This article is written by Taylor Greiner, Director of Innovation, US Signal

When we introduced OpenCloud, we knew our customers’ infrastructure needs would keep changing. We wanted a foundation that would let us respond.

Businesses were looking for more control over cloud costs, dependable performance, and greater flexibility in where they ran applications. Those needs haven’t gone away. Now, those same organizations are also evaluating AI, reconsidering compute architectures, and asking how to modernize without disrupting the systems they depend on.

That is where our decision to build US Signal OpenCloud on Apache CloudStack continues to matter. It gives us a foundation we can build on as those questions become more specific.

Last year, we shared how openness, collaboration, and customer feedback shaped OpenCloud’s first year. This year, we have more to show about where that approach is taking us.

Let Customer Needs Guide the Platform

For customers, the value of an open-source foundation comes down to what it allows them to do.

Can they provision resources when they need them? Can they understand what those resources will cost? Can they move and protect workloads without creating another dependency that becomes difficult to unwind?

These are practical questions. They shape how we develop OpenCloud and where we invest our engineering effort.

Predictable pricing, self-service provisioning, and direct access to technical support remain central to that experience. So does helping customers make a transition that works for their business. Choosing a new platform is only part of the job. Planning the migration, understanding application dependencies, and keeping users working through the change matter just as much.

Our Swanson, Martin & Bell case study offers a closer look at one organization’s move from VMware to OpenCloud and the considerations behind that decision.

ARM Compute Makes Workload Choice More Useful

One of this year’s additions is ARM Compute, powered by AmpereOne processors.

It reflects a question we increasingly discuss with customers: what does this workload actually need?

AI makes that question particularly useful. A team exploring an AI application may start by assuming it needs GPUs. Depending on the model, response requirements, and expected demand, that may be the right answer. Other workloads may run effectively on CPUs, including ARM-based infrastructure.

Adding ARM Compute to OpenCloud gives customers another option to evaluate alongside x86.

Compatible Linux applications, web services, containers, and certain AI inference workloads can benefit from ARM’s approach to throughput and efficiency. The fit still depends on the application, its software dependencies, and the performance the business needs.

That evaluation matters more than choosing an architecture because it is getting attention. We want customers to have useful choices and the support to assess them.

It also illustrates why we chose an open foundation. As compute requirements evolve, we can extend the platform around those requirements.

A Cloud Platform Includes How Workloads Are Protected

Compute is only one part of the infrastructure conversation.

Customers also need to know how they will move applications, maintain recovery options, and protect systems across different environments. Those concerns become especially important when an organization is changing platforms.

OpenCloud’s development has included services such as Desktop as a Service and SyncSafe Replication, which uses agent-based replication at the operating system layer to help move and protect workloads.

Partnership is an important part of this work. Building on CloudStack gives us access to an ecosystem of people tackling related challenges. Working together helps us bring capabilities into the platform that address how customers actually operate.

Bringing Those Questions to the CloudStack Community

The CloudStack Collaboration Conference gives us an opportunity to share what we are learning and hear how other operators and contributors are approaching similar decisions.

In my keynote, “The Next Era of Cloud Infrastructure: Bridging Enterprise IT to AI-Native Applications,” I’ll explore how infrastructure can support emerging application needs while continuing to serve the enterprise systems businesses rely on today.

Alex Dietrich will take a closer look at evolving compute requirements in their workshop, “Beyond Virtual Machines: GPU Infrastructure and the Future of CloudStack.”

I’ll also join Max Bozhenko for “Extending CloudStack Through Partnership: Modern Disaster Recovery for CloudStack,” connecting the partnership discussion to a critical operational requirement: recovery.

During the Hackathon, Alex and I will present “CloudStack’s Next Five Years: A Community Roadmap Discussion.” It is an opportunity for the community to discuss where the project should go and how the community can help shape that direction.

These sessions address different parts of the same challenge: building infrastructure that remains useful as the demands placed on it change.

Keep Building Around the Workload

Our experience with OpenCloud has reinforced the value of staying close to customer requirements and working openly with others to address them.

ARM Compute is one example of where that approach has taken us this year. Workload mobility, recovery, and desktop delivery are others. Each adds to what customers can do with the platform.

As we look ahead, we’ll continue evaluating new capabilities against practical questions: does this solve a customer problem, where does it fit, and can we deliver and support it well?

Those are the questions we’re bringing to the conference. We look forward to working through them with the CloudStack community.