The cloud has become the default answer to almost every infrastructure question. Need more computing power? Cloud. Need better reliability? Cloud. Need to reduce capital expenses? Cloud.
But the cloud is not a single thing, and migration is not always the right answer. Understanding when cloud infrastructure makes sense—and when it does not—is essential for making good technology decisions.
What "The Cloud" Actually Means
When people say "cloud," they might mean several different things:
Infrastructure as a Service (IaaS) — raw computing resources: virtual servers, storage, networking. AWS EC2, Azure VMs, Google Compute Engine. You manage the operating system and everything above it.
Platform as a Service (PaaS) — managed platforms for running applications: databases, application runtimes, container orchestration. AWS RDS, Azure App Service, Google Cloud Run. The provider manages the underlying infrastructure.
Software as a Service (SaaS) — complete applications delivered over the internet: Salesforce, Microsoft 365, Slack. You just use the software; someone else runs everything.
Each category involves different tradeoffs. Most businesses use a mix.
The Case for Cloud
Cloud infrastructure offers genuine advantages:
Elasticity. Scale up during busy periods, scale down during quiet ones. Pay for what you use rather than provisioning for peak capacity that sits idle most of the time.
Reduced capital expense. No server hardware to purchase and depreciate. Operational expense instead of capital expense, which affects cash flow and taxes differently.
Global reach. Deploy in data centers around the world without building or managing facilities. Reach customers with lower latency wherever they are.
Managed services. Databases, caching, message queues, machine learning—managed services handle operational complexity that would require specialized expertise to run yourself.
Built-in redundancy. Major cloud providers engineer for high availability in ways that would be prohibitively expensive to replicate in a single-office setup.
The Case Against Cloud (Sometimes)
Cloud is not automatically the right choice:
Cost at scale. For stable, predictable workloads, cloud computing often costs more than owned infrastructure over time. The elasticity premium makes sense when you need elasticity. If your workload is constant, you are paying for flexibility you do not use.
Data gravity. Moving data into cloud is cheap. Moving data out is expensive—both in direct egress fees and in the bandwidth limitations. Once your data is in a cloud platform, you have reduced practical options.
Vendor lock-in. Cloud-specific services (AWS Lambda, Azure Functions, Google BigQuery) offer powerful capabilities but create dependencies. Migration becomes increasingly expensive as you adopt more provider-specific features.
Compliance complexity. Some regulatory frameworks require data to stay in specific jurisdictions or on specific infrastructure. Cloud compliance is possible but requires careful architecture.
Loss of control. When the cloud provider has an outage, you have an outage. You cannot drive to the data center and restart a server. Your fate is tied to their operations.
The Cloud Exit Phenomenon
Something interesting has been happening: companies that were cloud-first are bringing workloads back on-premises.
Basecamp/37signals publicly documented their cloud exit, projecting $7 million in savings over five years by running their own servers. Dropbox moved most of its infrastructure off AWS. Many others have done the same quietly.
The pattern typically involves:
- Large, stable workloads with predictable resource needs
- Strong internal engineering capability
- Sufficient scale to justify dedicated infrastructure teams
- Workloads where cloud elasticity provides little value
This does not mean cloud is wrong—it means cloud is a tool that fits some situations better than others.
Hybrid Approaches
Most businesses end up with hybrid infrastructure:
- SaaS for email, collaboration, and standard business functions
- Cloud IaaS/PaaS for variable or scaling workloads
- On-premises or colocation for stable, high-volume workloads
- Edge computing for latency-sensitive applications
The goal is matching infrastructure to workload characteristics, not ideological commitment to any particular approach.
Making the Decision
When evaluating cloud migration, consider:
Total cost of ownership. Not just cloud fees—include the cost of migration, the cost of operating in cloud, the cost of potential exit, and the cost of lock-in. Compare to realistic on-premises costs including hardware refresh cycles, facilities, power, and staffing.
Workload characteristics. Variable workloads benefit more from cloud elasticity. Stable workloads may not justify the premium.
Team capabilities. Running your own infrastructure requires operational expertise. If you lack that expertise, managed cloud services may be worth the premium.
Strategic dependencies. What happens if your cloud provider raises prices significantly? Changes terms? Discontinues services you depend on? How concentrated do you want your vendor risk?
Regulatory requirements. Do compliance frameworks constrain your options? Is achieving compliance easier or harder in cloud versus on-premises?
Government and Specialized Clouds
Not all clouds are equal in terms of security and compliance:
GovCloud — specialized cloud regions that meet government security and compliance requirements. Higher cost, more restrictions, but may be required for certain contracts.
Industry-specific clouds — healthcare, financial services, and other regulated industries have cloud offerings with built-in compliance features.
Private clouds — cloud infrastructure in dedicated facilities, not shared with other customers. Higher cost, but addresses some multi-tenancy concerns.
The right choice depends on your regulatory environment and risk tolerance.
The Path Forward
If you are considering cloud migration:
- Inventory your workloads. Understand what you run, how it behaves, and what it requires.
- Calculate real costs. Model cloud costs for your actual usage patterns. Include egress, storage, and support—not just compute.
- Assess dependencies. What cloud-specific features would you adopt? What would migration cost if you needed to move later?
- Start small. Migrate less critical workloads first. Learn cloud operations before betting the business on them.
- Maintain optionality. Where possible, use portable technologies that do not lock you to a specific provider.
- Review regularly. The right answer today may not be the right answer in three years. Costs change, requirements change, capabilities change.
Cloud infrastructure is a powerful tool. Like any tool, its value depends on using it appropriately. Neither "all cloud" nor "no cloud" is likely the optimal answer. The businesses that thrive will be those that choose deliberately based on actual needs rather than following trends.