OpenAI Folds ChatGPT, Codex, and Atlas Into Desktop Superapp to Counter Anthropic

OpenAI confirmed Thursday that it will merge its ChatGPT desktop app, Codex coding platform, and Atlas web browser into a single desktop application, the Wall Street Journal first reported. The consolidation puts OpenAI President Greg Brockman in temporary charge of the product overhaul, while CEO of Applications Fidji Simo will lead the sales push for the unified app, an OpenAI spokesperson told Reuters. No launch date has been set, and the mobile ChatGPT app will remain separate.

The move marks a reversal from the startup approach that defined OpenAI's past year. CEO Sam Altman had previously likened the company's product expansion to "betting on a series of startups," according to Livemint. That era produced Atlas, Sora, Codex, e-commerce features, and a hardware device. Now Simo is consolidating. Fewer products, fewer teams, one app.

Fragmentation as admission

Simo spelled out the reasoning in an internal memo reviewed by the Journal. "We realised we were spreading our efforts across too many apps and stacks, and that we need to simplify our efforts," she wrote to employees on March 19. "That fragmentation has been slowing us down and making it harder to hit the quality bar we want."

Fragmentation is a polite word for what actually happened. OpenAI shipped Atlas in October, a browser that targeted Google's grip on how people use the web. It launched a standalone Codex desktop app for macOS in February. ChatGPT itself kept evolving with new features on a separate track. Three apps, three engineering teams, three user experiences. For a company that still operates more like a research lab than a shipping organization, that is a lot of exposed surface.

At an all-hands meeting earlier this month, Simo told staff that OpenAI was "orienting aggressively" toward high-productivity use cases, according to a partial transcript reviewed by CNBC. "What really matters for us right now is staying focused and executing extremely well," she said. The language carried the urgency of someone who arrived at a company that had been launching products faster than it could support them.

Simo came from Instacart, where she ran a consumer products operation at scale. She joined OpenAI in May 2025. Ten months later, she is reorganizing the entire desktop product line. That's fast, even by Silicon Valley standards. But Simo framed the consolidation as execution focus ahead of a potential IPO this year. "Really glad we're seizing this moment," she wrote on X.

The Anthropic problem

Product discipline becomes urgent when the competition stops being abstract. Anthropic has emerged as OpenAI's most pressing rival, particularly in developer tools and enterprise AI. The company hit a $5 billion annualized revenue run-rate last September with 300,000 enterprise customers, and its Claude Code product has been gaining traction with the developer audience OpenAI once owned outright, according to multiple industry reports.

One source described OpenAI as going into "code red" over the Claude Code threat, The Information reported. You could feel the anxiety in the product org, and it shows in the superapp decision. By folding Codex into the same shell as ChatGPT and Atlas, OpenAI can create tighter integration between coding, browsing, and conversational AI. If you're researching documentation in Atlas and want to hand that context directly to Codex without switching windows, a unified app makes that possible. Three separate products never could.

Simo addressed the competitive pressure on X after the Journal's report broke. "Companies go through phases of exploration and phases of refocus; both are critical," she wrote. "But when new bets start to work, like we're seeing now with Codex, it's very important to double down on them and avoid distractions."

The framing is deliberate. Concentration, not retreat. But the timing matters. This announcement arrived weeks after CNBC reported that OpenAI was preparing for a potential IPO this year, and days after the company confirmed its acquisition of Astral, a Python tooling startup. The audience for this move isn't just developers. It's investors who want to see a company that knows what it is building.

Microsoft's uncomfortable seat

OpenAI's push to own the desktop workspace creates a tension neither company seems eager to discuss publicly. Microsoft has invested more than $13 billion in OpenAI and holds exclusive rights to commercialize its technology in certain categories, Bloomberg reported. A unified OpenAI desktop app combining chat, coding, and web browsing competes directly with Microsoft 365 and its Copilot integrations across Windows, Edge, and Office.

