TL;DR: With Anthropic gaining ground, OpenAI wants to sharpen its focus and stop getting “distracted by side quests,” the Wall Street Journal reported this week. The new main mission: business customers and coding. It’s a notable switch-up for a company that turned ChatGPT into a byword for AI among the masses—and a sign of where the real money in the AI race is.

What happened: OpenAI applications chief Fidji Simo reportedly told employees at an all-hands meeting that a big strategic shift is in the works. The company’s tendency to “do everything all at once” has allowed Anthropic to catch up with a laser focus on enterprise use and coding. OpenAI wants to regain its lead in those two areas, where businesses are willing to shell out big dollars. OpenAI execs are apparently still deciding which projects to deprioritize in order to free up the needed resources.

Flavor of the month: OpenAI has taken on new projects at a dizzying pace over the past couple of years. One week it seems bent on disrupting Hollywood or TikTok, and the next it’s doubling down on shopping or healthcare.

Here’s a non-exhaustive list of some of OpenAI’s many “side quests”:

• A social feed for generated videos with the Sora 2 app
• A web browser called Atlas
• Shopping tools for ChatGPT, recently scaled back
• ChatGPT Health
• Humanoid robots?
• Consumer devices designed by Jony Ive
• An X-rated “adult mode” for ChatGPT

Actually, that last one is very much still on, despite technical setbacks and internal controversy, according to a separate Wall Street Journal article published Sunday. Sam Altman also seemed to confirm on X that OpenAI is not abandoning its hardware ambitions.

Codes red: The previous strategy was the sort of shotgun approach that might make sense for a first-mover with an early lead. But Anthropic’s narrower tack and Google’s Gemini improvements have now eroded that head start.

In December, Altman reportedly declared a “code red” as the latest Gemini release topped leaderboards, sidelining other projects in favor of core ChatGPT improvements (though new project launches seem to have continued apace in the time since). Simo reiterated a “code red” status in the recent all-hands, according to the Journal.

What’s the fuss?: This battleground centers on coding and enterprise for the same reason that Willie Sutton once famously said he robbed banks: because that’s where the money is. “Enterprise is where OpenAI can land larger, and predictable revenue and coding is a domain with clearer ROI,” Gartner analyst Arun Chandrasekaran tells Tech Brew.

OpenAI recently introduced paid ads into the free tier of ChatGPT, but this part of the business is still very nascent and no match yet for Google’s longstanding expertise.

“Consumer AI is commoditizing fast,” says Mitch Ashley, vice president and practice lead for software lifecycle engineering at the Futurum Group. “Enterprise is where durable revenue concentrates, and coding is where AI delivers measurable, defensible value to buyers.”

Back to business: While OpenAI is now known for its consumer product, the company has a long history of catering to enterprises and coders, too. OpenAI’s first commercial product in 2020 was an API geared toward developers and businesses, and it launched its first generative coding tool, Codex, in 2021. But everything changed after ChatGPT’s unexpectedly massive success in late 2022, when the direct consumer potential of generative AI became clear.

“OpenAI’s recent cadence around Codex and enterprise rollout suggests it wants to defend the developer mindshare it built early and convert it into broader business opportunities,” Chandrasekaran said.

What does this mean for you? “For consumers, it signals that ChatGPT's consumer features will increasingly serve as a funnel to enterprise, not a standalone destination,” Ashley said.

Bottom line: Fierce competition in the AI race is making OpenAI think twice about all its balls in the air. That could mean more focus on improving the core ChatGPT experience but also more prioritization of enterprise customers. There’s also a chance that less core OpenAI products, like Sora, could be neglected. —PK

A (YouTube) TV time machine

If you're nostalgic for the days of dozing off with the TV on only to get jolted awake by a televangelist screaming at you, here's your fix: Channel Surfer is a free web app that uses embedded YouTube videos to simulate the experience of flipping through cable TV channels, static and all.

The app's UI is a ’90s-style TV guide of everything that's "on" right now, across about 40 topic channels such as "AI & ML," "Comedy," "Gaming," and "Cooking," with the videos handpicked by developer Steven Irby. It's synced to a schedule, so everyone on the same channel sees the same video. You join midstream, just like real TV. You can even import your own YouTube channel subscriptions, though they don’t get neatly organized by category; they each become their own channel in the TV guide.

When I first launched the app, it loaded a "2 hour lofi hip hop mix" video—complete with TV scan lines. Switch to another channel, and you get a brief thwump—that muffled sound old CRTs used to make when turned on or off. I also caught a full-length movie from 1925, a physicist explaining the "illusion of time," an SNL skit, and System of a Down's “Chop Suey!” music video. The offerings here appear slop-free: It’s a lot of education channels, video essays, retro movies, and music.

The Good: It turns out one guy's taste is a more pleasant way to discover things on YouTube than an algorithm trained on everything you've ever watched at 2am. You can't pause it, only mute it. It’s a refreshing respite from the wild west of video content.

The Bad: It's just a web app right now, and it would be nice to have a dedicated app you could load on an actual TV (though you could probably just screencast it). While importing channels is fairly simple on desktop, there's no easy way (that I know of) to do that on mobile. There are still YouTube ads, but they’re unlikely to be as amusing as TV infomercials of old.

Verdict: Signal. Take a break from The Algorithm™. It’ll still be there if you want, trying to lure you back into its black hole. —WK

Drive anywhere in the US, and you'll see billboards for personal injury law firms, the Powerball jackpot, and a Cracker Barrel just off the next exit. Enter San Francisco, though, and it's a foreign country where a different language is spoken—that of tech bro. Nearly all the billboards in the city are now for AI companies, and what they actually sell is often inscrutable. A sampling of slogans: “Too much B2B SAAS,” “Own your nines,” “Close week hours,” “Book your Hotel on the Moon.” (The last one's easy to parse linguistically, but still no less head-scratching.)

The golden rule of advertising is to know your audience, and these companies certainly do. Most of these towering monuments to commerce are bought by venture-backed AI startups trying to reach a hyperspecific slice of Bay Area commuters: founders, engineers, and the people who write their checks. Outdoor ad revenue in SF grew 30% between 2023 and 2025, according to Outfront Media. A single digital billboard near SFO cost $7,000 a week for an eight-second spot (on rotation every 64 seconds) as of February (which, to be fair, is probably the cost of one software engineer's lunch stipend).

This deliberately exclusionary “if-you-know-you-know” approach has even spawned its own meta-industry: Vibe, a New York ad tech firm, has recently put up billboards with winking references to Jensen Huang (a hint of his leather jacket), Marc Andreessen (a glimpse of his bald head), and Tim Cook (an apple)—pitching startups to buy ads targeting major tech firms (for what purpose, only they know). Call it adception. When hundreds of AI startups are flush with VC cash, even one of the least technologically sophisticated advertising forms becomes a land grab. —WK

Chaos Brewing Meter: /5