AI Chips: What Are They?

According to some experts, AI chips work similarly to graphics chips in high-end video games. AI chip manufacturers claim that their products will make AI systems like ChatGPT faster and less expensive. This newfound interest in these chips coincides with what some experts believe is an AI revolution that has the potential to transform the IT sector. Jensen Huang is the leader of Nvidia, a manufacturer of graphics chips. Based in Santa Clara, California, is the corporation. Additionally, Nvidia designs AI chips. Last Thursday, the company’s stock value rose by 25% as a result of official predictions of a significant increase in income. Recently, the shares of the corporation was briefly valued at over $1 trillion.

Researcher Hanna Dohmen works in Washington, D.C. at the Center for Security and Emerging Technologies. “There really isn’t a completely agreed upon definition of AI chips,” the speaker stated. The phrase refers to computational hardware that has been specially designed to handle tasks including AI. AI chips are utilized, for instance, in “training” AI systems to solve complex issues. In 1993, three entrepreneurs established Nvidia with the goal of enhancing computer graphics capabilities. The business created a new semiconductor known as a graphics processing unit, or GPU, in a matter of years. Video games run faster on GPUs because they can handle multiple graphical tasks at once.

The procedure is referred to as parallel processing. It would be crucial for the advancement of AI and video games alike. A GPU-based network helped two University of Toronto students win the 2012 ImageNet AI competition. Compared to rivals, the neural network could detect photo images with far reduced mistake rates. Eleven years following the ImageNet competition, Nvidia is the primary provider of chips used in the development and maintenance of AI systems. The H100 GPU, one of its most recent devices, features 80 billion transistors. That is purportedly roughly 13 million more than the most recent top CPU from Apple, which is utilized in MacBook Pro laptops. But the expense of this technology is high. The H100 is offered by one internet vendor for $30,000. It is not Nvidia that manufactures these GPU chips. The business is dependent on producers like Samsung Electronics of South Korea and Taiwan Semiconductor Manufacturing.