In 1965, American engineer Gordon Moore predicted that the number of transistors on a microchip doubles approximately every two years. This rule of thumb has largely held, and defined the state of the art of computer technology for decades. But as increasingly microscopic transistors run into fundamental physical difficulties and increasing costs, people have started to talk about a “post-Moore” industry.
For the semiconductor world, this is a huge challenge. But senior Chinese officials see opportunities.
Moore’s law has been a sore spot for China’s semiconductor industry. The country has plenty of good chip design and “packaging” companies. But as chips grow smaller, Chinese-owned chip fabrication plants are a consistent generation or two behind the cutting edge.
In Focus: Semiconductors is an ongoing premium series, tracking China’s semiconductor boom in charts and deep-dives. Available to TechNode Squared subscribers.
Over the past month, the concept of the post-Moore era has been a major talking point, permeating the highest levels of China’s government and the chip industry.
“In the post-Moore era, we have a great room for innovation and catching up,” Wu Hanming, academician of the Chinese Academy of Engineering and a former vice president of Semiconductors Manufacturing International Corporation, told an industrial forum on May 12.
On May 14, a Chinese State Council technology commission led by Vice Premier Liu He discussed what it called “potential disruptive technologies in integrated circuits (IC)” in a post-Moore world, according to a statement (in Chinese) published on China’s central government website.
Beijing has bet on a wide range of technologies that it hopes will make the country a leader in some of the cutting-edge areas of semiconductors. But experts say it is an open question whether Moore’s Law is really coming to an end soon, and whether China will have any advantages in a post-Moore world.
Is Moore’s Law really ending?
Quantum limits or no, components are still getting smaller. The size of transistors in microchips has dropped as the law predicts—the most advanced chips mass-produced by TSMC now have minimum feature sizes of 5 nanometers. On May 6, American tech company IBM said it is developing 2-nanometer chip technology which would fit “up to 50 billion transistors on a chip the size of a fingernail.”
However, the process of cramming more transistors into chips is facing physical roadblocks. One of the most critical problems is electrical leakage, which makes chips less energy-efficient and generates more heat as transistors get smaller than 10 nanometers.
There are also economic limits. New processor technologies often require new machinery, and the price increases as the transistors get smaller.
The cost of setting up a 3-nm chip production line is around $31.4 billion, which is more than twice that of a 5-nm chip production line and four times of a 14-nm one, according to a report (in Chinese) by brokerage Kaiyuan Securities.
Moore’s Law “is coming to an end for industries that do not have the necessary volumes to be able to afford the skyrocketing costs of cutting-edge chip design and manufacturing,” Jan-Peter Kleinhans, a global semiconductor value chain expert at Berlin-based Stiftung Neue Verantwortung (SNV), told TechNode.
The cost of developing a cutting-edge 5-nm chipset is “only economically viable in markets with high volumes, such as consumer electronics,” he said.
Kleinhans expects the trajectory of scaling the number of transistors may continue for another 10 years. “But most companies will not be able to afford it if they don’t have the necessary high volumes,” he added.
What’s beyond Moore?
In the post-Moore world, semiconductor industry experts have discussed three potential routes that industry insiders call “more Moore,” “beyond silicon,” and “more than Moore.”
More Moore suggests that the shrinking predicted Moore’s Law will continue, just slower, with fab advances continuing to drive the industry.
“Beyond silicon” seeks breakthroughs to continue the Moore’s Law trend, looking at non-silicon materials such as gallium nitride (GaN) and silicon carbide (SiC). A shift from silicon might be crucial for the chip-making industry so it can build smaller devices, Intel’s former director of technology strategy, Paolo Gargini, said back in 2006.
“With innovations on the material level, these compound semiconductors can be significantly more performant and efficient than purely Silicon-based chips—especially for power and radio frequency semiconductors,” said Kleinhans.
Finally, the “more than Moore” approach focuses on increasing the computing power of microchips by improving chip design and the performance of software running on the system.
“Chip design is increasingly important to develop application-specific processors such as artificial intelligence accelerators that are significantly more power-efficient than general-purpose processors for machine learning tasks,” said Kleinhans.
Other possible breakthroughs in a post-Moore world include novel ways to make chips at current sizes, such as stacked ICs, which stack silicon wafers so that they behave as a single device to improve performance.
What is China doing?
China’s showiest post-Moore plan is quantum computing technology, which seeks to harness the principles of quantum mechanics to supercharge processing power. Quantum computers are made of quantum circuits that run quantum bits, or qubits, which are similar to the bits in traditional silicon-based wafers.
In October 2020, China’s top leaders held a “study session” exploring quantum technology, during which Chinese President Xi Jinping said that “developing quantum science and technology is of great scientific and strategic significance.” China is about as close to widespread use of this technology as other major countries: not close.
The government is also supporting research into closer-range technologies, such as new materials and design techniques. On Jan. 28, the National Natural Science Foundation of China, an affiliate of the State Council, published a list (in Chinese) of technologies that it would fund in the “post-Moore Era.” They include new semiconductor materials, new design techniques and architectures, and high-density stacking IC packaging.
The funding project aims to “develop revolutionary basic devices, integration methods and computing architectures … and enhance China’s independent innovation capability and international status in the semiconductor field,” the foundation said.
Is post-Moore a ‘fresh start’?
Commentators regularly predict that China’s semiconductor industry will get a “fresh start” as Moore’s Law comes to an end.
“The Post-Moore Era is a great opportunity for China to outdo the world’s cutting edge in semiconductors,” Kaiyuan Securities analyst Liu Xiang wrote in an investment note.
China can focus on what is beyond the current trajectory of semiconductor development, rather than play catch-up in current tracks of technology, Stewart Randall, Head of Electronics and Embedded Software at consultancy Intralink, told TechNode.
Randall agrees there will be a “fresh start” in a post-Moore world, and “everyone is starting from the same start line.”
But “I don’t think there is any specific area China has an advantage,” he said.
Peng Lianmao, an academician of the Chinese Academy of Science, told Chinese media in April that China could “overtake on a bend” by developing carbon-based semiconductor materials.
Peng said that China had “basically solved challenges faced by carbon-based integrated circuits” and commanded “a full set of carbon-based IC manufacturing technologies.”
Peng, who is also a director of Peking University’s Institute of Microelectronics, led a team studying carbon-based semiconductor materials, Chinese media reported in 2020. He said then that his team was developing a 90-nm carbon-based chipset that will have the equivalent capability of a 28-nm silicon-based chipset.
But Kleinhans of SNV doesn’t reckon that it would be easier for China to make more advanced chips, because “it needs even more collaboration across the three production steps: design, fabrication, and packaging.”
“It’s all about path-dependencies, so no one has a ‘fresh start,’” he said. “‘Post-Moore simply means that all other process steps besides fabrication have to step up their game to increase the performance and efficiency of chips. The slow-down of Moore’s Law is not a silver bullet to leap-frog to the top.”