Why Nobody Could Take the Browser Crown From Chrome, Not Even AI
By Creatives Takeover Editorial Team · September 23, 2026
Six companies built AI browsers to replace Chrome. Neither one worked.
Between October 2025 and the first half of 2026, the browser category saw more genuine competition than it had in over a decade. OpenAI launched ChatGPT Atlas on October 21, 2025. Perplexity took its Comet browser free worldwide on October 2, 2025. The Browser Company, having quietly put its earlier product Arc into maintenance mode, pivoted its entire team toward an AI native successor called Dia. Opera launched Neon at nearly $20 a month. Microsoft deepened Copilot inside Edge. Brave kept building out its Leo assistant across an installed base of more than 70 million users.
By mid-2026, all that effort had produced real, measurable growth. It had not produced a dent. StatCounter data still put Chrome above 65 percent of global browser traffic, roughly where it had sat for years. This is the part of the story that gets skipped in most coverage of the AI browser wars: it was never really a fair fight, and the clearest proof of that isn't the market share numbers. It's what happened when the United States government tried the same thing with the full force of federal antitrust law behind it, and still lost.
The Government Already Tried This, and Lost
In August 2024, US District Judge Amit Mehta ruled that Google had illegally maintained a monopoly in general search and search advertising, in violation of Section 2 of the Sherman Act, through a web of exclusionary distribution agreements with device manufacturers, browser developers and wireless carriers. The Department of Justice's proposed remedy was direct: force Google to divest Chrome. Perplexity, sensing an opening, submitted an unsolicited $34.5 billion offer to acquire the browser outright, even before Google had confirmed any intention to sell.
On September 2, 2025, Mehta issued his remedies ruling, and it did not include a breakup. He wrote that a Chrome divestiture would be "incredibly messy and highly risky," and that the government had "overreached" in seeking it. Google was ordered to end exclusive default agreements and share limited search data with qualified competitors, but it was explicitly permitted to keep paying Apple, Samsung and other manufacturers to preload its products, Chrome included, as long as those deals were no longer exclusive. Alphabet's stock rose 8 percent the same day.
This is worth sitting with. A federal court had already found Google guilty of illegally protecting its dominance. The government asked for the browser itself to be taken away as punishment. The judge said no, calling it too disruptive to unwind. If the US legal system, backed by a formal monopoly finding, could not separate Chrome from Google, the idea that a startup could out compete it into irrelevance was always going to be a much longer shot than the funding rounds suggested.
The Moat Was Never the Rendering Engine
Here is the detail that made the AI browser wars stranger than most people realized while they were happening: nearly every one of these challengers was built on Chromium, the open source browser engine Google itself maintains. Atlas runs on it. Comet runs on it. Edge runs on it. Even Dia, the most independently branded of the group, inherited Chromium's core from its Arc lineage. Brave and Opera Neon are Chromium browsers too.
That means the actual competition was never happening at the level of rendering speed, web compatibility or core browser engineering, the terrain where a genuinely better product could plausibly win. It was happening entirely in a thin AI layer stacked on top of infrastructure Google built and gives away for free. Every challenger was, in effect, running a marketing and feature campaign on top of Google's own foundation, which meant Google could always fold the winning ideas back into Chrome itself without having to rebuild anything underneath.
Distribution Beats Everything Else
The antitrust case laid out, in plain legal language, exactly what the real moat had always been. Judge Mehta found that Google's default placement on the iPhone, secured through billions of dollars in annual payments to Apple, had allowed the company to "evolve into an internet powerhouse, insulated from competitive threats." Chrome itself, the ruling noted, facilitates roughly a third of all Google web searches simply by existing as the pre-installed browser on the world's dominant mobile operating system.
Perplexity CEO Aravind Srinivas said the quiet part out loud in interviews well before Comet's launch, when he described the company's efforts to get phone manufacturers to preload Comet instead of Chrome: "It's not easy to convince mobile OEMs to change the default browser to Comet from Chrome." That single sentence describes the entire structural problem better than any market share chart. A better product still has to clear a distribution gate that the incumbent built, funded and legally defended, and that gate did not open just because a more capable AI assistant showed up on the other side of it.
The Growth Numbers That Still Weren't Enough
None of this means the AI browsers failed to attract users. Comet's growth was genuinely fast: from roughly 2.4 million monthly active users in early 2025 to somewhere between 10 and 18 million a year later, depending on the measurement window, with a run to the number three spot on the US App Store within 48 hours of its iOS launch in March 2026. Perplexity's broader platform, including Comet, crossed 100 million monthly active users across all products by April 2026.
