Calay News
ChatGPT, or generative artificial intelligence, is powered by tens of thousands of expensive and sophisticated electronic chips that have become a major problem for the tech giants involved in the AI race. The production of the vast majority of semiconductors used to power artificial intelligence models in the world relies on two companies: Nvidia for design and TSMC for manufacturing. Even if there is a supply, not every organization has the means to buy them. “Machines powerful enough to train generative AI models can cost hundreds of thousands of dollars,” says one expert.
Intel, the No. 1 US semiconductor company, which designs and manufactures the components, has lagged in this area. Although it continues to produce dedicated AI chips, which are less sophisticated but indispensable in many computer hardware applications, especially in the cloud, it is struggling to produce chips that can compete with Nvidia. The Californian group continues to produce chips dedicated to AI, which are less sophisticated but indispensable in many computing devices, especially in the cloud.
Microsoft has integrated generative AI into many online services but has not yet figured out how to generate additional revenue. Specialized site The Information reported that Microsoft has been developing since 2019 a chip called Athena, which is supposed to save money. Contacted by AFP, the group did not confirm this.
US President Joe Biden assured that the manufacture of these increasingly sophisticated chips is a national security issue, particularly in the face of China’s ambitions. Europe has also allocated a similar amount of funds to promote the industry. Western countries are looking to secure sources of supply for these electronic components present in smartphones, cars, refrigerators, and weapons. “God forbid, but if China were to attack Taiwan and TSMC were to stop producing, the global supply of chips would be reduced by 80%,” warns one analyst. But diversification will take time. “It takes 3–4 years to build a new specialized fab,” the analyst stresses.
