OpenAI just announced ChatGPT Atlas, their own web browser. The browser wars are officially underway.
This matters because whoever controls the browser controls how billions of people find information and make purchasing decisions. Right now, Chrome dominates with roughly 65-70% of global browser usage. But OpenAI, Perplexity, and Google are all fighting to own the next generation of browsing.
Here’s what’s happened, why it matters, and what I think happens next.
What Are AI Agent Browsers?
An AI agent browser is a regular web browser with a built-in large language model. The AI can browse the internet for you and take actions in the browser itself.
These browsers have a side panel where you chat with the AI. The AI can read pages, click buttons, fill out forms, and complete multi-step tasks on your behalf.
Right now, they’re slow and limited. But they’re improving quickly. I believe we’ll eventually have another ChatGPT moment once these browsers can truly perform tasks at a human level.
The Current Players
Perplexity Comet
Perplexity launched Comet as the first major AI browser. Initially, it cost $200 per month. They released a free version this month.
Comet is built on Chromium. It focuses on research and citations. The AI provides answers with clickable sources. It can organize tabs into projects and track your research goals across sessions.
Comet works on Windows and Mac. The core browser is free. Comet Plus costs $5 per month for premium features.
ChatGPT Atlas
OpenAI announced Atlas immediately after Perplexity released its free version. That probably wasn’t a coincidence.
Atlas integrates ChatGPT into every browsing session. The sidebar understands whatever page you’re viewing. It can answer questions, edit pages, and remember your browsing history for personalized suggestions.
The full Agent Mode is only available to ChatGPT Plus, Pro, and Business subscribers. This mode can open tabs, scroll pages, click buttons, and fill forms to complete complex tasks automatically.
Atlas currently only runs on macOS. Windows, iOS, and Android versions are coming soon.
Chrome with Gemini
Google already owns the most popular browser. They’re adding Gemini AI directly into Chrome.
Right now, Gemini in Chrome is limited. It can summarize articles and compare products. It works across multiple tabs. But it can’t take action on your behalf yet.
Google says agentic features are coming in 2026. When they arrive, Gemini will be able to book appointments, order groceries, and navigate web pages for you.
Chrome’s advantage is distribution. They don’t need to convince anyone to switch browsers. They already have 3 billion users.
Anthropic’s Claude Extension
Anthropic makes Claude, which I think has the best agent architecture right now. They’re not building a standalone browser. Instead, they have a Chrome extension in a limited preview.
The extension lets Claude see and act on your browser. It’s aimed at enterprise users, not consumers. Early tests show it can manage calendars, schedule meetings, draft emails, and handle expense reports.
Anthropic targets enterprises and developers. Their models are expensive. This positions them differently from OpenAI or Perplexity, which focus on consumers.
Why the Browser Wars Matter
Two things drive this competition: distribution and data.
Distribution
Browsers are the gateway to the internet. Whoever controls the browser controls how people find information and complete tasks online.
Chrome dominates now. Apple, Microsoft, OpenAI, and Perplexity are all fighting for position. Perplexity even made a public bid to buy Chrome outright.
AI companies want to own the platform, not just provide a chatbot. They want to control where people go for recommendations and decisions.
Data
Browsing data is valuable for training AI models. An AI browser can learn from millions of browsing sessions.
Every site visited and task delegated becomes potential training data. Even if companies say they won’t train models on user data without consent, they still gather aggregated usage patterns.
Google’s data from Chrome is perhaps its biggest advantage right now. All the model makers want that same data stream.
This is why OpenAI offers a week free of Pro if you set Atlas as your default browser.
Are These Features Free?
Basic AI assistance is generally free. Advanced agentic functions require paid subscriptions.
ChatGPT Atlas: The browser is free for all users. Full Agent Mode requires ChatGPT Plus, Pro, or Business ($20-$200 per month).
Perplexity Comet: The core browser and AI search are free. Comet Plus costs $5 per month for curated news feeds.
Chrome with Gemini: Free for all Chrome users in the U.S. Google has previously made Pro-only features free for desktop users.
I’m guessing we’re still a few years away from agentic capabilities being cheap enough to offer to everyone for free.
Who Wins the Browser Wars?
This is really Google’s game to lose.
Unlike OpenAI and Perplexity, they don’t need people to change browsers. Chrome is already the most popular browser. If Anthropic somehow wins, Google still wins because its extension runs on Chrome.
Google already has a product called Gemini with the same agentic capabilities as Comet and Atlas. I fully expect them to integrate this into Chrome in the near future.
Once these models become capable enough, people will choose whatever is most convenient. That’s probably whatever is already in your default browser.
The data Google collects from Chrome is its biggest advantage. All the model makers want is the data to train their models. But Google already has it.
What I Think Happens Next
Everything comes down to distribution and data at this point.
The models are becoming commoditized. They’re all getting sufficiently capable. Eventually, people will choose the most convenient option, not the smartest model.
Atlas can run multiple agents across multiple tabs right now. Each tab can work on a separate task simultaneously. This is the direction things are heading.
Once these browsers become capable enough, anyone using their computer for work will gladly pay for a subscription. We’re not there yet. But we’re getting closer.
I believe we’ll have another ChatGPT moment once we have a fully functional agent browser that performs tasks at a human level. That moment is coming.
For businesses, this means the way people find you is changing. People are already using ChatGPT and other AI tools to make purchasing decisions. As these browsers improve, more decisions will be made within AI interfaces.
That’s why AI optimization matters more than traditional SEO. You need to show up where people are actually searching. And increasingly, that’s inside large language models.
Ready to optimize your business for AI search? At TJ Digital, we help businesses show up in ChatGPT, Google’s AI mode, and other AI platforms. Get a free digital marketing audit to see how your business can stay ahead of these changes.