When does building Cloud Infrastructure (IaaS) make sense?

Building — running OpenStack or owned hardware — makes sense for organizations with large, predictable workloads, specialized infrastructure engineering capacity, and data residency requirements. Dropbox and 37signals are the most cited examples of companies that saved significantly by repatriating specific workloads.

When does buying Cloud Infrastructure (IaaS) make sense?

Buying makes sense for virtually all product teams. Public cloud removes the capital commitment and operational burden of hardware ownership, and the flexibility to scale up or down without procurement is worth the premium for variable or unpredictable workloads.

What are the main Cloud Infrastructure (IaaS) vendors?

Representative vendors include Microsoft Azure, Google Cloud Platform, AWS, Hetzner / DigitalOcean. B4 Pro scores the full set.

Are neoclouds worth considering for AI workloads?

Specialized cloud providers like CoreWeave and Lambda Labs offer GPU compute at three to six times lower cost than hyperscaler on-demand pricing for the right AI workloads — primarily inference and training that doesn't need the full hyperscaler service catalog. For high-volume, predictable AI compute, neoclouds are worth a direct cost comparison.

IT Operations · Engineering, IT & AI

Should you build or buy Cloud Infrastructure (IaaS)?

Cloud Infrastructure (IaaS) provides on-demand compute, storage, and networking resources over the internet, letting organizations run applications and workloads without owning physical servers. AWS, Azure, and Google Cloud Platform are the dominant providers; smaller clouds like Hetzner and DigitalOcean serve specific cost and geographic needs.

The build-vs-buy decision for Cloud Infrastructure turns on whether your workload profile is predictable and large-scale enough to justify the capital investment and engineering specialization required to operate your own hardware — as Dropbox and 37signals have done — or whether the operational simplicity and flexibility of hyperscaler pricing covers your economics; the decision has been stable for most teams but shifts at specific workload types.

Domain: IT Operations
Function: Engineering, IT & AI
Industries: Cross-industry

Last assessed June 2026 · re-scored quarterly via The Continuum.

Build it, buy it, or bridge?

	Build it	Buy it	Bridge (buy, then extend)
Cost shape	Significant upfront capex for hardware; low per-unit compute cost for steady-state workloads	Pay-as-you-go; budget providers (Hetzner, DigitalOcean) competitive for steady workloads	Colocation or private cloud for predictable workloads; public cloud for burst and variable demand
Time to value	Months to years to stand up owned infrastructure at meaningful scale	Compute available in minutes; no procurement, shipping, or racking required	Public cloud baseline immediate; private capacity added as workloads mature
Differentiation captured	Predictable cost at scale; specific hardware configurations (GPU density, storage ratios)	No differentiation — pure utility; competitive advantage happens above the infrastructure layer	Cost optimization on known workloads; flexibility retained for growth and experimentation
AI feasibility today	OpenStack and CloudStack are production-proven; requires specialized infrastructure engineering	Hyperscalers provide managed GPU instances; neocloud providers (CoreWeave, Lambda) offer 3-6x lower GPU cost	Public cloud for training experimentation; owned or neocloud for high-volume inference
Who it fits	Hyperscalers, telcos, sovereign cloud operators, and large enterprises with data residency requirements	The vast majority of SaaS and product companies; anyone without dedicated infra engineering	Companies with large steady-state compute alongside variable development and staging needs

The B4 call

B4 has a verdict for Cloud Infrastructure (IaaS).

Build, Buy, Bridge, or Beware, with the five-dimension scorecard and the reasoning behind it. Unlock the call, and every other category, with B4 Pro.

Unlock the verdict in B4 Pro →

When building Cloud Infrastructure (IaaS) makes sense

Running your own IaaS infrastructure makes economic sense in a narrow set of situations. Dropbox saved roughly $75 million over two years by moving predictable storage workloads off AWS. 37signals trimmed about $1 million annually by returning to owned hardware for steady-state compute. Those outcomes are real, but they represent specific workload profiles — large, predictable, storage-heavy or compute-heavy — not general conclusions. OpenStack and Apache CloudStack are mature, production-proven platforms used in telcos, sovereign cloud operations, and large enterprise private clouds. If your organization has specialized infrastructure engineering capacity, data residency requirements that limit hyperscaler options, or GPU-intensive workloads where neocloud providers offer three to six times the cost efficiency of on-demand hyperscaler pricing, the case for owning or colocating hardware is worth a serious calculation.

