The browser, the most fought-over piece of software on the internet, is genuinely in play again for the first time in a decade. Perplexity made its agentic Comet browser free on every major platform in March 2026, OpenAI's Atlas is pushing autonomous "agent mode" from a Mac-only beachhead, and analysts now project AI-native browsers could capture 15 to 20% of the browser market by the end of 2026, almost entirely at Chrome's expense. Google still holds around 90% share, but for the first time that number has a credible line of attack.

  • Comet went free everywhere. Perplexity dropped the paywall in March 2026; the full agentic browser now runs free on Mac, Windows, iOS, and Android.
  • Atlas is stronger but boxed in. OpenAI's Atlas has the deepest agent mode but remains macOS-only, with the best features behind Plus and Pro plans.
  • The projection. Analysts see AI browsers taking 15-20% of the market by end of 2026, primarily from Chrome.
  • The real battle. Not Comet vs Atlas, but AI-native browsers versus Chrome and Edge bolting AI onto the incumbent.
Two models of the AI browser AI-native browsers like Comet and Atlas build agentic browsing in from the start, while incumbents like Chrome and Edge add AI features on top of an existing browser. Where the AI lives AI-native Comet (free, all platforms) Atlas (Mac-only, paid agent) Agent mode, cross-tab reasoning, multi-step tasks built into the core Head start today, window closing Incumbent + AI Chrome (Gemini built in) Edge (Copilot mode) ~90% share, huge extension base, sync, dev tools, distribution AI as a feature, not the foundation genztech.blog
Fig 1 The strategic split: agentic browsing built into the core versus AI layered onto the browser a billion people already use.

What changed to make this a real contest?

Perplexity removed the two barriers that keep challenger browsers niche: cost and platform gaps. In March 2026 it made the full Comet browser, including the sidebar assistant, cross-tab reasoning, and agent mode, free across Mac, Windows, iOS, and Android, completing a rollout that began on desktop in 2025. Free-and-everywhere is how you actually take share from a default browser. Perplexity now reports over 100 million monthly active users across its products, and Comet also pushed into enterprise, where admins can deploy it silently across managed macOS and Windows fleets. Distribution, not model quality, is the lever here, and Perplexity pulled it.

RelatedManifest V3 Enforcement Reshapes Chrome Ad Blocking

Why is OpenAI's Atlas both stronger and stuck?

Atlas has arguably the most capable agent mode, able to execute multi-step tasks autonomously and draw on a user's ChatGPT history, but it is hobbled by reach. As of mid-2026 Atlas is macOS-only with no Windows build, which cuts it off from most of the world's computers, and its best agent features sit behind Plus at $20 a month or Pro at $200. A superior product on a fraction of the addressable market loses the distribution game to a good-enough product that runs free on everything. Atlas's advantage is real for power users tied to the OpenAI ecosystem; its ceiling is set by where it will not run.

DimensionCometAtlasChrome
PlatformsMac, Win, iOS, AndroidmacOS onlyEverywhere
PriceFreeFree tier, paid agentFree
Agent modeYes, fullYes, deepestGemini assist
Share postureChallenger, growingNiche, power usersIncumbent ~90%

Can Google and Microsoft simply absorb the threat?

That is the incumbents' whole strategy, and it might work. Chrome now bakes Gemini in for native page summaries, and Edge ships a Copilot mode, which means by late 2026 AI assistance is likely to be standard in every major browser. If that happens, the current edge held by Comet and Atlas, being AI-first, stops being a differentiator, and Chrome's structural advantages, the largest extension library, cross-device sync, best-in-class dev tools, and default distribution on billions of devices, reassert themselves. The challengers' window is open now precisely because the incumbents were slow, and it is closing as they catch up. The honest caveat: reliable "AI browser market share" is not yet measured by standard trackers, so the 15-20% projection is an analyst estimate, not audited data.

What to watch · 2026
  • Atlas on Windows. Mac-only caps Atlas hard. A Windows build would change the math overnight.
  • Measured share. The 15-20% figure is a projection. Watch Statcounter for the first real AI-browser numbers.
  • Agentic traffic. Sites are seeing more agent-driven visits. Watch how publishers and ad models react.
  • Google's defense. Gemini in Chrome is the counter. Watch whether it blunts the challengers before the window closes.

Our take

The browser war is genuinely back, but the winner may not be a browser you have heard of yet. Perplexity made the single smartest move available, going free on every platform, because that is how defaults actually get displaced, and it is why Comet, not the more capable Atlas, is the one to watch. Atlas being Mac-only and part-paywalled is a self-inflicted ceiling that no amount of agent-mode polish fixes. The uncomfortable truth for both challengers is that their core advantage, AI built in, has a short shelf life, since Chrome and Edge are stapling assistants onto browsers a billion people already use. Treat the 15-20% projection as a hypothesis, not a fact, and watch two things: whether Atlas ships on Windows, and whether Google's Gemini integration is good enough to make switching pointless before the challengers convert their head start into habit.

Primary sources

Original analysis by GenZTech. Figures current as of July 2026. Source: perplexity.ai/comet