Both companies maintain the partnership remains strong. Sure. But walk into any Microsoft product meeting and OpenAI's models are already embedded throughout the suite. An OpenAI superapp that pulls users into its own workspace, away from Microsoft's surfaces, strains a relationship that was already showing cracks. The Financial Times reported this month that Microsoft was weighing legal action over a $50 billion Amazon-OpenAI cloud deal.

For now, the standalone mobile ChatGPT app stays separate from the superapp, according to the Journal. That limits the immediate collision with Microsoft's consumer AI products. Desktop, though, is where enterprise users spend their hours. And enterprise is exactly where Simo is pointing this thing.

Agentic AI and the Astral bet

The superapp won't just bundle three existing products behind one icon. OpenAI intends to build it around agentic AI capabilities, allowing autonomous agents to perform tasks on a user's computer, writing software, analyzing data, executing multi-step workflows, with minimal human oversight, according to company statements.

That ambition got a boost the same day the superapp news dropped. OpenAI confirmed its acquisition of Astral, the Python tooling company behind the uv package manager and Ruff linter that developers already have open in their terminals every day. Financial terms were not disclosed. Astral's tools are already standard equipment for most Python developers and AI researchers, which gives OpenAI a deeper foothold in the development stack its users already inhabit.

And this is where the strategy coheres. Codex writes the code while Atlas handles the browsing and ChatGPT manages the conversation layer. Agentic capabilities let all three coordinate without constant human direction, and Astral's tooling keeps the Python stack tightly wired in. One window for all of it.

The risk is familiar. OpenAI is describing a product that does not exist yet, built on integration patterns that have tripped up larger companies with more shipping experience. Google has kept Gemini spread across Chrome, Workspace, and Android without cramming everything into a single surface. Anthropic has stayed focused on Claude as a conversational and coding tool without constructing a desktop shell around it. Microsoft has Copilot everywhere but has not consolidated it into a standalone workspace.

What discipline looks like from the outside

OpenAI chose the word "refocus." Competitors will read it differently. When a company that launched a browser, a video generator, a hardware device, and e-commerce features in a single year suddenly announces that fragmentation was slowing it down, the natural question is whether the expansion was strategy or impulse. Altman's "series of startups" line sounds less charming in retrospect.

Simo has the organizational authority, Brockman's backing, and a clear mandate from the board. What she does not have is a launch date or a track record at OpenAI longer than ten months. The enterprise admin console that Fortune 500 buyers expect does not appear to exist yet either. The last company to successfully build a desktop superapp that people actually wanted to use was, arguably, Microsoft. With Office. Thirty years ago.

Somewhere in San Francisco, engineers are already pulling Atlas code into the ChatGPT repository. Codex integration branches are being opened. Slack channels are being merged and deprecated. The hard part of consolidation is not the announcement. It's the six months after, when three products with three engineering cultures and three user bases have to become one thing that works better than what came before. Simo's memo acknowledged the problem. The superapp has to be the answer.

Meta Won Its Antitrust Case While Losing the Talent War That Actually Matters

Federal judges declared Meta faces "fierce competition" on Tuesday. The company's AI researchers apparently agree. They're just choosing to work for it instead. Judge James Boasberg dismissed the FTC

The Implicator

OpenAI's Atlas Browser Puts Google's Search Empire on Defense

The AI company's new product marks its most direct challenge yet to Alphabet's core business—and investors noticed Sam Altman unveiled OpenAI's Atlas browser Tuesday morning. The timing wasn't random

The Implicator

Atlassian buys Arc/Dia maker to build the “AI browser for work”

💡 TL;DR - The 30 Seconds Version 👉 Atlassian buys The Browser Company for $610 million cash to transform Dia into an AI browser designed specifically for enterprise knowledge work. 📊 Only 10%

The Implicator

Marcus Schuler

San Francisco

Editor-in-Chief and founder of Implicator.ai. Former ARD correspondent and senior broadcast journalist with 10+ years covering tech. Writes daily briefings on policy and market developments. Based in San Francisco. E-mail: [email protected]