Set against Chrome's installed base, the same growth curve tells a different story. Even optimistic industry projections put Comet's actual browser market share at roughly 2.5 percent by the end of 2026. Atlas remained macOS only for months after launch, cutting off the largest share of the desktop market by default. Dia carried a roughly $20 monthly subscription in a category where the incumbent is free. Ten million new users is a real achievement for a year old product. It is also, mathematically, nowhere close to a dethroning.
Google Didn't Rush, and Didn't Need To
Perhaps the clearest signal of how little urgency Google actually felt is the pace of its own response. Gemini arrived inside Chrome in September 2025 as a minor floating window, barely more than a bolted on feature. It wasn't until January 28, 2026, three months after Atlas and Comet had already launched, that Google moved Gemini into a persistent sidebar and introduced agentic "auto browse" features for its paying subscribers. The bigger move came at Google I/O in May 2026, when the company unveiled Gemini Spark, a persistent AI agent that runs continuously on Google's own cloud infrastructure and is built to operate directly inside Chrome, alongside WebMCP, an open standard designed to let websites expose their functions directly to browser based agents. By that point, Google also reported that AI Mode in Search had passed 1 billion monthly users.
Google could afford to watch its challengers ship first, study what worked, and fold the winning ideas into an installed base of billions rather than build a following from zero. That sequencing is not something available to any startup, no matter how good its product is.
Why "The AI Wasn't Good Enough Yet" Is the Wrong Lesson
The easy explanation for why AI browsers stalled is technical: they still hallucinate on complex tasks, agentic navigation succeeds roughly 80 percent of the time rather than reliably, and early security testing on Comet found phishing attempts slipping past its protections at a meaningful rate. All of that is true, and none of it is the actual reason Chrome kept its crown.
A technically flawless AI browser still runs on Chromium. It still needs a phone manufacturer willing to override a default that Google has spent billions of dollars a year defending, in agreements a federal court explicitly allowed the company to keep making. It still launches into a market where the incumbent can watch, wait, and ship the same feature into an audience that never had to download anything. The product gap was real, but it was never the binding constraint. The distribution gap was, and no browser, however well built, closes a distribution gap by being better.
What This Actually Means for a Founder Building Right Now
The instinctive takeaway is "build something so good users switch anyway," which undersells exactly how hard that bar is when the incumbent's advantage is structural rather than technical. The sharper version of the lesson is this: before entering a category with an entrenched leader, ask whether what's actually keeping people there is a product decision or a distribution arrangement, because those two problems require completely different strategies to solve, and only one of them can be beaten by writing better code.
Perplexity did the hard part. It shipped a genuinely well reviewed product, grew faster than almost any browser challenger in the category's history, and still ended up fighting over low single digit market share, because the real opponent was never Chrome the product. It was Chrome the default, backed by contracts, an operating system, and a legal system that, even after finding the company guilty, decided the browser itself wasn't worth the risk of pulling apart.
Five Things Worth Taking From This
A federal monopoly ruling and a startup's better product face the same wall for the same reason. The court found Google guilty of illegal conduct and still declined to separate it from Chrome, calling the risk of disruption too high. That is a stronger form of the exact obstacle every AI browser ran into.
Building on top of your competitor's infrastructure caps how much you can actually compete. Nearly every AI browser runs on Chromium, meaning the real contest was confined to a thin feature layer that the infrastructure's owner could always copy and fold back in.
Fast user growth and meaningful market share are not the same evidence of winning. Comet's climb from a few million to double digit millions of users was real. Projected against Chrome's base, it still landed around 2.5 percent, a reminder that growth rate and market position answer different questions.
An incumbent with distribution can let challengers move first and still win by moving second. Google's slowest, most incremental response to the AI browser wave still arrived inside an installed base no competitor could match, turning speed to market into a much smaller advantage than it looks like on paper.
The hardest constraint to solve with a better product is one that was never about the product. Every technical shortcoming in the early AI browsers was fixable with time and iteration. The default placement, the OEM contracts and the legal protection around them were not obstacles any of these companies had the leverage to remove.
Chrome didn't survive the AI browser wars because it was the best browser. It survived because the thing every challenger needed to take from it, the default, was defended by money, by contracts, and in the end, by a federal judge who looked directly at the case for breaking it up and decided it wasn't worth the risk.