A hyperscaler is one of the very largest cloud providers: companies that operate global networks of massive data centers and sell computing, storage, networking, databases, analytics, AI and hundreds of other services on demand to customers anywhere. The name comes from “hyperscale”, meaning infrastructure designed to grow quickly to very large size by adding standardized capacity. In everyday use, “the hyperscalers” usually means Amazon Web Services, Microsoft Azure and Google Cloud, though some definitions include a few other global providers.
At a glance
- The term describes the largest global public cloud providers; there is no official membership list.
- They offer the broadest catalog of services, from basic Infrastructure as a Service (IaaS) to managed databases and AI.
- They run many regions worldwide, each usually split into several availability zones.
- Pricing is usage-based, with discounts for committed spend; data transfer out and premium support add significant cost.
- Their scale brings breadth and reach, but also lock-in risk and limited negotiating leverage for smaller buyers.
What problem it solves
Most organizations cannot build global data center footprints, keep pace with new hardware, or develop hundreds of managed services. Hyperscalers do that once at enormous scale and rent it out. Even a small company can deploy in many countries, use managed databases, machine learning tools or content delivery without building them, and pay for what it uses.
The appeal for buyers is speed and breadth. A team can start a project the same day, scale it without a hardware purchase, and choose from a catalog that smaller providers rarely match. The trade-off is complexity: thousands of service options, intricate pricing and a strong pull to build more and more on one provider’s proprietary services.
How it works
Global footprint. Hyperscalers organize capacity into regions (geographic areas) that each contain several isolated data center groups called availability zones. Customers choose where their workloads and data run, which also matters for latency and data residency.
Service catalog. Services range from virtual machines and storage to managed databases, serverless functions, analytics, identity, security tools and AI models. Higher-level services save effort but are usually specific to one provider.
Buying models. Most customers start pay-as-you-go. Larger customers often sign committed-spend agreements, which trade a multi-year commitment for discounts, credits or support. Buying through marketplaces and resellers is also common. Many software vendors sell through the hyperscalers’ marketplaces, which can count toward a spending commitment, depending on the agreement.
Connectivity. Customers reach hyperscalers over the internet or through private connections from data centers and carrier networks, which can reduce data transfer costs and improve consistency.
For help choosing and negotiating, see our public cloud solution page.
When it matters for buyers
- When choosing a primary cloud. Compare service fit, regional coverage, existing licenses and skills, and support quality, not only price.
- When negotiating a commitment. Committed-spend agreements can lower rates but create a minimum you must pay; forecast conservatively and understand what counts toward it.
- When considering more than one provider. Multi-cloud strategies reduce dependency but add operational overhead.
- When costs climb. Data transfer, idle resources and premium support are common surprises; many companies review whether steady workloads belong in a hybrid design instead.
- When data location matters. Check which regions offer the services you need and whether they meet your residency or sovereignty requirements.
Questions to ask vendors
- Which regions and availability zones offer every service we plan to use?
- What committed-spend or reserved-capacity discounts are available, and what happens if we fall short of the commitment?
- How is data transfer billed, both out to the internet and between regions and zones?
- What does each support tier include, and what response times are committed?
- Which of the services we plan to use are proprietary, and how would we move off them?
- Do marketplace purchases count toward our commitment?
- If we buy through a reseller or partner, who provides support and who owns the billing relationship?
How it differs from colocation
A hyperscaler rents you finished cloud services; you do not manage its hardware. A colocation provider rents you space, power, cooling and connectivity in a data center, and you supply and run your own equipment. The two are linked: hyperscalers lease capacity from wholesale colocation operators in some markets, and many colocation facilities host private on-ramps to hyperscaler networks. Many buyers use both, with steady systems in colocation and variable or service-heavy workloads in the cloud. The broader category of providers is described under cloud computing and strategic cloud platform services.