The New Browser War: Why AI Companies Want to Own Your Starting Point

Authors Note: I first drafted this about nine months ago, before AI-native browsers were suddenly everywhere and before Perplexity and Arc became regular talking points. Google was already experimenting with Gemini-powered summaries in Search — but it still felt early enough that I wasn’t quite sure where things were heading. I never hit publish. So this is me pulling it out of the archive, giving it a quick tidy, and finally letting it see daylight.

Once Upon a Time, Browsers Were Just Windows

The early web wasn’t pretty. Functional, text-heavy, faintly academic. And then Mosaic arrived. Marc Andreessen didn’t invent the browser, but he did help build the first one that supported inline images — which suddenly made the web feel like a place you could explore, not just query.

From Mosaic came Netscape, which dominated until Microsoft realised the browser was becoming a strategic beachhead. Internet Explorer was bundled with Windows, and that was that. IE won the first browser war so completely that, for years, millions of people genuinely believed the blue “e” was the internet.

But dominance breeds complacency. After Microsoft was hit with antitrust action and IE stagnated, Mozilla seized the moment. Firefox felt modern again and briefly became the browser for people who cared.

That run ended when Chrome arrived. Yes, it was lighter and faster. Yes, Google had more resources. But the deeper reason Chrome won is simpler: Google had already become the gateway to the web.

And Then the Browser Became the Search Engine

Google made the web feel navigable. Instead of remembering where to go, you just typed what you wanted. And when Chrome collapsed the address bar and the search bar into one, something subtle but consequential happened: the browser became the search engine. Most online journeys now began with a question, not a URL. And our habits shifted with it. We stopped wandering link to link and started treating Google like home base — a place to ask, collect the fact we needed, and retreat back to before starting the next search.

And that shift was enormously powerful. Once Google owned the on-ramp to the internet — the place where every journey started, the habit people repeated dozens of times a day — its position became unshakeable. Controlling discovery meant controlling attention, and controlling attention meant controlling the economics of the web.

It’s a big part of why Google didn’t just become a successful tech company, but one of the most valuable businesses on the planet. When you own the front door, everything else flows through you.

That shift turned search into the primary interface of the internet — and quietly set the stage for the next one.

Google’s Slow Slide Into “Answer Mode”

It started innocuously. Google began showing quick answers in a sidebar: population sizes, opening hours, definitions — often pulled straight from Wikipedia or structured data. Then came featured snippets: those neat, extracted paragraphs placed above all the organic results.

Useful, but they pulled attention (and clicks) away from the sources underneath. Why navigate to a new site when Google could give you the gist without you ever leaving? And as these quick answers expanded, content sites that relied on search traffic for ad revenue began to feel the impact. Google — and later Chrome — kept more and more of the user’s time inside its own ecosystem. The search engine wasn’t just organising the web; it was absorbing behaviours that had previously belonged to publishers.

Fast-forward to today and Google is generating full AI-powered summaries. First using Gemini, and now increasingly blending that into the core Search Experience, the company no longer just points you to the web — it does the reading for you. And this shift was already underway before I wrote the original draft.

Which brings us to the third interface: the answer engine.

The Quiet Rise of the Chat Layer

A new wave of AI tools — from OpenAI’s browsing-enabled ChatGPT, to Perplexity, to Arc’s AI features — no longer just surface links. Increasingly, they summarise the web for you.

You ask a question, and instead of ten blue links, you get a tidy paragraph. One answer. Neatly packaged. Delivered by a chatbot that’s read the web on your behalf.

As The Browser Company put it in a recent letter:

“Chat interfaces are already acting like browsers.
They search, read, generate, respond.
They interact with APIs, LLMs, databases.”

In this model, the search engine doesn’t disappear — it just stops being the first place we turn. And I’ve felt that shift myself. For years, Google was the starting point for 90% of my web sessions. Now, almost without noticing, that starting point has become ChatGPT. It’s where I go first — not to browse the web, but to understand it.

And once you see that pattern in your own behaviour, it becomes obvious why OpenAI wants to do what Google did with Chrome: own the app, own the wrapper, own the starting point. If the chat interface becomes the new gateway to the web, the real prize isn’t the model — it’s the browser that delivers it.

Why AI Companies Are Coming for the Browser

If chat becomes the first place people turn, the browser becomes strategic all over again. Not as a passive window onto the web, but as the delivery mechanism for whichever model sits behind the interface. And that’s why the model companies suddenly care about browsers in a way they didn’t a year ago. If you can make your AI the default starting point — the place people open reflexively to ask, plan, compare, book — you don’t just win usage. You win distribution. And controlling distribution has always been the most valuable position in tech.

Google understood this better than anyone. Chrome wasn’t just a great browser; it was a way of ensuring that every question flowed through Google Search. Owning the browser meant owning the audience. Owning the audience meant owning the economics.

The AI companies have absorbed that lesson. They don’t just want to answer questions — they want to control where the questions begin. And the moment the chat interface becomes the default way people interact with the web, the browser turns into the new front door. Whenever there’s a new front door, there’s a fight over who gets to guard it.

Google will want to keep that position. It needs Chrome — and the search box embedded inside it — to remain the starting point. But the model makers want their own equivalent of Chrome, because whoever owns the entry point owns the flow of attention, commerce, and data.

This is why OpenAI is shipping apps and wrappers. It’s why Perplexity is experimenting with its own desktop experience. It’s why Arc is repositioning itself as an “AI-native browser.” They all know the same thing: the interface is becoming more valuable than the content underneath it.

If AI answers most of our questions before we ever reach the underlying sites, the browser becomes the choke-point — the distribution layer, the place where the value pools. Which explains the urgency. Whoever wins the browser wins the interface. And whoever wins the interface wins the next decade of the internet.

What Comes Next?

If I had to bet, I’d place it on a simple but profound shift: the chat interface replacing the search box. Not as a novelty or optional mode, but as the primary way we ask the internet for help. The moment that happens, the centre of gravity moves — and suddenly there’s a race for who owns the browser that hosts that interface. Will it be Google, clinging to its incumbency? Or one of the model companies building faster than Google can integrate?

If this shift plays out the way I think it will, we’ll see a steep drop in human traffic to the web. Not because people don’t care about the web, but because most of their questions will be answered before they ever reach it. And those answers won’t come from a single page; they’ll be mashed, averaged, and synthesised across dozens. Incredibly useful for users, but devastating for publishers. If fewer people arrive, fewer people click, and fewer people even know which sources shaped the answer, the open web starts to thin out.

Even so, browsing won’t disappear. People will always want to visit specific pages — to read something at the source, to inspect details, to use tools, to get the full context. So these AI-first browsers will still need to be capable browsers. Browsing just becomes the secondary activity rather than the default one.

And then there’s transacting. We can already see where this is heading. The model companies want you to buy straight through the chat interface. MCP is the early signal: restaurant bookings, hotel reservations, flight search — all the things Google’s been awkwardly bolting onto Search for a decade — but handled conversationally instead of through clunky forms and ad-choked GUIs. Why fight a badly optimised booking widget when you can simply say, “Find me a table at seven,” and be done?

So the future isn’t just a new search paradigm. It’s a new interface for information, navigation, and commerce — all bundled into the conversational layer. And the real battle will be over who gets to be the place where that conversation happens.