In 1990, at CERN, Tim Berners-Lee built the first web browser on a NeXT computer. He called it WorldWideWeb (and later renamed it Nexus to disambiguate from the network it browsed). It was a browser and a WYSIWYG editor — the original vision of the web was that anyone could read and write hypertext from the same tool. That vision quickly bifurcated, and the editor half mostly faded.

In 1991, CERN released a portable text-only client called the Line Mode Browser, written by Nicola Pellow, so the web could be reached from terminals that weren't NeXT cubes. Over the next two years, several small graphical browsers appeared (ViolaWWW, Erwise, MidasWWW, Samba), each used by a few hundred to a few thousand people.

The browser that broke the web into the mainstream was NCSA Mosaic, released by the National Center for Supercomputing Applications at the University of Illinois in January 1993. Mosaic ran on Unix, then Windows and Mac, displayed inline images, and had an installer non-academics could actually use. A team led by Marc Andreessen and Eric Bina built it.

In 1994, Andreessen left NCSA, teamed up with Silicon Graphics founder Jim Clark, and started a company originally called Mosaic Communications. NCSA's lawyers objected to the name; the company became Netscape, and on December 15, 1994, it shipped Netscape Navigator 1.0. Within a year Navigator had something like 75% of the browser market. The web exploded out of academia and into living rooms.

Microsoft was late to the web. In 1995 it licensed code from Spyglass — a company that had itself licensed Mosaic from NCSA — and shipped Internet Explorer 1.0 on August 16, 1995, as part of the optional Microsoft Plus! pack for Windows 95. Almost nobody used it. IE 2 and IE 3 closed the technical gap with Netscape; IE 4, in late 1997, was the first version Microsoft genuinely treated as strategic. It was bundled directly with Windows 98 and integrated into the operating system through "Active Desktop."

Microsoft had a specific advantage Netscape couldn't match: every new PC came with Windows, which now came with IE. Netscape Navigator cost money to download (or you got it for free with magazines and at universities). IE was already there. By the late 1990s IE's market share was climbing past Netscape's, fast.

On May 18, 1998, the U.S. Department of Justice and twenty state attorneys general sued Microsoft for using its Windows monopoly to crush competition in browsers. United States v. Microsoft went to trial in October 1998 in front of Judge Thomas Penfield Jackson. The findings of fact came down on November 5, 1999: Microsoft had monopoly power and had abused it. On April 3, 2000, Jackson ruled Microsoft had violated the Sherman Antitrust Act, and on June 7, 2000 he ordered the company broken into two — an OS company and an applications company.

The breakup never happened. In June 2001, the D.C. Circuit Court of Appeals reversed the breakup remedy and removed Jackson from the case (he had given press interviews about Microsoft during the trial), while upholding the underlying antitrust findings. The case settled in November 2001 with behavioral remedies — Microsoft had to disclose APIs, allow OEMs to install competing software, and accept ongoing oversight. No structural separation.

Netscape didn't survive the war. In January 1998 it announced that it would release Navigator's source code; the Mozilla.org project went live February 23, 1998, and the code was released March 31. AOL announced it was buying Netscape that November for around $4.2 billion in stock; the deal closed in March 1999. Netscape the browser limped through a few more releases and was effectively retired in 2008. The seeds of the comeback, though, were already in the Mozilla source release.

Internet Explorer 6 shipped in August 2001 with Windows XP and proceeded to dominate the web for half a decade. Its market share peaked at over 90%. Microsoft, having won, disbanded most of the standalone IE team around 2003. There was no IE 7 in the pipeline. The web stagnated. Web developers spent years working around IE 6's quirks while the rest of the platform sat still.

In November 2004, the Mozilla Foundation — formed July 15, 2003 out of the wreckage of Netscape — released Firefox 1.0. It was small, fast, free, open-source, ad-supported by Google's search box, and it shipped with three things IE 6 didn't have: tabbed browsing, a built-in pop-up blocker, and an extension system that let third parties add features without waiting for the browser vendor.

Firefox grew explosively. By 2009 it was at roughly a quarter of the market and still climbing. Microsoft, jolted out of complacency, reassembled an IE team and shipped IE 7 on October 18, 2006. The first browser war had ended in a Microsoft monopoly; the second one was already underway, and this time the challenger had the moral high ground of open source and open standards.

On September 2, 2008 — over a decade after Netscape, four years after Firefox 1.0, and seven years into IE's dominance — Google launched Chrome. Google announced it the day before via, of all things, a 38-page comic book drawn by Scott McCloud explaining how the browser worked.

Chrome did three things that, in retrospect, decided the next decade. First, it had a brand-new JavaScript engine called V8, written from scratch in Denmark, that ran JavaScript an order of magnitude faster than the competition. Suddenly web apps that had been dog-slow felt instantaneous, and the platform's ambitions could grow accordingly. Second, it used a multi-process architecture in which each tab ran in its own OS process, sandboxed from the others. A crashing tab no longer killed the whole browser; a malicious tab no longer easily reached the rest of your machine. Third, the user interface was almost defiantly minimal — one combined address-and-search bar (the "Omnibox"), tabs above the address bar, and almost nothing else.

Chrome was built on WebKit, the rendering engine Apple had developed for Safari (itself a fork of the older KHTML engine from KDE). For the first five years Chrome and Safari shared upstream rendering code. On April 3, 2013, Google forked WebKit into a new engine called Blink and went its own way — partly for engineering reasons, partly to avoid having to coordinate with Apple on every change.

