Train the algorithms artificial intelligence most important ones that are going around today is very expensive. Not only that, the processing power they need is also immense. For this reason, in recent weeks Sam Altman, creator of OpenAI and who is at the forefront of AI, came out to say that there are two things that will greatly impact how we will use artificial intelligence: processing power that we can achieve and its price. The more powerful and cheaper the systems are, the more these algorithms can be used.

That’s where it comes in Microsoft, which is the gold mine of OpenAI. Rumors suggest that the company founded by Bill Gates would put US$ 100,000 millions in the manufacture of a supercomputer that will be installed in the United States to be able train and drive OpenAI’s upcoming artificial intelligence models.

Read also: Artificial intelligence and quantum computing: the explosive combo for cybersecurity

Estimates are that Stargate will take between five and six years to complete and that it will have five manufacturing phases. Microsoft would go for the third. And the price would be more than 100 times more expensive than some of the largest data centers operating today.

Energy consumption and nuclear fusion

Added to all this is another thing that Altman has been talking about for a long time: the energy consumption. The Stargate scale would require immense energy sources and, for that reason, they are exploring whether the solution is nuclear fusion, which is a technique that imitates the reactions that occur within the Sun to produce clean and cheap energy.

Even Altman invested millions of dollars in a venture to develop this type of energy, called Helion Energy. Helion already signed an agreement with Microsoft at the beginning of last year to be able to provide it with electricity generated through nuclear fusion starting in 2028.

This is how data centers grow

Microsoft has been involved in the world of data centers for a long time. To realize the magnitude of the project, we can take what the company has been doing recently.

Last year Microsoft began construction of a new data center US$ 1,000 million in Wisconsin that analysts say is intended to house the chips to train OpenAI’s next-generation models.

Read also: We test an artificial intelligence designed to chat and spy on the soul

This data center could cost up to $10 billion once the price of specialized NVIDIA chips used for artificial intelligence applications is taken into account.. Stargate would be about 10 times more expensive. But betting $100 billion is a lot, even for Microsoft. In fact, it is more than three times what the company spent in 2023 and double what it plans to spend by 2024.

If you compare it to Amazon Web Services, which is Amazon’s data center arm, that’s two-thirds of the total amount they said they plan to spend on all new data centers over the next 15 years.

Why Stargate?

The name of the supercomputer comes from a science fiction film, also called Stargate, published in 1994 where a series of archaeologists discover an interstellar teleportation device that takes them to an alien world.

The movie trailer describes this device as “the key to the past, the door to the future, the step to discovery”. Quite aligned with what Microsoft and OpenAI want to achieve with this supercomputer: the famous artificial general intelligence, which would be an algorithm that would have the ability to learn like humans do and even surpass our intelligence.

Other news that may interest you