CapEx
Definition
Capital Expenditure - upfront costs for buying physical equipment like servers and data centers that you own. Like purchasing a house instead of renting.
Use Cases
- Netflix: Shift from owning data center infrastructure to using cloud infrastructure to scale streaming globally. — Migrated away from company-managed data centers and rebuilt core systems on AWS using a pay-as-you-go model, reducing the need to purchase and refresh large amounts of owned server and storage hardware. (Improved ability to scale capacity on demand and reduced reliance on large upfront infrastructure purchases, aligning costs more closely with usage.)
- Capital One: Modernize IT and increase agility by moving from owned infrastructure to cloud services. — Adopted AWS broadly and reduced dependence on traditional data center hardware purchases by using managed services and elastic compute/storage instead of buying servers for peak capacity. (Faster delivery of digital capabilities and less need for large, periodic hardware refresh cycles typically funded as CapEx.)
- Dropbox: Control long-term infrastructure economics by investing in owned infrastructure for predictable workloads. — Built and expanded its own infrastructure (including custom storage systems) for parts of its workload, which required upfront investment in servers, storage, networking, and facilities—classic CapEx—while still using cloud where it made sense. (For steady-state workloads, owning infrastructure can improve unit economics over time, but it requires significant upfront spending and ongoing operations.)
Frequently Asked Questions
- What's the difference between CapEx and OpEx in cloud computing?
- CapEx is money spent upfront to buy and own long-lived assets like servers, storage, and data center equipment. OpEx is ongoing spending to operate the business, like monthly cloud bills for compute, storage, and managed services. Cloud adoption often shifts costs from CapEx (buying hardware) to OpEx (paying for usage).
- When should a business choose CapEx instead of cloud (OpEx)?
- CapEx can make sense when workloads are stable and predictable, you have strong in-house operations expertise, regulatory or data residency needs require owned facilities, or you want full control over hardware choices. Cloud (OpEx) is often better when demand is variable, you need fast time-to-market, or you want to avoid large upfront purchases and hardware refresh cycles.
- How much does CapEx cost for on-prem servers and a data center?
- CapEx varies widely based on capacity, redundancy, performance, and facility requirements. Major cost drivers include servers/GPUs, storage arrays, networking gear, racks and cabling, power and cooling, physical security, backup systems, and multi-site disaster recovery. You also need to plan for refresh cycles (often every 3–5 years) and additional costs that are not CapEx (like staff, maintenance contracts, and electricity), which are typically OpEx.
Category: business
Difficulty: intermediate
Related Terms
See Also