Originally, Browser Wars were fought on terms like speed, features and browser compliance. Today there’s a new war driven by artificial intelligence and Agentic features built into browsers. The next-generation browsers promise to act as smart agents or researchers that can understand content and perform tasks on behalf of users.
As of July 2025, Google Chrome has an iron grip on the browser market with 66.6% global market share, followed by Safari at 18%, Edge at 5.2%, and Firefox struggling at just 2.9%.
But, why would anyone build a new browser in 2025? AI features in early attempts have shown a revolutionary shift in capabilities. Although none of the projects so far like Anthropic’s Computer us or OpenAI’s Operator truly caught on, what’s clear is that a browser can be an intelligent assistant, and that’s an opportunity to challenge Google’s Chrome.
Google’s Strategy
Google’s approach to the AI browser race has been two-fold: enhance search with generative AI and gradually inject AI assistance into Chrome. Google introduced its Search Generative Experience (SGE) – a lab feature where search results are augmented with AI-generated summaries and answers at the top of the page. But they’re going to have to try harder.
The stakes couldn’t be higher: Google’s search advertising generated $49.4 billion in Q3 2024 alone, representing 75% of its total ad sales. Every percentage point of browser market share lost to AI competitors threatens this revenue stream, especially as these new browsers keep users within conversational interfaces rather than directing them to ad-supported search results.
Project Mariner, unveiled in December 2024 and currently available to Google AI Ultra subscribers ($249.99/month), represents the search giant’s ambitious attempt to transform Chrome into an autonomous web agent. TechCrunch Built on Gemini 2.0, Mariner can take control of the browser, navigating websites, filling forms, and completing multi-step tasks based on natural language instructions.
Mariner’s current limitations reveal weaknesses that could be exploited by othres. Actions execute with 5-second delays between movements, and like othes, the system only operates in active browser tabs requiring supervision, and it cannot complete purchases or accept terms of service without user intervention.
OpenAI Enters and Re-Enters the Fray
According to reports OpenAI is close to releasing an AI-powered web browser with a built-in ChatGPT-based agent. This browser is expected to have ChatGPT as a “native” part of the interface, meaning users can converse with the AI anywhere, anytime, without needing to visit the ChatGPT website.
An OpenAI exec even testified that OpenAI would be interested in buying Chrome if Google were forced to sell it. Instead, OpenAI opted to build from scratch. This opens a direct challenge to Google on its home turf, which OpenAI had done earlier by creating an extension that reset the default search in browsers like Chrome. This browser could mark the beginning of a very different web experience where your primary interface is a chat box that does things, rather than a search box that returns links.
The strategic rationale is clear: direct access to user browsing data without Google intermediation, positioning to challenge Chrome’s dominance, and creating new revenue streams beyond API and subscription models. OpenAI has even hired two longtime Google VPs who were part of the original Chrome development team.
The Perplexity Factor
Perplexity AI is well known for its AI-powered answer engine. Recently, the company went a step further and launched its own web browser called “Comet.” Comet was built with AI at its core from the ground up. Comet includes the “Comet Assistant,” an embedded AI agent that can interpret page content, automate workflows, and answer complex questions using context from what you’re browsing.
Perplexity targets technical users – developers, researchers, data professionals, who might benefit from an AI that can handle dense information and multi-step tasks. A data analyst could use Comet to pull data from multiple sources and have the assistant consolidate it, or a developer could ask the assistant to navigate API documentation and extract a specific code snippet. Voice and text queries are supported, so you could speak a command. Perplexity is a research assistant. but it is only currently accessible with a steep $200/month plan.
Will it work?
These new generation of browsers are aiming to be the “native interface” of the browser. If these ideas catch on, the way we think about “using the web” could change and be more about dialoguing with an assistant than clicking links and reading.
Today’s browser war is all about intelligence. It’s a battle to see who can transform the web from a static place you navigate yourself into a dynamic environment that navigates for you. In this new war, data, AI models, and integration finesse are the weapons, and both the giants and upstarts have entered the arena.
