By Ben Cohen
The latest battle in the most expensive competition in the history of capitalism officially began a few years ago with a message on Slack.
"I wanted to show off a new tool I've been hacking on," an engineer named Boris Cherny wrote in 2024 to his colleagues at Anthropic.
Before long, everyone inside the company was using his new tool. These days, it's known by millions of people as Claude Code. It has conquered Silicon Valley, jolted Wall Street and even managed to scramble Anthropic's fiercest rival.
Along the way, it also revealed something elemental about business success.
Long before Cherny pressed send on that fateful message, Anthropic's founders had chosen a counterintuitive strategy that would come to define the company. They decided to focus ruthlessly on coders and enterprise business customers, rather than everyday consumers. Because of their decision, Anthropic's engineers have an unusually clear sense of the company's identity. And when Cherny was tasked with building a coding tool for users, he knew exactly who those users were. The result was Claude Code.
That focus turned a side project within the company into the company's hit product.
In any business, focus is an edge. But it's especially valuable in the business where the possibilities are infinite and the stakes feel existential.
And when companies can build almost anything, the trickiest thing is knowing what not to build.
OpenAI is learning that lesson the hard way right now. As my colleague Berber Jin reported this week, OpenAI's executives are refocusing the company and racing out major strategic changes to meet the demand for business productivity.
It's a notable shift from the industry giant that practically invented the consumer market when it unleashed ChatGPT, which is still the most popular AI product ever made. But more people and companies are spending their time and money with Claude and Google's Gemini. And with AI evolving -- from humans chatting with bots to bots working for their humans -- OpenAI had to adapt.
And with AI evolving -- from humans chatting with bots to bots working for their humans -- OpenAI had to adapt.
"We cannot miss this moment," Fidji Simo, the company's CEO of applications, told staff, "because we are distracted by side quests."
As it happens, the company that wrote the playbook on resisting distractions is just down the road.
"The key for us is focus," Apple CEO Tim Cook once told me, "which is more about saying no than it is about saying yes."
That principle has been essential to Apple for the past 50 years. In the early days, the company even published a philosophical document that singled out "focus" as a top priority.
"In order to do a good job of those things that we decide to do," wrote Mike Markkula, Apple's first investor and chairman, "we must eliminate all of the unimportant opportunities."
A half-century later, Cook put it this way: "You're saying no to really, really good ideas so you can make room for the great ones."
Like the other insanely great ideas from Apple, this one came from the mind of Steve Jobs. He may have tantalized the crowds with promises of one more thing, but he always believed in the power of fewer things.
That sort of focus doesn't come naturally to all of today's entrepreneurs who grew up idolizing Jobs. To start their companies, founders have to be creative, restless, slightly delusional and outrageously ambitious. Almost inevitably, they get bored of what they have built and feel the itch to build again. The same drive that pushed them to do one thing pulls them to the next thing.
But when Jobs sold NeXT and came back to Apple, he saved the company by simplifying the product line. He narrowed the organizational focus to fit neatly inside a 2×2 grid -- consumer and pro, desktop and laptop. And his monastic clarity led to magical products. By not making a PDA, Apple had the resources to make the iPod, then the iPhone.
In other words, he eliminated an unimportant opportunity to pursue the right ones.
"I'm actually as proud of many of the things we haven't done," Jobs said in 2008, "as the things we have done."
OpenAI has spent the past year trying to do everything. A video generator and web browser. ChatGPT ads. Secret hardware devices! Most promising of all, the company released an impressive AI coding platform called Codex, which is now part of its planned "superapp." CEO Sam Altman has described this scattershot approach as "betting on a series of startups" inside his own company.
But when it's time to focus on the winning bets, founder types often bring in skilled operators. Altman hired Simo, the former Instacart CEO and head of Facebook's app, to do precisely what she's doing right now.
"When new bets start to work," Simo wrote on X this week, "it's very important to double down on them."
At Anthropic, CEO Dario Amodei has taken a radically different approach and placed his chips on fewer bets. No browsers. No video slop. No hardware devices. When the company bought a Super Bowl ad for Claude, it was to brag about keeping Claude ad-free.
For most of the AI boom, it didn't seem to matter if the leading frontier labs were disciplined or distracted. OpenAI was so far ahead of the pack that it was hard to imagine any rivals catching up. But over the past year, the race tightened. And not long ago, OpenAI suddenly found itself under such intense pressure from a resurgent Google that Altman declared a "code red."
Anthropic, meanwhile, kept focusing on another kind of code.
Claude Code was released to the public last year, months after it flickered to life on Cherny's screen. Anthropic relied on business customers for 80% of the company's revenue -- and they couldn't get enough of the product. Then the best AI coding tool got much better.
By the end of the year, as OpenAI was fending off Google, Anthropic unveiled a model so powerful that it made people who don't know code into coders and made actual coders feel like they were strapped to a jetpack. It turned the lucrative talent of coding into the act of telling Claude what to code.
And it showed why the smartest bet in AI was saying no to almost everything -- and yes to the guy hacking away on a new tool.
Write to Ben Cohen at ben.cohen@wsj.com
(END) Dow Jones Newswires
March 20, 2026 21:00 ET (01:00 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
