With the release of their new GeForce RTX 2000 series GPUs, Nvidia took an opportunity to share some thoughts on Moore’s Law, the well-known prediction that integrated circuit densities will double every two years. Moore’s Law is dead, and we have killed it, said Nvidia CEO Jensen Huang at the company’s GPU Technology Conference. But wait!
At a corporate launch ceremony on Tuesday, Intel CEO Pat Gelsinger declared that Moore’s Law, a maxim from the firm’s creator that dates back to the 1960s, is “alive and thriving.” According to Gordon Moore’s idea, processors will continue to get quicker and less expensive over time.
The message being preached by Nvidia, which is currently nearly three times more valuable than Intel, is very different. Moore’s Law has come to an end, according to co-founder and CEO Jensen Huang last week.
Huang warned investors after introducing new goods that “the approach of deploying brute force transistors and the achievements of Moore’s law has basically run its course.”
The distinction highlights the glaring disparity between Intel and other American semiconductor firms. While Nvidia and other companies mostly rely on third-party foundries outside of the United States, Intel has pledged to keep making some of its chips.
Moore predicted that the number of transistors on a chip will double every two years, increasing computing power. This is known as Moore’s Law. Making transistors smaller is necessary to increase the number of transistors on a chip, which calls for breakthroughs in manufacturing technology.
For many years, Intel was the industry pioneer in semiconductor production, regularly producing chips with the highest transistor densities. However, Taiwan Semiconductor Manufacturing Company and Samsung have just outpaced Intel, since they can now make CPUs with 5-nanometer transistors, but Intel is still to 10-nanometer and 7-nanometer technology.
To regain “performance leadership,” one of Intel’s primary corporate objectives under Gelsinger, is that the company’s chips must be equally quick and effective as those produced by competitors at independent foundries. While developing a new node with fewer transistors typically takes two years, Intel plans to progress its manufacturing five “nodes,” or five transistor sizes, in four years to catch up.
Since Intel is still actively working to fit more transistors onto a single chip, Moore’s Law must continue to hold true.
However, size has its limits since eventually transistors get so tiny that they encounter a physics issue. Gelsinger referred to that as a “day of reckoning” on Tuesday.
In order to continue packing more transistors onto each chip even as they become smaller and can only be measured in angstroms, or a hundred-millionth of a centimetre, Gelsinger noted that Intel is focusing on manufacturing innovations including new lithography techniques and RibbonFET design.
We want to have 100 billion transistors on a single package by 2025. We are on target to have a trillion transistors in a single package by the end of the decade, according to Gelsinger.
‘The path forward’
The world’s largest chipmaker, TSMC, which now uses the most cutting-edge semiconductor manufacturing processes, produces Nvidia’s most recent processors. Nvidia creates semiconductors, but is less concerned with the production end.
Nvidia’s solution to the engineering problem of producing smaller transistors is a theory Huang refers to as “accelerated computing,” as opposed to Moore’s Law. In his ideal world, the processor that performs demanding applications like artificial intelligence the best is the graphics processor that Nvidia creates. In other words, Intel’s expertise is less necessary.
According to Huang, there are no longer any possibilities to follow Moore’s law’s price-performance curve going ahead. Therefore, after 15 years — almost 20 years — of pursuing accelerated computing, I believe that, very roughly speaking, virtually it’s conventional wisdom that accelerated computing is really the path ahead if you want to be able to conduct larger-scale computing and to do it in a cost-effective fashion.
Intel announced new chips and software on Tuesday as it attempts to bounce back from years of slipping performance and profits. Over the past five years, Intel’s stock has lost 28% of its value, while Nvidia’s stock price is up over 180% (even after dropping 58% in 2022).
Intel announced new Core desktop processors for gamers and consumers with improved performance, a graphics chip for data centers called Ponte Vecchio and software called Unison that enables Windows PCs to pair with Android phones and iPhones to send texts and make calls.
