Linode under Akamai: what changed, what did not, and who it suits

Linode spent two decades as the sensible choice for developers who wanted a real server without a console full of services they would never use. Akamai bought it in 2022, and the fear at the time was that the pricing and the simplicity would go. Neither did.

What arrived instead was scale: a substantially larger set of locations, backed by one of the largest content delivery networks in existence.

The plan families

• Shared CPU. The entry line, starting with the very small Nanode. Cores are shared, which is fine for anything without a sustained CPU floor.
• Dedicated CPU. Guaranteed cores, no contention. The correct choice for build servers, game servers, databases and anything where latency consistency matters.
• High Memory. More memory per core for caches and in memory workloads.
• Premium. Newer processor generations with a performance floor, at a premium over the standard dedicated plans.
• GPU and accelerated. NVIDIA hardware for training, inference and rendering, billed by the hour.

Pricing is flat monthly with an hourly cap, so an instance destroyed after three days costs three days. That is simpler to budget against than the metered models on the hyperscalers, and it is why so many small teams have stayed.

The transfer pool

Each instance contributes a monthly outbound transfer allowance to an account pool. You are charged only for what the account collectively exceeds. Inbound is free.

The practical effect is the same as on DigitalOcean: a fleet of quiet instances builds headroom for a busy one, and internal traffic on private addresses does not consume the pool at all. Keep service to service traffic on the private network and the pool usually lasts.

What the Akamai footprint changed

The location list expanded well beyond the original set, which matters for two reasons. Latency sensitive workloads can sit closer to users, and regulated workloads can sit in the right jurisdiction without leaving the platform.

There is also tighter integration with Akamai’s delivery and security products for teams that want them. That is optional; you can still run a plain Linode instance and never touch any of it.

The account level restriction

Outbound SMTP is blocked on new accounts. Ports 25, 465 and 587 are restricted until you request access and the account has some history. This is a deliberate anti abuse measure and every reputable provider does something equivalent.

If your application sends mail directly, this is the setting that decides whether it works. It is an account property, which is why listings distinguish between standard accounts and accounts with port 25 already open. The wider picture is in outbound port 25 on cloud servers.

The rest of the platform

• Object Storage. S3 compatible, priced as a base subscription with included capacity and transfer.
• Block Storage. Volumes attached to instances, billed per gigabyte month.
• NodeBalancers. Managed load balancing at a flat monthly rate.
• Managed Databases. PostgreSQL and MySQL with backups and failover.
• LKE. Managed Kubernetes where you pay for the worker instances rather than the control plane.
• VPC and Cloud Firewall. Private networking and stateful filtering, both included rather than metered.

Where Linode fits

It is the right answer for teams who want predictable monthly costs, a control panel that does not require training, and a real Linux machine rather than a platform abstraction. The documentation is genuinely good and the support has a reputation it has kept.

It is the wrong answer if you need a deep managed service catalogue, unusual compliance certifications, or the kind of enterprise procurement relationship the hyperscalers exist to provide.

The closest comparison is DigitalOcean, which competes for the same customer with a very similar shape. That one is worth reading in full: DigitalOcean vs Linode. If price is the deciding factor, Hetzner undercuts both in Europe.