The A.I. Bubble Is Coming for Your Browser

There’s an old business maxim dating to the California gold rush: it’s easier to make money selling picks and shovels to aspiring miners than to strike it rich finding gold. Artificial intelligence is in a picks-and-shovels phase right now. If gold, in this metaphor, is artificial general intelligence—a machine smarter than a human—or some version of a digital god, then tech companies are snapping up the tools to create one, including graphics-processing units, data centers, and trained A.I. models. That scramble is why Mark Zuckerberg is paying a twenty-four-year-old A.I. researcher two hundred and fifty million dollars to work at Meta, and why Sam Altman, the C.E.O. of OpenAI, recently said that the company would spend “trillions of dollars” building infrastructure. It’s why Google spent nearly two and a half billion dollars to hire some of the leaders and research staff of Windsurf, an A.I. coding tool, and to license its technology. Insiders have described San Francisco as being in a new state of A.I.-gold-rush fervor, but the true gold has yet to be found; all of the major generative artificial-intelligence businesses are unprofitable. The race is on to find something, anything, that works.

This is also why Atlassian, the conglomerate behind workplace-productivity software including Jira and Trello, acquired a niche A.I. startup called the Browser Company this month for more than six hundred million dollars. Whereas many tech deals are conducted in equity of ambiguous value, this one was done in cold, hard cash, though the product itself is by no means a sure bet. The Browser Company’s core offering, an A.I.-powered “smart” web browser called Dia, was not publicly available before the purchase and had fewer than a hundred thousand active invite-only beta testers. Still, the leadership of Atlassian deemed it a worthy investment because it hints at how the tech industry hopes we’re all going to interact with A.I. in the future—as a seamless element of our daily lives. Josh Miller, a co-founder of the Browser Company, told me, “The race is to figure out the iPhone-app equivalent of this era.”

In some ways, the A.I. bubble is not unlike that of the dot-com era, but, as a recent piece in Forbes put it, this time “the numbers seem to be 10 times bigger, and the burn is faster.” ChatGPT and its ilk have hundreds of millions of users, and the A.I. industry is already meaningfully affecting the U.S. economy. In the second quarter of this year, G.D.P. grew by three per cent; A.I. capital expenditures may have accounted for nearly half of that, and spending on the construction of data centers is on track to outpace that of building offices. Some kind of market correction is inevitable, as even A.I.’s boosters have acknowledged, and the current round of consolidation could be a sign of contraction. One or two of the crop of A.I. startups will become the new Apple or Google; the rest will likely disappear without a trace. “We’re about to enter a winter of companies shutting down,” Miller said. Regarding the A.I.-browser market, of which Dia is a part, he added, “The winner is going to be decided in the next twelve months.”

Dia seeks to build A.I. interactivity into every aspect of the browser interface, redefining how we move around the web. Right now, A.I. tools such as ChatGPT operate in their own windows, working in isolation from the rest of your onscreen activities. Dia wants to make A.I. chat and generative tools available every time you pull up a site or service, from e-mail to online shopping to media platforms. The artificial intelligence will be able to digest the information you’re looking up and port it from one tab to another. You can, for instance, ask Dia to summarize and send an e-mail about a slew of Amazon links you have open, or just have the A.I. buy products for you, using a planned agentic function that the company calls “self-driving.” Dia remembers what you’ve browsed—it “knows a lot about you,” Miller said—and it will integrate a generative-text function into the cursor, so that you can ask the A.I. questions or get suggestions at any point. This model builds on the success of Arc, a newfangled browser that the company built during the pandemic and released in 2022. (To the chagrin of Arc’s earliest users, its development was halted, in favor of Dia, early this year.)

Mike Cannon-Brookes, the C.E.O. and co-founder of Atlassian, became interested in the acquisition of the Browser Company as a user and a fan of Arc. What intrigued him most was how the browser’s interface made A.I. feel familiar. Traditional browsers have tried to graft A.I. functions onto their existing software, but, as Cannon-Brookes told me, “Like most new technologies, if we just try to apply the old designs to it, it doesn’t really work very well.” No one knows what the future of A.I.-user experience will look like, and the public so far has exhibited a limited appetite for the A.I. tools that companies are shoving into the corners of their apps with abandon. Most of us likely still ignore A.I., just as we ignored the rash of pop-up windows in the nineties; this didn’t stop Google from announcing, last week, a major A.I. update for Google Chrome which heavily resembles Dia. Josh Stutt, the head of marketing at The49, an A.I.-focussed startup incubator, told me, “There is a lot more money floating around than we have answers and use cases.” Hence the over-all feeling of spaghetti being thrown at so many screens. Some startup founders might aim to become the Zuckerberg of a new generation, piloting their own gargantuan tech corporations, Stutt said, but “a lot of them are looking to build something cool, flip it, and call it a day.” In other words, it might be time to sell your picks-and-shovels company before the bubble pops.