Built to Grow: How One SaaS Company Scaled from a Closet Server Room to a Full Data Center Without Starting Over

For most early-stage software companies, the server room is an afterthought. You buy what you need, stack it in whatever space is available, and promise yourself you'll sort it out properly when the funding comes in. That was exactly the posture Marcus Ellery adopted when he co-founded CloudBridge Analytics out of a shared office in Austin, Texas, in 2018.

By 2020, CloudBridge had 50 employees, a product gaining traction in the mid-market HR software space, and an infrastructure problem that could no longer be deferred.

"We were running workloads on a mix of leased colocation hardware and some older rack units we'd bought secondhand," Ellery recalled during a recent conversation with Dell Experience. "Every time we added a customer segment or launched a new module, we had performance conversations that nobody wanted to have."

The company faced a fork in the road that many founders in similar positions recognize: go all-in on public cloud, invest in purpose-built on-premises infrastructure, or find some middle ground. What Ellery and his team ultimately chose was a modular, incremental strategy built around Dell PowerEdge servers—and the lessons from that decision are instructive for any growing tech company wrestling with the same questions.

Why Ripping It All Out Wasn't the Answer

The instinct to start fresh is understandable. When your existing infrastructure is a patchwork of mismatched hardware and borrowed configurations, a clean slate feels appealing. But Ellery's head of infrastructure, Diana Tran, pushed back on that impulse early.

"A full replacement isn't just a capital expenditure problem—it's an operational disruption problem," Tran explained. "You're migrating workloads, retraining your team, renegotiating vendor contracts, and doing all of that while the business keeps moving. We didn't have the runway to absorb that kind of disruption."

Instead, Tran began evaluating server platforms that could be expanded incrementally, with consistent management tooling across generations of hardware. After a competitive review that included HPE and Lenovo offerings, CloudBridge selected Dell PowerEdge rack servers as the foundation of their rebuilt environment.

The decision came down to several factors: the consistency of Dell's OpenManage management interface across hardware generations, the availability of modular storage and networking expansion options, and Dell's ProSupport enterprise service agreements, which gave a lean IT team confidence that help was available around the clock.

Scaling in Phases, Not Leaps

CloudBridge's infrastructure expansion happened in three distinct phases over roughly four years, each timed to company milestones rather than arbitrary hardware refresh cycles.

The first phase, in late 2020, involved deploying a cluster of PowerEdge R740 servers to replace the colocation hardware and establish a consistent baseline. The team standardized on a single management platform and, critically, selected a chassis and networking architecture that would accommodate future expansion without requiring new cabling or rack restructuring.

"We were deliberate about buying slightly more chassis capacity than we needed on day one," Tran said. "That upfront cost felt uncomfortable at the time, but it meant we didn't have to rebuild the physical layer every time we added compute."

By 2022, CloudBridge had grown to approximately 200 employees and was processing significantly higher data volumes as its analytics product matured. A second infrastructure phase added PowerEdge R750 nodes into the existing racks, with Dell's iDRAC remote management tools allowing the team to bring new hardware online with minimal physical intervention.

The third phase, completed in early 2024 as the company crossed 500 employees, introduced PowerEdge servers with NVMe storage configurations to support the company's newer AI-assisted reporting features. Crucially, this expansion integrated into the same management framework the team had been using since 2020.

"We didn't have to retrain anyone," Ellery noted. "The tooling was familiar. The support relationship was established. That continuity has real dollar value that doesn't always show up in the initial cost comparison."

What the Numbers Actually Showed

CloudBridge's finance team conducted an informal cost comparison in 2023, modeling what a full public cloud migration in 2020 would have cost against their actual infrastructure spend over the same period.

The results were nuanced. For pure compute costs at their current scale, on-premises infrastructure carried a clear advantage—particularly given the predictability of their workloads. The company's analytics platform processes batch jobs on defined schedules, which makes reserved or on-demand cloud pricing less competitive than it might be for more variable workloads.

However, Ellery was candid that the comparison isn't universally favorable. "If we were a different kind of company—more variable traffic, faster geographic expansion—the math might look different. We're not here to tell every startup to go on-prem. We're saying this approach worked for our specific profile."

The more compelling figure was total infrastructure management hours. By maintaining a consistent hardware and software stack, Tran's team spent significantly less time on platform-specific troubleshooting and vendor coordination than peer companies of similar size that had pursued more fragmented infrastructure strategies.

Practical Takeaways for Growing Tech Companies

For founders and infrastructure leaders facing similar inflection points, CloudBridge's experience surfaces a handful of actionable principles.

Standardize early, even when it feels premature. Choosing a consistent hardware vendor and management platform before you feel the pressure of scale gives you options later. Heterogeneous environments are harder and more expensive to manage at every stage.

Size your physical infrastructure for the next phase, not the current one. Buying slightly ahead of demand on chassis and networking capacity avoids costly physical restructuring during periods of high operational pressure.

Evaluate total cost of ownership across a realistic time horizon. Per-unit compute comparisons between cloud and on-premises rarely tell the full story. Factor in management overhead, support costs, and the operational disruption cost of platform migrations.

Treat vendor relationships as infrastructure assets. CloudBridge's established ProSupport relationship meant faster resolution times and access to Dell engineering resources during critical scaling moments. That relationship took time to build and had real value when it mattered.

A Foundation, Not a Destination

CloudBridge Analytics is not finished growing, and Ellery is candid that the infrastructure strategy will continue to evolve. The company is currently evaluating a hybrid approach that would add cloud bursting capacity for peak processing periods while maintaining its on-premises core.

"We're not ideologically committed to any single model," he said. "We're committed to making decisions that give us control, predictability, and the ability to adapt. The Dell infrastructure has done that for us."

For other SaaS companies watching their headcount and their server rooms grow simultaneously, that posture—pragmatic, incremental, and grounded in operational reality—may be the most valuable thing CloudBridge has to offer.