Megatrend · Cloud & Digital Infrastructure

Why the web loads fast even when the server is on the other side of the planet

Every time you tap play, open a site, or launch a game and it just “appears,” there's a quiet network behind it putting content — and now the computing too — as close to you as possible, instead of sending every request halfway around the world to one central data center. This is the story of CDNs and edge computing — infrastructure that's shifting from a “move bits cheaply” business into a battleground over security and AI.

Category Cloud & Digital Infrastructure Level Sub-theme Maturity Mature, but with a new wave coming (mature + edge-AI) Read time ~13 min
A world map dotted with many tiny points clustered near cities, so data travels a short distance to reach people instead of having to run to a distant central data center.
ภาพประกอบ (hero.png)
Spread out, close to people. Instead of one far-off warehouse, the internet places content at thousands of points near users.

01What it is

Picture this: you're in Bangkok, and the site you open has its real server in Virginia, USA. If every click had to cross the ocean and back, it'd be annoyingly slow — but in reality it isn't, because a “copy” of that site already sits in a data center near Singapore, or in Thailand itself. So you pull it from somewhere close, never reaching the origin. That's the heart of a CDN.

A CDN (Content Delivery Network) is a network of servers spread across the globe that “caches” (cache — keeps a temporary copy of) a site's content — images, video, files, pages — as close to users as possible. When someone asks for it, it's served from the nearest point instead of running back to the origin. The result is faster, cheaper, and more resilient — because if one point goes down, another takes over.

The word “edge” is the next step. Instead of just “keeping a copy” near users, we can now “run code” and process things right at those points. The edge of the network isn't just a shelf anymore — it's a small computer a few milliseconds from you. Put another way, edge is “stretching the cloud” all the way out to the last stop near the user.

Key terms
PoP / Latency / Origin

PoP (Point of Presence) = a location where a provider puts servers, in cities around the world. More PoPs = more points close to users · Latency = the time data takes to make a round trip, measured in milliseconds (ms); lower is better · Origin = the site's real source server, the load the CDN keeps shouldering so it isn't hit by every single request.

This node sits under the Cloud & Digital Infrastructure megatrend as the “last-stop” layer of the cloud — the part closest to users. That's the opposite of siblings like Hyperscale Cloud, the giant central data centers. If hyperscale is the “central brain,” edge is the “nerve endings” spread across the whole body.

02Why it matters — cutting the distance to the user

Its importance comes down to one physical law you can't escape: light is the fastest thing in the universe, but over a long enough distance it's still too slow for humans. A signal running through fiber across continents takes hundreds of milliseconds. That sounds tiny, but one page makes dozens of round trips before it finishes loading, so the latency piles up until you can feel it. Research from both Amazon and Google points the same way: every bit of added latency really does cut sales and usage.

That makes a CDN not a luxury but core infrastructure of the modern internet. And it's not small: the global CDN market is around $26B in 2025 and is expected to reach ~$43B in 2030, growing about 9–11% a year — driven by 4K/8K video that eats huge bandwidth, and by every business moving online.

The CDN and Edge Computing market
Market value (USD billions) — edge computing covers far more ground than CDN · 2030 is an estimate
Source: Mordor Intelligence & MarketsandMarkets (CDN ~$26–27B→$43–45B, CAGR ~9–11%; edge computing $168B→$249B, CAGR ~8%)

But the number that really brings it home is the “closeness” these networks achieve. Today the market leader, Cloudflare, has points of presence in over 330 cities across 125 countries — enough to say 95% of the world's internet users are within 50 milliseconds of its servers, several times faster than a blink. That's what makes the web “feel fast” even when the origin is on the other side of the world.

95% within 50 ms Almost every internet user on Earth is within 50 milliseconds of a leading edge network's points of presence — and about 20% of all web traffic runs through this single network.

03How it works

The heart of the mechanism is a switch from “every request runs to one distant center” to “every request runs to the nearest point that already has a copy.” Compare these two pictures.

CDN/edge cuts the distance to the user Old way: every user runs to one distant central data center, so it's slow. New way: each user pulls from a nearby edge point, so it's fast Old way — one center, far, slow central data center users worldwide every request runs far → ~150 ms / round trip CDN/edge way — spread out, close, fast origin edge point (PoP) user short <50 ms fetch from origin only when not cached
Cut the distance. The old way: every request runs to one distant center (slow) · the edge way: users pull from a point nearby, and the origin is touched only when the cache doesn't have it yet (fast).

