Expensive and complex chips used by the tens of thousands in programs like ChatGPT have become a major issue for the tech giants racing in artificial intelligence.
Expensive and complex, chips used in the tens of thousands in programs like ChatGBT have become a major issue for the tech giants racing in artificial intelligence from Silicon Valley to Taiwan.
The startup OpenAI that built ChatGBT has been at the forefront with its main investor, the Microsoft Group, since the generative AI program was so successful that it attracted millions of users due to its ability to Production of all kinds of texts on request in the spoken language.
However, “the only entity that achieves financial profits is the company + NVIDIA +,” noted Alan Priestley, an analyst at the technology research company “Gartner”. Although “Microsoft” has integrated generative artificial intelligence into a number of its online services, “it has not yet figured out how to generate additional revenue from it,” according to the expert.
The vast majority of semiconductors used in training artificial intelligence models in the world are produced by two companies: Californian “Nvidia” for design and “TSMC” for manufacturing in Taiwan.
NVIDIA is famous for its graphics processors (GPU), which allows especially the use of high-definition video games or video conferencing.
However, these chips, which are capable of processing large amounts of data, have recently enabled the emergence of the latest generation of artificial intelligence, which all technology companies are seeking to adopt, while some governments view it as a matter of national security.
“urgent”
“The capabilities and versatility of generative AI make companies feel it is urgent to develop and implement a strategy in this area,” Jensen Huang, CEO of NVIDIA Corporation, noted during a meeting with analysts in February.
As for independent analyst Jack Gold, he saw a “great appetite” for these chipsets, adding, “I don’t know if Nvidia is really able to provide them,” she confirmed.
Not all institutions have the financial capabilities to obtain these technologies. The expert explained that “the cost of powerful machines needed to train generative AI programs can run into the hundreds of thousands of dollars,” not to mention that they consume a great deal of energy.
There was a delay in the production of “Intel”, which is the first American company to design and manufacture semiconductors.
“Intel has rested on its laurels for a long time,” independent analyst Jack Gold told AFP. He added that it “is trying to produce chips that enable it to compete with + Nvidia + (…) but this field is new for it, as it has not manufactured graphic processing units for at least ten years.”
The California-based group nonetheless produces lower-end AI chips that are essential in many computer equipment, especially in the field of cloud computing.
“We’re working to make artificial intelligence accessible to everyone,” Intel CEO Pat Gelsinger told analysts Thursday. He expected some AI capabilities to move from servers to computers, and in this area the company has “advantages that you can use.”
The tech giants are also working on producing their own chips. In this context, the specialized website “The Information” reported that “Microsoft” has been seeking since 2019 to design a chip called “Athena”, which is supposed to ensure financial savings. The group did not confirm this information in connection with it.
Jack Gold saw that “competition is intensifying in this system.”
“God forbid”
Last September, US President Joe Biden considered the manufacturing of these evolving chips a matter of “national security”, especially in the face of Chinese ambitions.
He was speaking during the inauguration of a semiconductor factory in the state of Ohio (eastern United States), and stressed on that day the importance of a law passed on his initiative, according to which appropriations amounting to 52 billion dollars were allocated to revive semiconductor production.
Europe has allocated similar funds to encourage this sector as well. Intel has invested heavily in production and research on both continents.
Through this, Western countries seek to provide job opportunities, but more importantly to ensure the availability of these electronic components used in smart phones, cars, refrigerators, advanced weapons, and so on.
“If, God forbid, China attacks Taiwan and TSMC stops production, the global supply of chips will drop by up to 80 percent,” Jack Gold said.
But diversifying sources will take time. The analyst explained that “the construction of a new specialized factory takes between three and four years.”
Keen to remain a leader in the latest generation of chips synonymous with development in the field of artificial intelligence, the United States has also taken measures to restrict China’s access to key semiconductor manufacturing technologies.
