March 25, 2026 by Cogent Solutions Engineering

Enterprise computing is a pendulum that moves in regular cycles: from centralization to decentralization and back again. Mainframes gave way to distributed servers, which were consolidated with virtualization. Then came public cloud, which provide rapid deployment and the promise of infinite scalability.

Read on for more

But today, we see many companies arriving at the same conclusion: not all workloads require the flexibility that public clouds provide, when more cost-effective alternatives are available. The initial enthusiasm for cloud services needs to confront the reality of each specific deployment.

When companies have bursty, unpredictable or short-lived workloads, public cloud platforms are a powerful tool. Cloud platforms are also very flexible for service development and prototyping. However, as those workloads grow in maturity, predictability and scale, the economics of running them on public clouds needs to be validated.

There are multiple factors that might lead to public cloud platforms no longer being the most cost-effective solution: always-on, predictable workloads, data egress and movement costs, east-west traffic at scale, latency and performance guarantees, and persistent data at volume.

Always-on, predictable workloads
When an application continuously runs at the same utilization, the economic advantage of on-demand computes decreases, and companies may no longer want to pay a premium for the flexibility.

Data egress and movement costs
Outbound data transfer remains one of the least visible and fastest growing components of cloud spend. Video delivery, AI model distribution, analytics exports, backups, and replication can silently turn into recurring expenses that rival compute costs.

East-west traffic at scale
As architectures span regions, availability zones, and services, internal bandwidth charges add up. What looks inexpensive per gigabyte becomes enormous at scale — especially for data-intensive applications distributed across multiple arbitrary cloud regions.

Latency and performance guarantees
Trading platforms, real-time analytics, streaming services, and interactive applications often require deterministic performance. Shared cloud networks usually mean premium service tiers or specialized services to meet that objective.

Persistent data at volume
Large datasets that are accessed frequently — not archived — often cost far more over time than owning or colocating dedicated storage infrastructure.

Every user, from enterprises to startups, eventually need to optimize their computing and data storage expenditures. Applications that might not require high-scale flexibility should not have to pay premium rates.

When companies start to re-evaluate their maturing workloads, they might recognize the value of a hybrid compute structure – dedicated infrastructure for the more predictable applications, and cloud for newer applications and rapidly changing developments.

Beyond cost-effectiveness, regulatory and data sovereignty issues might also suggest to carefully segment applications and data between public cloud environments and dedicated infrastructure.

Colocation in Carrier Neutral Data Centers allows you to have dedicated servers that deliver consistent performance and give you full control, at a predictable cost. Connecting such infrastructure to your offices, end users, or to your cloud-based applications is straightforward.

All Cogent connectivity products (Internet, VPN, Optical Wavelengths) support optimal connectivity to cloud service providers. The same connectivity products also securely extend your network to your dedicated infrastructure, with over 1700 Carrier Neutral Data Centers around the world connected to Cogent’s backbone. For further integration, Cogent Colocation and Utility Computing are available in 187 Cogent data centers in North America and Europe, to help build the dedicated infrastructure you require.

As organizations evolve their compute architectures, Cogent supports every iteration: your data, your choices. Cogent can support your applications deployment and deliver optimal connectivity.

About the Author:
The Cogent Solutions Engineering team is a group of experienced, technically-versed and business-minded individuals who understand the challenges facing our global customer base, and how Cogent’s set of products and services can address these, from their daily involvement in customer conversations, review of individual customer requirements, solution design, and resulting product development initiatives.