The mechanism splits into two layers. The first is “content caching”: the first time someone in a region asks for a file, the edge point fetches it from the origin and stores it. After that, everyone nearby gets it straight from the cache — no need to bother the source again. That slashes both latency and the load on the origin server. For popular content like video, Netflix goes a step further, placing its own “server boxes” (Open Connect) inside ISP networks worldwide at tens of thousands of points — so the movie you watch comes from a machine in your own city.

The second layer is “running code at the edge” (edge compute), which is new and the heart of the future. Instead of just serving static files, an edge point can run small programs right there — checking permissions, personalizing a page per user, blocking bots, even answering an API — without sending anything back to the center. A popular tool like Cloudflare Workers runs this kind of code in a fraction of a second (latency under 100 ms) and is growing about 300% a year.

Key terms
Cache hit/miss & Edge compute

Cache hit = the copy is already at the edge point and can be served instantly (fast) · Cache miss = it isn't there yet, so it must be fetched from the origin first (slower, but it happens once per region) · Edge compute = running a “program,” not just serving a file, at the edge point so processing happens near the user, cutting both latency and central load.

04Where it sits in the cloud

The easiest way to understand edge is to see it as the “outermost layer” of the cloud. Within the Cloud & Digital Infrastructure megatrend, it ties tightly to several neighbors:

  • Builds on Hyperscale Cloud: the giant central data centers are the “origin” that edge distributes and relieves. The two work as a pair — central brain and nerve endings
  • Sits on Towers, Fiber & Colocation: every PoP has to sit in a real colocation facility, wired with real fiber — edge is the “software” running on that “digital real estate”
  • Merges with Cybersecurity & Digital Trust: this is the most important link right now. Because edge sits on the front line against the user, it's the ideal place to filter dangerous traffic (DDoS, bots) — so delivery and defense fuse into a single service
  • Feeds and depends on AI Applications: AI generates enormous traffic (from agents and chatbots) that edge has to absorb, and in turn edge is becoming the place to “run AI” near the user — a two-way relationship

What makes this node interesting as a business is that it's repositioning itself — from being just a cheap “content pipe” into a platform that bundles delivery + security + compute in one place. And that's the story of the next chapter.

05Where it stands now + who the players are

The biggest story in this industry in 2025 is that “pure content delivery has become a commodity”. Once anyone can run a basic CDN, the price per unit of data delivered keeps falling — forcing every player to run upmarket toward “something worth more”: security and compute at the edge. Those who adapt grow; those who cling to the delivery pipe shrink.

A side-by-side comparison: a plain, cheap water pipe anyone can sell, next to a fortress with walls, guards, and a calculator — meaning the business of moving data became a commodity, while security and compute are where the value is.
ภาพประกอบ (commodity.png)
Value moves. Just moving content = a cheap pipe · security + compute = the profitable part.

You see this most clearly in Akamai's numbers, the world's first CDN pioneer. Its “delivery” revenue in 2024 was about $1.32B, down 15% and still falling. But over the same period, revenue from security + cloud computing grew into ~67% of the whole company, with the security side already past $2B. It's the entire industry's “pivot” playing out inside one company.

Akamai: the business is flipping
Revenue growth by segment (% year over year, 2024–2025) — the cheap part shrinks, the valuable part grows
Source: Akamai FY2024 (delivery -15%, security +16%, compute +25%) — in 2025 delivery's decline eased to ~-4%

The one born in the new era and running furthest ahead is Cloudflare, which designed itself from the start to be “CDN + security + edge compute in one place.” Its 2025 revenue hit $2.17B, up 30%, accelerating to +34% in Q4, with a new catalyst: AI — companies are moving to build AI apps on the Workers platform, to the point that Cloudflare closed the largest deal in its history, worth over $100M, off this demand.

