Sam Altman needs 100 million AI GPUs worth $3 trillion for his vision — after OpenAI was forced to do “unnatural things”

by

Do you recall the incident when OpenAI CEO Sam Altman was referred to as a “podcasting bro” by TSMC executives, due to his bold and visionary AI plan that aimed to construct 36 semiconductor plants and additional data centers at an estimated cost of $7 trillion and over several years? This ambitious project was intended to bring his dream to fruition.

It’s increasingly clear that advanced AI systems, like those capable of generating content, tend to require significant resources such as vast amounts of computational power and cooling water, as they grow more complex and achieve greater milestones.

It’s been said that OpenAI has expressed concerns about Microsoft not being able to fulfill its cloud computing requirements, as stipulated in their multi-billion-dollar deal. The creators of ChatGPT hinted that if another AI company manages to develop Artificial General Intelligence (AGI) before Microsoft, the blame would likely lie with them.

This year, OpenAI revealed its ambitious Stargate plan, a $500 billion initiative aimed at building AI-focused data centers nationwide in the U.S., which led to Microsoft losing its position as their cloud provider.

It’s worth noting that according to a recent report by The Wall Street Journal, the massive project in question seems to be facing challenges taking flight. Instead, it appears that OpenAI might focus on establishing a smaller data center before the end of this year.

As a tech enthusiast, I’m thrilled to share that Sam Altman just hinted at OpenAI ramping up to have over a million GPUs active by the year’s end! This exciting update follows his disclosure about the AI pioneer needing to pull off some rather unusual maneuvers due to a GPU shortage triggered by the viral Studio Ghibli memes from their GPT-4o image generator. Unusual steps included borrowing computational power from research projects and temporarily dialing back certain features.

By the end of this year, we’ll have more than a million graphics processing units (GPUs) up and running! I’m incredibly proud of the team, but now it’s time for them to buckle down and brainstorm ways to increase that number by 100 times. Ha! Let’s see if they can make it happen! (July 20, 2025)

During roughly an hour, ChatGPT attracted more than a million new users, which was so significant that the company’s executives humorously suggested it came close to melting their Graphics Processing Units (GPUs) due to the overwhelming demand.

I observed Sam Altman expressing great pride in his team but emphasizing the crucial challenge of scaling OpenAI’s current 1 million GPUs to an ambitious target of 100 million. Tom’s Hardware speculates that this expansion could potentially cost the AI company up to $3 trillion. For a point of comparison, this figure falls just short of Microsoft’s recent market capitalization, which surpassed the $4 trillion milestone.

The constant development of AI underscores its growing need for increased computational resources to support complex advancements, a viewpoint shared by AI safety researcher and director of the Cyber Security Laboratory at the University of Louisville, Roman Yampolskiy. According to him, if sufficient computing power and financial resources were widely accessible, Artificial General Intelligence (AGI) could potentially be realized today by an AI company.

It’s intriguing to consider whether OpenAI could manage to amass over $3 trillion to purchase additional GPUs, boosting their AI capabilities. The question then arises: Would this substantial amount be sufficient for them to accomplish extraordinary milestones such as Artificial General Intelligence (AGI) or even superintelligence?

Read More

2025-07-22 13:09