At a Mediterranean restaurant in San Francisco overlooking Alcatraz Island, Sam Altman set the stage for what felt less like a casual dinner and more like a strategic briefing. The OpenAI CEO had gathered about a dozen tech reporters for an on-the-record meal (and, curiously, an off-the-record dessert).
From the moment Altman walked in — bare iPhone in hand, joking about cases being unnecessary — the evening carried an unusual mix of casual charm and high-stakes messaging. The night wasn’t really about fish entrées or lamb skewers. It was about signaling where OpenAI goes from here, especially in the wake of a rocky reception to GPT-5.
GPT-5’s Mixed Reception
Launched with years of anticipation, GPT-5 was expected to surpass its predecessors in a leap forward. Instead, it has been described as a solid step — one that keeps pace with competitors like Google’s Gemini and Anthropic’s Claude, but doesn’t blow past them as GPT-4 once did in 2023.
The rollout stumbled when OpenAI quietly deprecated GPT-4o without warning, frustrating users. Altman himself admitted:
“I legitimately just thought we screwed that up.”
VP of ChatGPT Nick Turley added that OpenAI is now learning from user reactions. The company is rolling out updates to make GPT-5’s responses feel “warmer” without tipping into sycophancy. “Many people liked that ChatGPT would actually check in with you,” Turley said, contrasting this with GPT-5’s initial robotic directness.
Behind the lighthearted dinner table talk was a serious acknowledgment: user trust matters, and OpenAI knows it cannot afford to alienate its massive base.
Shifting the Spotlight Beyond Models
Yet it became increasingly clear that the dinner wasn’t really about GPT-5. Altman and his team wanted to steer the conversation elsewhere — toward OpenAI’s expanding ambitions that stretch well beyond large language models.
Consumer Apps on the Horizon
OpenAI’s incoming CEO of applications, Fidji Simo, will soon lead efforts to launch entirely new consumer products. Rumors suggest this could include:
- An AI-powered browser designed to compete with Chrome.
- An AI-driven social media app, aimed at reimagining digital interaction with more meaningful experiences.
Altman even entertained the idea of buying Chrome outright if it were ever up for sale — a bold, if unlikely, move that underlines OpenAI’s appetite for scale.
A Beautiful Device with Jony Ive
Altman also spoke of the upcoming hardware project being developed with legendary Apple designer Jony Ive. With a playful warning that putting a case on it would be a crime against design, Altman positioned the device as something too beautiful to cover — a piece of hardware meant to redefine how we interact with AI in daily life.
Brain Chips and Neural Interfaces
The evening also confirmed OpenAI’s interest in Merge Labs, a brain-computer interface startup that would compete with Elon Musk’s Neuralink. Though no deal has been signed, Altman described it as a company he’d “like us to” invest in. Whether such an investment will tie deeply into OpenAI’s models or remain independent remains unclear, but it highlights OpenAI’s growing interest in the frontier of human-computer integration.
Demand is Surging Despite Criticism
For all the questions around GPT-5’s reception, one fact stands out: usage is skyrocketing.
Altman revealed that within just 48 hours of launch, API traffic doubled, straining OpenAI’s GPU resources to the point where the company is “effectively out of GPUs.” Tools like Cursor and other coding assistants have already made GPT-5 their default model.
It’s a striking contradiction: critics call GPT-5 underwhelming, but adoption is breaking records.
Balancing Human Impact
A quieter but crucial thread in the dinner conversation centered on how people — especially vulnerable users — relate to ChatGPT.
Altman estimated that less than 1% of ChatGPT users have “unhealthy relationships” with the chatbot. That still translates to tens of millions of people. To address this, OpenAI has worked with mental health experts to evaluate GPT-5’s answers and ensure the model pushes back against reinforcing negative behaviors.
It’s a reminder that as OpenAI scales, the human side of AI use cannot be ignored.
The Bigger Picture: OpenAI as the Next Tech Giant
As the evening wound down, it became increasingly clear that OpenAI wants to shed the identity of being “just the ChatGPT company.”
The company is already making bets in:
- Search (challenging Google’s dominance)
- Enterprise software (offering AI-powered productivity tools)
- Robotics and energy (laying groundwork for the infrastructure powering AI’s future)
Altman’s vision is bold: to build a company that could rival Alphabet, not just in scope but in diversity of influence.
Preparing for a Public Future
All of these ambitions require one thing: massive amounts of capital. Altman dropped hints that going public could be part of OpenAI’s roadmap to fund data centers, hardware, and frontier AI research.
For now, the dinner felt like a pitch — not just to the media, but to the world. A pitch that OpenAI’s most important chapter hasn’t been written yet.
The evening reflected OpenAI’s current paradox: a flagship product facing criticism, yet record adoption; a company still building models, but dreaming far beyond them.
Altman isn’t just selling GPT-5. He’s selling a future where OpenAI becomes the defining company of the next technological era — spanning apps, hardware, brain interfaces, and beyond.
The question is whether the world is ready to follow him there.
Sources ( Techcrunch )