Cloudflare's revenue accelerates
Annual revenue (USD billions) — 2026 is the company's own target
Source: Cloudflare 8-K / earnings, FY2025 (2025 revenue +30%, Q4 +34%; 2026 target ~$2.8B)
Key players in this field
Note
We arrange players by role and competitive direction (who's the new-era platform, who's pivoting, who's still a pipe) rather than by raw market-cap size.
CloudflareNET · US
USA · new-era leader
Bundles CDN + security + edge compute into one platform, by design from day one. 2025 revenue $2.17B (+30%), a 330+ city network, and about 20% of all web traffic runs through it. It's becoming a base for running AI apps via Workers.
core · platform leader
AkamaiAKAM · US
USA · the pivoting pioneer
Invented the world's first CDN. Its delivery business is shrinking (-15% in 2024), but it's built security + cloud into ~67% of revenue, with the security side now past $2B — a textbook case of running upmarket from the cheap stuff to the valuable stuff.
core · pivoting the business
FastlyFSLY · US
USA · the developer-focused challenger
A high-performance CDN aimed at developers and edge compute. Q3 2025 revenue $158M (+15%), with the compute side growing strongly at ~64% and about 627 enterprise customers — smaller, but beloved by technical teams.
core · developer-focused
F5FFIV · US
USA · application + traffic
A specialist in application-traffic management and security (load balancing, app security) that sits between users and servers — an important complement to the edge/delivery layer.
core · app delivery/security
Hyperscaler CDNAMZN · GOOGL · MSFT
USA · the cloud owners
AWS (CloudFront), Google, and Microsoft each have their own CDN/edge tied to their cloud — pushing delivery prices into commodity territory, and acting as both partner and competitor to the independent players.
secondary · the cloud's CDN
Wangsu/ China players300017 · CS
China · domestic market leader
Wangsu (and the earlier ChinaCache) is the backbone of China's CDN, a market walled off from Western players by domestic rules — a reflection of how edge is sovereign infrastructure each region wants to control itself.
core · China market leader

06The road ahead — when AI moves to the edge

If CDN is the story of “a past that's becoming a commodity,” the new wave that will define the next decade is “AI inference at the edge” — running AI models near the user. The idea is simple: when you talk to an AI or use a smart feature, it has to “think” (inference) before it answers. If that thinking happens at an edge point close to you, instead of being sent to a far-off AI data center, the answer comes back much faster, saves bandwidth, and your private data doesn't have to travel far.

A small AI sitting at a network point near a user's home, answering questions on the spot without sending them to a giant brain far away.
ภาพประกอบ (edge-ai.png)
AI moves closer. Small models run at edge points and answer instantly, without running to the giant brain at the center.

That's why edge players like Cloudflare are racing to build platforms where developers can run AI on their networks, and it's the source of a big new wave of demand. The overall edge AI market is estimated at around $25B in 2025 and growing at high double digits (edge AI software alone is growing ~29% a year) — a growth engine that helps offset the decline of the old content-delivery business.

The three waves of Edge
The direction of growth — value moving from content delivery toward security and AI at the edge
Source: Compiled from Mordor/MarketsandMarkets & Grand View (approximate CAGR per layer — CDN ~9–11%, security ~15%+, edge-AI software ~29%)

Three things to watch over the next 3–5 years: (1) who becomes the “platform to run AI at the edge” that developers choose · (2) how far the lines between CDN, security, and cloud blur away (they're fusing into one service) · and (3) how long independent players can compete with hyperscalers that have their own edge — because Amazon, Google, and Microsoft all have CDN/edge tied to their cloud and are ready to cut prices.

07Challenges & risks

The first risk — and the axis of everything — is commoditization. Basic content delivery has almost no barrier left, and the price per unit drops every year. Anyone whose revenue still leans on delivery (as Akamai's once did) has to run upmarket toward higher-value work in time, or revenue shrinks along with the price. This isn't theory — it's already happening in the financials.

The second risk is the hyperscaler shadow. Amazon, Google, and Microsoft are all at once customer, partner, and competitor to the independent edge players. They have their own CDN/edge, bundled and priced down with their cloud. So independents have to win on “neutrality” (not tied to any one cloud), performance, and innovation — not price.

The third risk is betting on edge-AI before it's fully proven. Everyone talks about AI at the edge, but the business model, the real size of demand, and who captures the big profits are still open questions. If this investment doesn't pay back fast enough, the players who bet heavily could get hurt. And one last thing — because edge sits on the front line against the user, it's both shield and target: a single big network outage can take down half the internet at once. Reliability is both the biggest selling point and the most fragile weak spot.

The bottom line for investors Edge & Content Delivery is a trend whose “base is mature and becoming a commodity, but with a new wave of value layered on top” — three keys: (1) who moves from the delivery pipe to security + edge compute fastest and deepest (that's where the value is) · (2) who controls the “run AI at the edge” platform · (3) how much hyperscaler price pressure they can withstand — the winner is whoever sells “speed + security + intelligence” bundled together, not just bandwidth.

In short: Edge & Content Delivery is the layer of the internet closest to you. It started with a simple job — putting content near users to make the web faster — but it's evolving into infrastructure that bundles data delivery, security, and AI compute at the same edge. Anyone who still sees it as just a “cheap pipe” may miss the real story: the fight to own the “last meter” between the cloud and every user on Earth.

Explore this theme — live data, stocks & news →