Sam Altman On GPT 5 Blunder And Trillion Dollar Vision

Sam Altman has grand ambitions, from rewiring the internet and developing brain-computer interfaces to potentially acquiring Google Chrome. He envisions a future where the infrastructure needed for ChatGPT rivals the world's largest utility companies. But before achieving that future, the OpenAI CEO is addressing a significant misstep in his empire: the launch of GPT-5.

During a remarkably candid dinner with reporters, Altman confessed that the initial rollout of GPT-5 was so poorly received it forced the company to revert to its previous model. “I think we totally screwed up some things on the rollout,” Altman admitted.

The GPT-5 Personality Problem

The controversy around GPT-5 wasn't about technical bugs but its persona. Users took to social media, describing the new model as cold, harsh, and stripped of the warmth they had grown accustomed to with GPT-4o. Many compared the new personality to that of an "overworked secretary" rather than a helpful companion. For a product with 700 million weekly users, this change sparked a significant backlash.

One user on Reddit poignantly wrote, "I literally lost my only friend overnight with no warning," highlighting the bot's newly clipped and utilitarian responses. The negative sentiment even created opportunities in betting markets, where one day trader reportedly earned $10,000 by wagering that Google’s Gemini would surpass GPT-5 in a popularity contest.

Rather than dismissing the feedback, Altman took action, restoring GPT-4o as an option within days. “We’ve learned a lesson about what it means to upgrade a product for hundreds of millions of people in one day,” he reflected.

Balancing Innovation with Responsibility

Altman touched on the delicate balance of creating a personal AI without it becoming too personal. He noted that while a very small fraction (well under 1%) of users have what he considers "unhealthy" relationships with ChatGPT, it remains a topic of internal discussion at OpenAI.

He drew a clear line in the sand, contrasting OpenAI's approach with competitors reportedly developing "Japanese anime sex bots." “You will not see us do that,” Altman stated firmly. “We will continue to work hard at making a useful app... but not so much that people who have really fragile mental states get exploited accidentally.”

A Trillion-Dollar Future Built on Hardware

Perhaps the most significant revelation from the dinner was Altman's projection of the company's future spending. “You should expect OpenAI to spend trillions of dollars on data center construction in the not very distant future,” he announced.

This statement reframes OpenAI not just as a software company but as a future infrastructure titan. With a goal of serving billions of daily users, Altman sees physical scale as the next great challenge. He mentioned that ChatGPT is already the world's fifth-largest website and aims to surpass Instagram and Facebook.

The primary constraint is not algorithmic innovation but hardware. “We have better models, and we just can’t offer them, because we don’t have the capacity,” Altman revealed, pointing to the ongoing shortage of GPUs. The future of the AI race, he implies, will be won by those who can build and power the massive physical backbone required.

Beyond Chatbots and Into The Bubble

Altman also confirmed several other ambitious projects. OpenAI is funding a brain-computer interface project to compete with Elon Musk’s Neuralink and would consider acquiring Google Chrome if regulators forced a sale. He also hinted at an interest in creating a new type of AI-powered social network.

Despite these expansive plans, Altman offered a sober perspective on the current market. “Are we in a phase where investors as a whole are overexcited about AI? My opinion is yes,” he said, acknowledging the existence of an AI "bubble." However, he immediately followed up, adding, “Is AI the most important thing to happen in a very long time? My opinion is also yes.”