The Browser Wars Are Back

OpenAI's Chrome

So OpenAI is making a web browser. Which is a thing that startups do when they have a lot of money and are looking for new ways to spend it. But of course, this isn't really about building a better Chrome. It's about building a home for the AI agents that are supposed to take over the world, and more importantly, it's a direct shot at Google's entire business model. Here's Reuters with the news:

OpenAI is close to releasing an AI-powered web browser that will challenge Alphabet's market-dominating Google Chrome, three people familiar with the matter told Reuters.

The browser is slated to launch in the coming weeks, three of the people said, and aims to use artificial intelligence to fundamentally change how consumers browse the web. It will give OpenAI more direct access to a cornerstone of Google's success: user data.

A web browser would allow OpenAI to directly integrate its AI agent products such as Operator into the browsing experience, enabling the browser to carry out tasks on behalf of the user, the people said.

Yes, a browser is the perfect place to put your little AI minions to work, booking flights and filling out forms. More importantly, it's the perfect place to watch what users are doing. Data is the oil, compute is the engine, and the browser is the oil rig.

And this is all happening because Google is, maybe, a little bit on its back foot. The whole premise of the AI chatbot revolution is that people might not want to sift through ten blue links anymore; they just want an answer. We have seen some evidence of this. Apple executive Eddy Cue testified in May that for the first time ever, search queries on the iPhone's Safari browser had dipped, a change he directly attributed to users turning to AI. Apple itself is now reportedly weighing replacing its own Siri AI with models from Anthropic or OpenAI and is looking to integrate AI search options directly into Safari.

So you can see the play. Google's castle is built on two things: a dominant search engine and a dominant browser (Chrome) that feeds the search engine. OpenAI and others are attacking the search engine with chatbots. Now, OpenAI is attacking the browser, too. It is not alone. AI search startup Perplexity has already launched its own AI-native browser, Comet. This is a battle for the internet's front door.

The ironic twist, as is so often the case in tech, is that OpenAI is building its Google-killer on top of Google's own open-source code, Chromium. It's a bit like building a Pepsi bottling plant using a blueprint that Coca-Cola gave you for free.

And the move makes sense in the context of OpenAI's other grand ambitions. The company just spent $6.5 billion to buy Jony Ive's hardware startup to build a new "family of devices." This isn't just a software play; it's a full-stack play. They don't just want to build a better model; they want to own the entire interface through which you interact with AI, whether it's a physical device in your pocket or the browser on your screen. The fight is no longer about who has the best search results. It's about who gets to mediate your entire digital existence.

The Talent War

There is a fun game being played in Big Tech right now, where the goal is to figure out just how much money an AI researcher is worth. For a while, the number seemed to be in the low eight figures. Then Sam Altman went on a podcast and said Meta was offering his OpenAI employees signing bonuses as high as $100 million. That is, as they say, a lot of money. But apparently, it was not enough. Which brings us to the numbers. From Bloomberg:

Meta Platforms Inc. has made unusually high compensation offers to new members of its "superintelligence" team — including a more than $200 million package for a former Apple Inc. distinguished engineer.

Meta hired Ruoming Pang, who ran Apple's AI models team, with a pay package in the hundreds of millions over a several-year period, according to people with knowledge of the matter… Apple didn't try to match the offer, as it far exceeds pay at the company for leaders other than Chief Executive Officer Tim Cook.

There are a few ways to think about this. One is that Mark Zuckerberg is suffering from a severe case of what Deirdre Bosa of CNBC has called "strategic FOMO." After his Llama 4 models had a less-than-stellar debut, he seems to have decided that the best path forward is to simply buy all the AI talent in the world. He tried to buy Perplexity, he took a $14.3 billion stake in Scale AI to get its CEO, and now he is paying nine figures to poach individual engineers.

Another way to think about this is a classic "acqui-hire," but one where the asset being acquired is not a company but a single human brain. The modern AI startup is a strange corporate financing puzzle. A few brilliant people get together, raise a few billion dollars from investors, spend most of it on Nvidia chips and electricity, and then a bigger company comes along and just hires all the brilliant people, leaving the investors with a pile of rapidly depreciating hardware. It is an interesting business model.

In this case, the talent war seems to be mostly between Meta and OpenAI. Sam Altman insists his people are staying because they believe in OpenAI's mission and culture and want to be on the winning team. Zuckerberg's counterargument appears to be a dump truck full of cash.

But the most interesting part of this story is not who is fighting, but who is not. Apple, a company with more cash than most countries, looked at a $200-million-plus offer for one of its top AI minds and said "nah, you can have him." This is, from a certain point of view, fiscally responsible. From another, it is a disaster. If you can't or won't pay to keep your best people, you will not have the best people. And if you don't have the best people, you cannot build the best technology.

And so you end up in a situation where, as we discussed last week, Apple is now reportedly considering licensing AI models from Anthropic or OpenAI to power Siri. It is a wonderfully ironic feedback loop. The reason Apple has to license outside AI is because its own AI isn't good enough. And one reason its AI isn't good enough is because it won't pay to keep the people who could make it better, who are now going to work for the companies Apple might have to pay for the AI. You never want to be on a board where the clever strategic move is to rent the technology built by the employees you just lost.

The Scoreboard

• AI: Amazon Web Services Is Building Equipment to Cool Nvidia GPUs as AI Boom Accelerates (CNBC)
• Social Media: Linda Yaccarino Steps Down as CEO of Elon Musk’s X (WSJ)
• Consumer: Samsung Bets on Slimmer AI Foldable Phones to Challenge Apple, Chinese Rivals (SCMP)