OpenAI boss suggests states build their own infrastructure for artificial intelligence

The day after a controversy launched by his financial director, Sam Altman assured Thursday that he was not seeking “government guarantees for OpenAI data centers”.

OpenAI is going backwards. His boss, Sam Altman, called on Thursday, November 6, governments around the world to build their “own infrastructure” for artificial intelligence (AI) and assured that he himself was not looking for “public guarantees” for the investments of hundreds of billions of dollars in his company, creator of ChatGPT. In a long message posted onhe thus tried to put an end to the controversy the day after his financial director’s comments.

Wednesday, during a conference organized in California by the Wall Street JournalSarah Friar, had used the terms of “safety net” and of “guarantee” to describe the role she saw the federal state playing in financing OpenAI. After a first wave of criticism, she retracted in a published message on LinkedInsaying wanting to want “clarifier” that the company does not “was not seeking a government guarantee” for its loans.

But the controversy has not died down, as concerns grow about the ability of tech giants to absorb the colossal costs of the race for AI. “There will be no federal bailout for AI. The United States has at least five large pioneering companies in this field. If one of them goes bankrupt, others will take its place”had reacted on David Sacks, Donald Trump’s advisor on AI. “Governments should not pick winners and losers, and taxpayers should not bail out companies that make bad business decisions”added Sam Altman a few hours later.

OpenAI has concluded a series of agreements in recent months that commit it to acquiring, over eight years, $1.4 trillion in additional computing capacity from chip manufacturers and “cloud” (remote computing) players. This total amount concerns a growing number of analysts and market observers, who compare it to OpenAI’s annual revenues, so far announced at $13 billion for 2025.

The creator of ChatGPT, however, loses billions of dollars each year and does not plan to be profitable before 2029, which means that in addition to capital increases, OpenAI will have to borrow massively on the markets, directly or indirectly. “The risk for OpenAI of not having enough computing power is greater and more likely than the risk of having too much”declared the 40-year-old entrepreneur again, saying he wanted “a world where AI is abundant and cheap”.

Sam Altman mentioned, on this subject, the possibility for OpenAI to “rent computing capacities to other companies (and people)“a project that he had never mentioned before, his company not being able, until now, to meet its own needs in this area. If the idea came to fruition, the start-up would therefore pose itself as a competitor to cloud providers, from Amazon to Oracle, including its preferred partner and key shareholder, Microsoft.