Akamai

What Akamai actually is

Akamai is the company that invented the content delivery network, and that history matters when you evaluate it. Tom Leighton and Danny Lewin worked out the routing mathematics at MIT, incorporated the company in August 1998, and launched commercial service in 1999 with Yahoo as a charter customer. Nearly three decades later Akamai is a public company (Nasdaq: AKAM), headquartered in Cambridge, Massachusetts, and it operates what is still the most widely distributed delivery network in the industry: 4,400+ edge points of presence across 130+ countries, riding on 1,200+ networks, with edge capacity the company puts at over 1 PBps. Third-party counts of individual servers hover around the 350,000 mark.

The product surface is broader than the old "CDN" label suggests. Akamai sells three overlapping things: content delivery, edge security (Web Application Firewall, bot management, API security, and the Prolexic DDoS scrubbing platform), and — since the 2022 Linode acquisition — Akamai Cloud compute. The compute line is now organized as a continuum: 25+ core compute regions carrying the heavier storage and database workloads, 10 distributed sites for latency-sensitive work, and the edge PoPs themselves running lightweight logic through EdgeWorkers and Akamai Functions. On ServerSpotter we file Akamai under CDN providers, edge computing, and firewall & DDoS, and that combination is the honest description: this is a delivery-and-security vendor that now also rents compute, not a general cloud provider with a CDN add-on.

In our view the distinction that matters for buyers is that Akamai sells outcomes to enterprises — global media delivery, uptime during attacks, predictable latency — rather than self-serve infrastructure. If your mental model is "spin up a VM with a credit card," you are shopping in the wrong aisle, even though the Linode heritage means Akamai Cloud compute can be bought that way.

Pricing: what you really pay

We list no entry price for Akamai, because there is no public list price to verify. Delivery and security contracts are quote-based, negotiated per customer against traffic commitments, regions, and which security modules you bundle. Procurement runs through sales, contracts are typically annual or longer, and the number on your invoice is a function of volume commitments rather than any published rate card. Anyone quoting you a specific "Akamai price" without knowing your traffic profile is guessing.

The practical consequence is a real budget floor. This is enterprise procurement: expect a sales cycle, a solutions engineer, and a minimum commit that puts it out of reach for personal projects and most small businesses. There is no free tier for the delivery or security stack; evaluation happens through sales-run trials and proofs of concept. If you want metered, self-serve delivery with transparent rates, the honest alternatives are the self-serve CDN vendors — Cloudflare's free and pro tiers, Fastly, or the budget CDNs — and we would point individuals and small teams there without hesitation.

Akamai Cloud compute is the exception. Those plans follow the Linode tradition of published per-hour rates for shared and dedicated instances, object storage, and managed databases, and they can be bought online without a contract. But treat that as a separate buying motion from the edge and security products this profile covers. Our guidance: if you only need compute, evaluate Akamai Cloud on its own merits and on our benchmarks page alongside Hetzner, DigitalOcean, and the rest of the self-serve crowd. If you need global delivery plus DDoS scrubbing, budget for a negotiation and compare the resulting quote against Cloudflare's and Fastly's enterprise tiers.

Performance & infrastructure

The infrastructure numbers are the reason enterprises pay up. 4,400+ edge PoPs in 130+ countries means an Akamai edge server is close to almost any user on earth, and sitting on 1,200+ networks means the mapping system can route around congestion at the last-mile level, which is where delivery actually breaks. Edge capacity above 1 PBps is the kind of headroom that absorbs traffic spikes — product launches, live sports streams, viral events — without drama. Akamai has delivered some of the largest live-streaming events on the public internet, and that track record is not marketing copy; it is the core of what the company has sold for 25 years.

The security side is equally serious. Prolexic, the DDoS platform, runs 32 anycast scrubbing centers with more than 20 Tbps of dedicated mitigation capacity, zero-second mitigation SLAs, and a 100% platform availability SLA, with network connectivity SLAs of 99.9% to 99.99% depending on how you connect. In practice that means attacks are absorbed upstream of your origin before they touch your infrastructure. IPv6 is supported at the edge — dual-stack delivery with both A and AAAA records, backed by an IPv6 footprint that has been expanding for years. GPU resources are available in the compute regions for inference workloads.

