The first thing Akamai delivered over the commercial internet was almost nothing: a single pixel, tucked deep inside Disney's website in February 1999. It was a modest proof that a radical idea from an MIT mathematics lab could survive outside the lab. Put copies of digital content on servers nearer to the people asking for it, then use algorithms to choose the best route. A month later, the young company faced less modest assignments - a burst of March Madness traffic for ESPN and the rush to watch a new Star Wars trailer.
Those events supplied a compact education in internet physics. Popularity arrives unevenly. Distance creates delay. A server that looks ample on Tuesday can become a smoking crater on Friday. Akamai turned the chaos into a service, launched commercially in April 1999 and named Yahoo as a charter customer. The business was built to make itself disappear. When it worked, a page simply loaded or a video simply played.
A quarter-century later, the company is still selling proximity. What changed is everything proximity can be used for. The server close enough to speed a stream is also close enough to inspect a request, block an attack, execute code, store data or run an AI model. Akamai now describes itself as a cybersecurity and cloud computing company. That is not merely positioning. Security produced $2.243 billion of its $4.208 billion in 2025 revenue. Delivery remains large, but the corporate center of gravity has shifted.
01 / OriginThe World's Wait Problem
Akamai's origin story begins in 1995, when World Wide Web inventor Tim Berners-Lee challenged colleagues at MIT to do something about looming web congestion. Applied-mathematics professor Tom Leighton saw a routing problem. Graduate student Danny Lewin helped turn that insight into algorithms for distributing and replicating content across a large network of servers. The pair entered an MIT entrepreneurship competition with business student Preetish Nijhawan. In August 1998, Leighton and Lewin incorporated Akamai, with Jonathan Seelig and Randall Kaplan joining the founding team.
The idea sounds obvious now because it worked. In the late 1990s, the normal model was to serve a website from an origin data center, even if the visitor was continents away. Akamai placed content inside many networks and directed each request toward a nearby healthy server. It shortened the trip and spread demand. Its expertise became a peculiar blend of applied math, distributed systems, relationships with network operators and round-the-clock operations.
Akamai's most valuable product may be its position in the path - close to the user, close to the application and close to trouble.
02 / ProductOne Network, Three Jobs
The current portfolio can be read as three layers. Delivery products such as Ion, Adaptive Media Delivery, API Acceleration and Global Traffic Management keep applications, software downloads and video moving. EdgeWorkers lets developers run JavaScript near users; EdgeKV stores small amounts of data nearby; mPulse measures what people actually experience in their browsers. This is the original Akamai proposition made programmable.
Security converts network visibility into defense. App & API Protector sits in front of internet-facing software. API Security finds and analyzes interfaces that organizations may not know they expose. Prolexic absorbs distributed denial-of-service attacks. Bot Manager separates humans from automated traffic. Guardicore Segmentation limits how far an intruder can move inside a network. Akamai also sells zero-trust access, DNS protection, account-abuse defenses and managed expertise for organizations that do not want to operate every control themselves.
Cloud is the newer layer. The 2022 purchase of Linode supplied developer-friendly compute, storage and managed Kubernetes. Akamai Connected Cloud, now presented simply as Akamai Cloud, combines those core services with the edge network. It is not an attempt to reproduce every obscure service in a hyperscaler's catalog. The differentiation is distribution: put workloads, data and security nearer to where demand happens, with simpler pricing and less exposure to data-egress charges.
Delivery
Cache, route and optimize websites, APIs, software and streams.
Security
Inspect traffic, stop attacks and contain movement across networks.
Cloud
Place compute, data and AI inference closer to users.
03 / MarketThe Company Behind the Customer
Akamai sells mostly to organizations for which a slow page, failed stream or breached API becomes expensive immediately. Its customer base spans media, games, retail, banking, telecommunications, healthcare, software and government. The company says all of the top 10 video streaming services, all of the top 10 video game companies, all of the top 10 banking companies and all six branches of the U.S. military trust it. Customer announcements offer public glimpses: ITV used Akamai to help launch the Freely streaming service, while Miele has described using the platform for ecommerce.
