Summary

  • Solomon e3 says its total monthly cloud spending fell by more than half after a review and a move of its development and MVP environment to NexQloud.
  • The result combines retired idle resources, resized capacity and cheaper compute. Selected AWS services remain, and an independent energy study is still a plan.

Three changes behind one bill

The striking number in Solomon e3’s 4 September announcement is a reduction of more than half in total monthly cloud spending. The more revealing detail is the work that preceded it. Engineers compared the customer’s bill, codebase and infrastructure definitions, then removed idle resources, resized capacity and moved compute. Several purchasing and engineering decisions sit inside the reported result.

Solomon e3 describes an energy-infrastructure and geospatial platform. Its development and minimum viable product environment now runs on NexQloud One, according to the announcement. The review applied the principles of NexQloud’s Qlarity offering by hand. It is not evidence that a software tool independently discovered and delivered the entire saving.

The companies also cite compute costing about a third less than AWS. That comparison uses identical configurations at published on-demand list prices in us-east-2, without Savings Plans. It answers a narrower question than the customer’s before-and-after spending figure.

Reported measure What it covers
More than half less monthly spending Solomon e3’s total cloud bill, including AWS services still used
About a third less compute cost A stated list-price comparison for matching configurations and specified purchasing terms
Lower resource use Retirement and resizing identified in the review, without a published contribution for each

These measures should not be added together or treated as interchangeable. A resource no longer required has a different economic story from a resource bought at a lower rate. The release does not provide the dollar baseline, matched measurement periods or contribution of each intervention needed to reproduce the aggregate saving.

A selective move, with data services left behind

The reported environment runs its portal, gateway and API as containers on managed Kubernetes, alongside self-hosted OpenSearch and PostGIS. The announcement says application code was not rewritten. DynamoDB, S3 and AWS Glue remain as bridge components while Postgres with PostGIS is built out as the production data layer.

That is a selective migration, not a completed exit of the whole production estate from AWS. Retaining useful data services can allow application work to move before every dependency is replaced. It also leaves responsibilities and costs across more than one environment. The absence of an application rewrite does not establish the absence of integration or operating effort.

The FinOps Foundation’s usage-optimization guidance treats sizing and unnecessary consumption separately from rate optimization, which concerns what an organisation pays for the resources it uses. The latter warns that the two sets of actions interact. In this case, a like-for-like price comparison would need to be distinguished from savings achieved by changing what was running.

The energy question remains open

The release points to a NanoServer using an Intel Core i9-14900T and cites 35 watts. Intel’s specifications identify that as processor base power and separately list 106 watts of maximum turbo power. Neither figure measures the energy consumed by the complete service for a useful unit of work.

Solomon e3 and NexQloud say they will commission an independent academic review of cost, compute and energy efficiency. That is a proposed study, not a published finding. The claim of no unplanned downtime also comes from NexQloud’s monitoring; this reporting did not inspect the underlying availability record.

The linked proof page was not accessible during research, and no original invoices, code or operating logs were examined. The case is therefore an attributed customer announcement with a plausible mechanism, not an independently reproduced benchmark. Its commercial significance is the combination of cleanup and selective portability. Whether another buyer can reproduce the result depends on its starting waste, purchasing terms, retained services and required operating standard.

Sources