When buying Cloud Infrastructure (IaaS) makes sense

Buying public cloud IaaS is the right call for the overwhelming majority of product teams. Pay-as-you-go pricing removes the capital commitment and operational burden of owning hardware, and budget providers like Hetzner and DigitalOcean make the economics accessible for workloads that don't need hyperscaler scale or feature breadth. The competitive advantage for most companies comes from the applications and data running on the infrastructure, not from the infrastructure itself. The refactoring cost to migrate specific services off cloud, combined with the ongoing engineering cost of running hardware, typically erases the apparent savings for teams without dedicated infrastructure specialists. About 85% of workloads stay on public cloud rather than repatriating, which reflects where the economics actually land for the median organization.

AWS, Azure, and Google Cloud Platform are the platform. The build case here means running OpenStack or Apache CloudStack, which is a real and documented choice for telcos, sovereign cloud operators, and large enterprises with specific data-residency or cost profiles. Dropbox saved roughly $75 million moving off AWS for predictable storage workloads. 37signals trimmed about $1 million annually by returning to owned hardware for steady-state compute. Those numbers are real, but they represent specific workload profiles, not general conclusions.

For the vast majority of product teams, the buy case is obvious and the economics are sound. Pay-as-you-go pricing and budget providers like Hetzner and DigitalOcean keep the floor accessible. The calculus shifts when your workload is predictable enough to benefit from owned hardware, your engineering team has the infrastructure specialization to operate it, and the refactoring cost of moving specific services is justified by the long-term savings. GPU compute is the current exception, where specialized cloud providers offer three to six times the cost efficiency of hyperscaler on-demand pricing for the right workloads.

Representative vendors

AWSMicrosoft Azure and 3 more, scored in B4 Pro

B4 Pro

Get B4's actual call on Cloud Infrastructure (IaaS)

→ B4's call for Cloud Infrastructure (IaaS): Build, Buy, Bridge, or Beware
→ The five-dimension scorecard and the scoring rationale
→ All 5 vendors with pricing and positioning
→ Quarterly re-scores that feed the MCP live, so your agents always query the current call
→ MCP server plus API and SDK access, and CSV/JSON export

Upgrade to B4 Pro

Prefer to read first? The book covers the framework end to end.

Frequently asked

What is Cloud Infrastructure (IaaS)?: Cloud Infrastructure (IaaS) provides on-demand compute, storage, and networking resources over the internet, letting organizations run applications and workloads without owning physical servers. AWS, Azure, and Google Cloud Platform are the dominant providers; smaller clouds like Hetzner and DigitalOcean serve specific cost and geographic needs.
When does building Cloud Infrastructure (IaaS) make sense?: Building — running OpenStack or owned hardware — makes sense for organizations with large, predictable workloads, specialized infrastructure engineering capacity, and data residency requirements. Dropbox and 37signals are the most cited examples of companies that saved significantly by repatriating specific workloads.
When does buying Cloud Infrastructure (IaaS) make sense?: Buying makes sense for virtually all product teams. Public cloud removes the capital commitment and operational burden of hardware ownership, and the flexibility to scale up or down without procurement is worth the premium for variable or unpredictable workloads.
What are the main Cloud Infrastructure (IaaS) vendors?: Representative vendors include Microsoft Azure, Google Cloud Platform, AWS, Hetzner / DigitalOcean. B4 Pro scores the full set.
Are neoclouds worth considering for AI workloads?: Specialized cloud providers like CoreWeave and Lambda Labs offer GPU compute at three to six times lower cost than hyperscaler on-demand pricing for the right AI workloads — primarily inference and training that doesn't need the full hyperscaler service catalog. For high-volume, predictable AI compute, neoclouds are worth a direct cost comparison.

The B4 Index scores every software category on two axes, strategic differentiation and AI feasibility, to classify it Build, Buy, Bridge, or Beware. See the full methodology.

More in IT Operations

Build or buy IT Service Management (ITSM)? Build or buy IT Asset Management (ITAM)? Build or buy Software Asset Management (SAM)? Build or buy Cloud Cost Management / FinOps? Build or buy SaaS Management? Build or buy Configuration Management Database (CMDB)? Build or buy Network Monitoring? Build or buy Unified Endpoint Management (UEM)? Build or buy SD-WAN? Build or buy Data Center Infrastructure Management (DCIM)? Build or buy Bare Metal Cloud / Dedicated Server Provisioning? Build or buy Certificate Lifecycle Management / PKI Automation?

The Build Report

Bi-weekly analysis of software categories through the B4 Framework. What to build, what to buy, and how to use AI to make better decisions for your company.