Linux instead of Windows Server.
  • Storage optimization: $1.1 million/month through S3 Intelligent-Tiering, deleting 2PB of orphaned snapshots, and consolidating EBS volumes.

On an annualized basis, the savings surpass $60 million—roughly the cost of ExxonMobil’s entire 2022 renewable energy R&D budget, insiders noted. The most surprising finding, according to AWS, was the sheer scale of Windows Server licensing waste. Because ExxonMobil had originally migrated SQL Server workloads in lift-and-shift mode, it was paying for Windows Server licenses on 1,200 EC2 instances that were only running SQL services. By switching to the bring-your-own-license (BYOL) model and running SQL Server on Linux, the OLA freed $340,000 per month without any code changes.

The SQL Server Licensing Sinkhole

For Windows-focused enterprises, SQL Server licensing is often the largest single cost driver in the cloud—and the least understood. The pivotal point is License Mobility through Software Assurance. Many customers don’t realize that SQL Server licenses with active Software Assurance can be deployed on AWS on Linux hosts, completely avoiding Windows Server licensing costs. ExxonMobil had 850 SQL Server Standard and Enterprise core licenses that were assigned to Windows Server instances; the OLA re-hosted 620 of those workloads onto Amazon Linux 2 instances with zero application changes, saving $680 per core per year.

Even more impactful was the find that 40% of ExxonMobil’s SQL Server instances were using features available in SQL Server Web Edition, which costs up to 70% less. The assessment detailed a phased upgrade path: relicense on Web Edition where possible, then use AWS Launch Wizard to rebuild Windows-based failover clusters into Linux-based Always On availability groups. The project is expected to net an additional $9 million in annual savings once complete.

Oracle Modernization: From Lock-in to Freedom

ExxonMobil’s technical debt was most visible in its Oracle fleet—410 databases running on 540 EC2 instances, with many still running Oracle 11g (released 2009) and requiring extended support. The OLA used the AWS Schema Conversion Tool to analyze these databases and found that 62% could be migrated to Amazon Aurora PostgreSQL with minimal refactoring. The projected one-time migration cost of $2.3 million would be recouped in under five months through avoided Oracle license fees.

Crucially, the assessment flagged 120 Oracle databases that were part of closed-vendor applications where the ISV wouldn’t certify PostgreSQL. For those, AWS recommended Oracle on Amazon RDS with the License Included model, which eliminated the need for the customer to manage core factor and compliance. The net effect was a 45% reduction in monthly Oracle spend, proving that even partial modernization yields massive returns.

FinOps for the Megacorp

The ExxonMobil case validates the FinOps Foundation’s latest maturity model, which emphasizes that optimization must be an ongoing team sport, not a one-time audit. After the OLA, ExxonMobil created a dedicated Cloud Economics Office (CEO) with 18 full-time staff who use AWS Cost Explorer, Compute Optimizer, and Trusted Advisor to enforce tagging policies, rightsize weekly, and govern license allocation. The company also adopted Amazon DevOps Guru for automated detection of underutilized databases and EC2 instances.

“The OLA isn't just a report; it’s a cultural injection,” said J.R. Storment, Executive Director of the FinOps Foundation, in a statement. “ExxonMobil now has a living FinOps practice that treats cloud costs as a first-order design constraint, not an afterthought. That’s the difference between one-time savings and sustained advantage.”

Lessons for Every Windows IT Leader

While few organizations match ExxonMobil’s scale, the dynamics uncovered by the OLA apply to any company running Windows and SQL Server workloads on AWS:

  1. Audit your Windows Server licenses immediately. If SQL Server is the only Microsoft product on an EC2 instance, you’re likely overpaying. Migrate to Linux and use BYOL.
  2. Check your Oracle core factor. Oracle’s cloud licensing policy counts AWS vCPUs differently than on-premise sockets. An OLA can recompute your effective licensing position and often finds you’re overlicensed.
  3. S3 storage should never be static. Enable Intelligent-Tiering globally and set lifecycle policies for snapshots and versioning. ExxonMobil was shocked to find terabytes of objects that hadn’t been accessed since 2018.
  4. Software Assurance is your hidden asset. If you have it, you have license mobility for SQL Server—use it.
  5. Modernize before you optimize. Often, a migration to a cloud-native service (e.g., Amazon RDS, Aurora, DynamoDB) eliminates the licensing variable entirely, making further tweaks unnecessary.

The Road Ahead: AI and Sustainability

ExxonMobil plans to invest a portion of the savings into GPU clusters for seismic processing AI and generative models for exploration, breathing life into its 2030 digital transformation goals. The energy company also highlighted the sustainability win: reducing its AWS footprint by 60% in CPU hours directly lowered its data center power consumption by an estimated 12 GWh per year—a number that helps its public ESG commitments.

AWS is expected to announce an enhanced OLA for AI/ML workloads later this year, which will analyze GPU utilization, Spot Instance strategies, and SageMaker endpoint sizing. For ExxonMobil, that could mean another round of savings as it embarks on its AI-first strategy.

A New Cloud Savings Playbook

The ExxonMobil OLA demonstrates that the biggest cloud waste isn’t idle VMs—it’s the misalignment between software licensing and modern infrastructure. By treating the cloud as a dynamic, architecturally fluid environment rather than a static data center, the company achieved what AWS calls “the largest license-optimized savings event in our history.” For enterprise IT teams watching their cloud bills climb, the message is clear: schedule your OLA before the next finance review.