The buyer might be a chief information security officer worried about ransomware, a streaming engineer planning for a live final, a retailer distinguishing shoppers from inventory-scraping bots or a developer deploying an application in several markets. Akamai charges through subscriptions and consumption-based contracts tied to traffic, capacity, users, protected assets and service levels. Professional and managed services add human expertise. Large customers often buy several products, which makes the shared platform commercially important as well as technically elegant.
That map also explains the messy competitive set. Cloudflare and Fastly challenge Akamai in edge delivery and developer services. Amazon Web Services, Microsoft Azure and Google Cloud dominate centralized cloud infrastructure. Zscaler, Palo Alto Networks, F5, Imperva, Radware, Netskope and others meet it in pieces of security. Akamai's pitch is that enterprises can buy a distributed layer that spans all three categories. The risk is equally plain: specialists can go deeper and hyperscalers can spend more.
04 / NextLatency Gets an AI Sequel
Artificial intelligence has given Akamai a familiar problem wearing expensive new hardware. Training giant models rewards enormous centralized clusters. Inference - the moment a deployed model answers a user or agent - rewards speed, predictable cost and proximity to data. In March 2026, Akamai said it would acquire thousands of NVIDIA Blackwell GPUs for a distributed AI platform. It is working with VAST Data on data-intensive inference and promoting Akamai Inference Cloud as a place to run models nearer to users.
The financial signal arrived in May: Akamai disclosed that a leading U.S.-based frontier model provider had committed $1.8 billion over seven years to cloud infrastructure services. The customer was not named. In the same quarter, cloud infrastructure revenue rose 40 percent from a year earlier to $95 million, while security rose 11 percent to $590 million. Delivery and other cloud applications declined 7 percent to $389 million. The transition is visible in four numbers: a mature engine shrinking, two newer engines growing, and one unusually large commitment making the cloud bet harder to dismiss.
AI also creates traffic that needs policing. Akamai reported a 300 percent annual increase in AI bot activity. Its 2026 products reflect the ambiguity. Firewall for AI protects AI applications. AI Brand Presence helps companies see AI crawlers and prepare content for machine-mediated search. A partnership with Visa seeks to authenticate shopping agents, because tomorrow's merchant cannot simply block every bot. The useful question becomes which machine is visiting, whom it represents and what it is allowed to do.
LayerX, acquired in July 2026 for about $205 million, pushes that control into ordinary browsers. Renamed Akamai Workforce Protector, its technology observes how employees interact with web content, SaaS tools, prompts and file uploads without requiring a proprietary browser. It extends the security perimeter to the point where workers meet generative AI. The acquisition also fits Akamai's long method of reinvention: buy a focused product, attach it to the global platform, then sell it to enterprise customers already passing through the network.
05 / VerdictA Bet on Being Nearby
Akamai is neither an insurgent startup nor a sleepy utility. It is a 1990s internet company trying to make old infrastructure useful for each new era. The CDN created reach. Reach produced traffic intelligence. Traffic intelligence supported security. Linode added general-purpose cloud building blocks. GPUs now test whether the same distributed footprint can become meaningful AI infrastructure.
Customers can use Akamai narrowly - to protect an API, absorb an attack, speed an ecommerce site or stream a match - or as a broader operating layer. The practical appeal is consolidation near the network edge. The practical caution is complexity: combining many acquired and homegrown products into a coherent experience is difficult, and the alternatives are formidable.
Still, Akamai owns something hard to manufacture quickly: a working relationship with the internet's geography. Its software and servers already sit near audiences, applications and adversaries. In 1999, that closeness helped a Star Wars trailer arrive before impatience did. In 2026, Akamai wants it to help an AI answer faster, a bank stop a bot and an employee keep sensitive data out of the wrong prompt. Different request, same instinct: meet the internet before the internet reaches the customer.