TechCompare LogoTechCompare

What is a /16 subnet? 65,536 IPs and when a network needs that many addresses

A /16 is a planning container, not a subnet you'd actually assign to a single broadcast domain. Use it as the top-level allocation for a large site, then subnet it down to /22 or /24 for actual deployment. In cloud environments, a /16 VPC gives you room to carve out /20 subnets per availability zone.

A /16 subnet, written as 255.255.0.0, provides 65,536 total addresses with 65,534 usable for hosts. This is a large subnet, typically seen in enterprise environments, university campus networks, and cloud VPC configurations like AWS's default VPC sizing.

By TechCompare · Updated

CIDR
/16
Class B
Total addresses
65,536
65,534 usable for hosts
Network size
Enterprise
Larger deployments

How this is calculated

A /16 means the first 16 bits are the network portion, leaving 16 bits for hosts (2^16 = 65,536). In practice, you'd almost never put 65,000 devices on a single flat subnet. Broadcast traffic alone would cripple performance. Instead, /16 blocks are almost always sub-divided into smaller subnets (/20, /22, /24) for different departments, floors, or security zones. The /16 acts as a supernet or summary route that encompasses all the smaller subnets underneath it.

Verdict

The math is 16 bits reserved for the network and 16 bits for hosts, so 65,536 total addresses and 65,534 usable. The reason you don't put 65,000 hosts on a flat L2 segment is broadcast traffic: ARP, DHCP, mDNS, and NetBIOS chatter would saturate the segment long before you filled the address space. Real deployments use the /16 as a supernet or summary route and carve it into operational subnets underneath, often /20 per availability zone (4 zones in AWS = 4 /20s inside a /16) and /22 per application tier. That's why AWS ships its default VPC at /16, not because anyone expects a flat 65K-host LAN.

More Subnet scenarios

Frequently asked questions

Why would anyone need a /16 subnet?
It's typically used as an allocation block that gets sub-divided further. A university might get a /16 from their regional internet registry and carve it into /24s per building.
Why would a cloud VPC use a /16 instead of multiple /24s?
Because a /16 VPC gives you one IP namespace with room to carve out /20 or /22 subnets per availability zone, each with thousands of usable IPs for auto-scaling. Multiple /24 VPCs would force inter-VPC routing, peering complexity, and IP overlap risk. The /16 supernet is the AWS default VPC size.
How do I subdivide a /16 into useful smaller subnets?
Common practice is: /20 per availability zone (4 zones per region in AWS, so a /16 fits 4 /20s), /22 per application tier, and /28 per isolated service cluster. This is hierarchical subnetting: each level carves a fraction of the parent block. Cloud providers' CIDR calculators and ipcalc do this math for you.