Chrome arrived late. Within five years it was the most-used browser on the planet, and it has stayed there. As of 2025, it's used by roughly two thirds of all web users globally. The shift was driven by a combination of genuine technical lead, aggressive distribution (bundled with the Google Toolbar, then the Adobe Flash and Avast installers, then with Android), and the network effect of being recommended by every other Google product.

Chrome's dominance reshaped how the web evolves. Google's rendering engine, Google's V8, and Google's view of where the platform should go effectively set the agenda. Many of the modern web's most important features — Service Workers, Web Components, WebRTC, WebGL 2, WebAssembly, and dozens of "Web*" APIs — were shipped first in Chrome and only later picked up by other browsers. Whether you think this is a triumph of momentum or a danger of monoculture depends on your priors.

Microsoft eventually surrendered the engine race. After years of trying to keep up first with Trident (the IE engine) and then with EdgeHTML (the engine in the original Windows 10 Edge, launched July 29, 2015), Microsoft announced on December 6, 2018 that it would rebuild Edge on top of Chromium. The new Chromium-based Edge launched January 15, 2020. The browser wars had effectively been replaced with a single shared engine and a thin layer of brand on top.

The list of browsers built on Blink today reads like a small industry: Chrome, Edge, Opera (which switched from its own Presto engine to WebKit and then to Blink in 2013), Brave, Vivaldi, Samsung Internet, and dozens of smaller projects. WebKit lives on in Safari and — until 2024 — every browser on iOS, because Apple required all iOS browsers to use WebKit underneath. The European Union's Digital Markets Act loosened that requirement in 2024, allowing non-WebKit browsers on iOS within the EU. Adoption so far has been small.

Firefox's market share today is around 3%. By any commercial metric, it lost. By the metric that matters most to the health of the open web, it's the reason there is still a non-Chromium engine in widespread use. Firefox's Gecko engine remains the only mainstream rendering engine not derived from KHTML's family tree, and Mozilla's continued participation in the standards process is a check on whatever Google decides the web should do next.

On November 14, 2017, Firefox 57 — known as Quantum — landed. It was a major rewrite that pulled in components from Mozilla's experimental Servo project, including a parallel CSS engine (Stylo) written in Rust. The result was a browser that competed with Chrome on speed and memory for the first time in years. It came too late to reverse the market trajectory but kept Firefox technically credible.

Mozilla also shaped the modern web in ways that don't show up in browser-share charts: it was the original home of the extension model, the early driving force behind the Web Platform docs (MDN), an early advocate of HTTPS-by-default, and the consistent voice arguing for user privacy as a first-class concern in browser design.

Until the mid-2000s, debugging a web page meant sprinkling alert() statements through your JavaScript and reloading. The DOM was mostly opaque; the network was a black box; CSS was inferred by guessing. Web development was a craft people defended only because they loved the result.

In January 2006, Joe Hewitt — one of the original Firefox developers — released Firebug as a Firefox extension. Firebug let you click an element on the page and see its live HTML, its applied CSS, and its computed styles, in the same window, with edits that took effect immediately. It had a JavaScript console, a network panel, a DOM tree, and a step-through debugger. Compared to what came before, it was magic.

Every major browser's developer tools today are descendants of Firebug. When Chrome launched in 2008 it included an inspector based on WebKit's Web Inspector, which had been heavily inspired by Firebug. Over the next several years Chrome DevTools surpassed Firebug; Firefox eventually built its own first-party DevTools that matched Chrome's; Firebug itself was officially retired in October 2017 when Firefox Quantum landed.

What modern devtools added on top of Firebug's foundations is itself a long list: source maps (so you can debug your original code even when the browser sees only a minified bundle), accessibility audits, performance profilers with flame graphs, mobile-device emulation, network throttling to simulate a 3G connection, the ability to edit a running page's source and have it hot-reload, headless automation through the DevTools Protocol (which is what powers Puppeteer and Playwright), and, since 2016, Lighthouse — a built-in audit tool that scores a page on performance, accessibility, SEO, and best practices.

It's easy to forget how recent all of this is. The shift from "guess and reload" to "step through, profile, audit, and emulate" happened almost entirely in the last twenty years, and it's the reason web development can be a serious engineering discipline rather than a folk practice.

Three rendering engines remain in widespread use: Blink (Chrome, Edge, Opera, Brave, Vivaldi, Samsung Internet, and most smaller browsers), WebKit (Safari, and every browser on iOS until very recently), and Gecko (Firefox and its derivatives). Everything else either runs on top of one of these or has been wound down.

The current shape of the market is: Chrome around 65%, Safari around 18%, Edge a few percent, Firefox around 3%, and a long tail of niche browsers — Brave, Opera, Vivaldi, DuckDuckGo, Samsung Internet on Android, and a few more — sharing what's left. Arc, the most-discussed new browser of the early 2020s, has effectively been wound down by its maker (The Browser Company) in favor of a different product.

Beyond the browsers themselves, the active fights are about privacy (third-party cookies, fingerprinting, tracking protection, alternative attribution proposals), platform power (whether the dominant browser sets the standards, and how regulators should respond), and the browser as application platform (Progressive Web Apps, WebAssembly, WebGPU, and the long-running question of how much of native software the browser can absorb). The browser stopped being just a document viewer a long time ago. The argument now is about what it should be instead.