For the majority of digital enthusiasts, "Moore's Law" is a very well-known and familiar law.
This law was first proposed in 1965 by Gordon Moore (Figure 1), one of the founders of Intel Corporation, who believed that the number of transistors that could fit on an integrated circuit could double every 18 to 24 months, while the cost would be halved accordingly.
This law has been the cornerstone of the computer hardware development industry for the past few decades, and the relevant industry chain manufacturers have been working closely with each other to work towards this goal. For a long time, the actual development of the chip industry is basically consistent with this law, so this law has had a very significant, far-reaching and positive impact on the entire technology industry.
However, in recent years, the performance of the chip industry has been unsatisfactory, whether it is the GPU of the graphics card, the CPU of Intel and AMD, or the SoC of the mobile phone, they have all fallen into such a very embarrassing situation:
The number of transistors is increasing, the manufacturing cost is getting higher and higher, but the performance has not been greatly improved, the "perception is not strong", and the power consumption is getting higher and higher, so Intel is also nicknamed "toothpaste factory" by netizens.
Therefore, in this context, many people, including industry professionals, such as Nvidia CEO Jensen Huang, believe that "Moore's Law is dead" and that the law is no longer suitable for guiding the future development of the chip industry, but there are also different opinions in the industry, and there has been debate.
Recently, Intel CEO Kill Singer talked about his views on Moore's Law in a speech, which is available for reference.
He believes that Moore's Law is not dead now, "still alive", but that "we are no longer in the ** era of Moore's Law, and now (technological development) is much more difficult". Now the pace of development has slowed down, and the number of Moore's Law transistors can be doubled in about two years, and now the actual situation is that it will take about three years.
In fact, this is not the first time that Gelsinger has made a strong argument in favor of Moore's Law, and since he became CEO in 2021, he has made similar statements on several occasions, and in order to prove his point, he laid out some arguments and Intel's future development plans.
According to Kilsinger, Intel could make chips with 1 trillion transistors by 2030, up from about 100 billion transistors in a single package today, and at least 10 times more in the future.
In order to achieve this goal, Intel will focus on four directions in the future: new RibbonFET transistor technology, PowerVIA power supply technology, next-generation process nodes and 3D chip stacking technology, which will help further promote the rapid development of chips in the future, and will generally follow Moore's Law.
Kielsinger also proposed a "Super Moore's Law", also known as "Moore's Law 2."0", claiming that in the future it can be achieved by using 25D and 3D chip packaging technologies, such as FoverOS, continue to dramatically increase the number of transistors inside a chip.
As for Moore's Law, Kielse's views have changed and softened in recent years. A few years ago, he was confident that Moore's Law would continue to guide chip development, but now says the pace of development has slowed and it may take three years to double the number of transistors.
On December 21, Intel received a high-NA EUV lithography machine (Figure 4) delivered by Dutch lithography giant ASML, which can be used to manufacture chips at the 2nm node, and is expected to start mass production in 2025 based on 18A (18 nanometers) process of chips, ambitious.
Generally speaking, Kilsinger and Intel are very optimistic and confident about the technological development of the chip field in the next five years, and whether Intel's various planning goals can be successfully achieved and whether they can take off the hat of "toothpaste factory" are expected to be the most critical period in the next one or two years, and the last two years will be known.