What we cannot do is hand you our own latency or throughput numbers for Akamai. Our dated benchmark suite on benchmarks covers cloud compute instances — Hetzner, DigitalOcean, UpCloud, Vultr, Linode — not CDN edge performance, and we do not invent measurements we have not taken. So treat the performance story above as the vendor's documented architecture plus two decades of public track record, and validate it yourself with a trial on your actual user geographies before you sign. That is how we would evaluate any delivery network.

Where it falls short

The obvious weakness is accessibility. No public pricing, no self-serve onboarding for the products that matter here, and a procurement process designed for six- and seven-figure contracts. For a startup, a blog, a small game community, or a regional business, Akamai is not a realistic option — not because the product fails, but because the buying motion assumes a procurement team.

Second, complexity. The product catalog is deep and heavily modular: delivery, security, compute, and dozens of named products (EdgeWorkers, Ion, App & API Protector, Prolexic, Edge DNS, IP Accelerator) that overlap and are priced separately. Getting the right bundle at the right price requires expertise, and misconfiguration is common enough that many customers route through a managed service partner. The documentation is thorough but written for practitioners who already speak the vocabulary.

Third, cost at scale. Aggressive egress pricing on the cloud side is a genuine strength — a direct inheritance from the CDN heritage where data transfer is the core competency — but enterprise delivery contracts carry volume commitments, and exceeding them is expensive. Buyers should model worst-case traffic, not average.

Fourth, fit. Akamai is not a full hyperscaler. If you want managed Kubernetes with a thousand integrated services, Azure, AWS, or Google Cloud are the comparison set, and Akamai's 25+ core compute regions cannot match their breadth. The compute line is strong for the workloads it targets — streaming backends, ecommerce catalogs, game services, inference endpoints — and thin outside them.

Who should pick this

Our pick for large media and streaming platforms, global gaming backends, major ecommerce operations, and any organization whose revenue depends on staying online through targeted DDoS attacks. If you deliver video at national scale, run live events, or are a regular attack target, Akamai's combination of edge reach and Prolexic scrubbing is the most battle-tested option on the market, and the premium buys you operational calm.

Also reasonable for enterprises with data-residency requirements spread across many geographies: the distributed sites exist precisely to keep workloads inside specific jurisdictions while the edge handles delivery.

We would not pick it for individuals, small businesses, regional sites, or teams that want transparent per-hour pricing and a control panel instead of an account manager. Cloudflare covers the self-serve end of this market, Fastly competes on edge compute ergonomics, and AWS CloudFront is the default for teams already deep in AWS. Our advice is blunt: if you cannot articulate why you need 4,400 points of presence, you probably do not, and a simpler provider will serve you better at a fraction of the cost.

Key takeaways

• The original CDN, founded 1998 from MIT research; 4,400+ edge PoPs across 130+ countries with 1+ PBps of edge capacity.
• Prolexic DDoS defense is first-rate: 32 scrubbing centers, 20+ Tbps mitigation capacity, zero-second mitigation and a 100% platform availability SLA.
• No public list prices for delivery and security — quote-based enterprise contracts with real procurement overhead. We verify no entry price and quote none.
• Akamai Cloud compute (ex-Linode) is a separate, self-serve motion with published rates; judge it on its own, not as part of an edge contract.
• We would pick it for global media, gaming, and attack-prone enterprises; we would send individuals and SMBs to self-serve providers instead.
• Weak spot is fit and accessibility rather than capability: overkill for small workloads, and no match for hyperscaler breadth in managed services.

Sources

• https://www.akamai.com/us/en/about/facts-figures.jsp
• https://en.wikipedia.org/wiki/Akamai_Technologies
• https://www.akamai.com/products/prolexic-solutions
• https://techdocs.akamai.com/api-definitions/docs/set-up-edge-servers
• https://www.akamai.com/site/en/documents/research-paper/2025/idc-report-cloud-evolution.pdf