🔺 “Code Red” is how American hospitals would alert their staff about fire.
Sam Altman has signaled “Code Red” to his team a week ago. Last Monday The Information wrote about a “leaked” memo.
🔥What is burning? Most likely, it is investors’ cash burning while OpenAI multiplies its costs, hiring more and more people, renting more and more data centers and opening satellite offices.
There is no secret that OpenAI has zero product discipline. I’m happy to remind you a long list of products and features offered by company.
Custom GPTs and "GPT Store" - this was OpenAI's attempt to create marketplace where users could build and monetize their own AI assistants. The store launched with fanfare but never gained real traction, most custom GPTs have handful of users at best.
Sora feed turned their video generation model into full social network where people share AI-generated clips. This pivot from tool to platform happened without clear business model or user demand.
Atlas is OpenAI's web browser project, attempting to rebuild browsing experience around AI assistance from ground up. This puts them in direct competition with Google Chrome and Microsoft Edge, both backed by companies with decades of browser development experience.
Pulse was designed as personal information feed that would curate and summarize content across user's digital life. Project got shelved in recent memo, indicating it never found product-market fit despite significant investment.
Operator is autonomous agent that handles web tasks, booking appointments, filling forms, making purchases. This overlaps significantly with features already built into ChatGPT and raises questions about why it needs to be separate product.
Codex powers command line coding tool and underlies GitHub Copilot. Unlike other products on this list, Codex actually generates revenue through Microsoft partnership, but OpenAI seems uncertain whether to compete with their own licensee or support them.
Aardvark is agentic security researcher that autonomously searches for vulnerabilities in code. This is highly specialized tool aimed at narrow market, strange priority for company claiming to focus on consumer experience.
And finally there is mysterious hardware device being developed with Jonny Ive, Apple's former design chief. Nobody outside OpenAI knows what this device does or when it will ship, but it represents massive investment in unproven hardware category.
Altman’s memo wasn’t addressing the issue of lack of focus in company’s strategy, it was addressed to OpenAI employees who shall concentrate on priority projects.
But I know that OpenAI leadership is worried about a lack of focus that 8000-people “startup” demonstrated in the last year.
How do I know? From LinkedIn profile of Kevin Weil.
Kevin - ex-Twitter, ex-Instagram, ex-Meta, joined OpenAI in a summer of 2024 - was company’s Chief Product Officer.
In October 2025, his title changed to “VP, OpenAI for Science”. We should read this as definitive sign that Mr. Weil is removed from steering all OpenAI products besides “OpenAI for Science”, whatever it is.
In his memo, Altman (allegedly) wrote, “We are at a critical time for ChatGPT” and explained that the company will push back its work on advertising integration, AI agents for health and shopping, and a personal assistant feature called Pulse. Furthermore, Sam Altman encouraged temporary team transfers and established daily calls for employees responsible for enhancing the chatbot.
100% of my business experience screams “b/s” hearing how Altman focused his message at his employees. When a leader of a company decides to raise headcount to more than 8000 people, and then calls everyone to focus on company’s singular product - this is deranged. A focused AI lab needs 2-3 times less people, and I won’t be surprised to see staff reductions at OpenAI.
Still the elephant in the room is an issue of “consumer focused AI” - this seems to be investor-driven strategy that OpenAI tries to implement. Few months ago, Brad Gerstner [GP in Altimeter Capital, invested more than $250 million in OpenAI] was repeatedly saying that OpenAI is a consumer-facing business.
Altman’s alleged messaging to his team have included three points, all of them “customer-focused: make ChatGPT more personal, make ChatGPT less restrictive, make ChatGPT faster.
In my humble opinion, there is no such thing as “consumer AI”. The fact that Anthropic’s LLMs (Haiku, Sone, Opus) are considered one of the industry’s best and are top choice for people making money with AI - mostly developers - shows that if you grew a good model, it will be great “employee” and “companion”.
Every attempt to make ChatGPT most attractive to consumers only led to unwanted results - (a) model becoming sycophantic, and this becomes really dangerous as soon as a user has a dangerous idea and starts to discuss it with LLM (sycophantic LLM can lead a person with suicidal thoughts to untimely end) or (b) model becoming too “paternal”, restricting discussion topics, refusing to provide information, avoiding giving precise numbers, etc.
AI lab trying to win a game of “consumer AI” - something that OpenAI trying to do for very long time - does not seem realistic in a current competitive environment.
Consumers are won (a) at scale; (b) by price.
OpenAI is a pioneer in a “consumer AI” field, it is a hottest AI company in a world with great team. Their results so far: 800 million weekly users, ~2 billion daily queries, expected 2025 revenue $13 billion.
Those are beautiful numbers but you must place them into the context. Google’s main “surface” - web search - has 5.1 billion users, handles approximately 96 billion search queries weekly. All this leads to revenue of $350 billion. Meta (Facebook, Instagram, WhatsApp) has 7 billion users and $165 billion of revenue.