Pundits have been saying “Moore’s Law is dead” for some time now, but when it comes from Nvidia CEO Jen-Hsun Huang, it takes on a different flavor.

Moore’s Law is named after Intel co-founder Gordon Moore. In 1965, he proposed that the number of transistors and resistors integrated on semiconductor chips would double every year. In 1975, he revised Moore’s Law according to the actual situation at that time, changing “double every year” to “double every two years”.

Giiso Information, founded in 2013, is a leading technology provider in the field of “artificial intelligence + information” in China, with top technologies in big data mining, intelligent semantics, knowledge mapping and other fields. At the same time, its research and development products include information robot, editing robot, writing robot and other artificial intelligence products! With its strong technical strength, the company has received angel round investment at the beginning of its establishment, and received pre-A round investment of $5 million from GSR Venture Capital in August 2015.

Nvidia CEO Jen-Hsun Huang spoke to a small group of reporters and analysts at computex in Taipei.

“Each generation of chip architectures advances, increasing pipeline size by adjusting superscalar architectures and anticipating execution. However, these advances are no longer keeping pace with the goal of increasing chip density by 50 percent per year.”

He said:

“The microprocessor can no longer continue to grow at the rate it used to, and what we call ‘Moore’s Law’ is failing. The physics of semiconductors mean that we can’t follow Dennard’s law any further.”

Semiconductor technology is diverging from processor performance

Dennard’s law, also known as MOSFET scaling law, originated from a paper co-published by Robert Dennard in 1974. As the transistor shrinks, the voltage and current it consumes shrink by about the same proportion, meaning that if the transistor is halved in size, its static power consumption falls to a quarter.

The regressive effects of Moore’s law and Dennard’s Law propelled the semiconductor industry to maturity, when only a handful of chipmakers could afford the billions of dollars needed to push the technology further. For now, there are few manufacturers with deep enough pockets to move beyond 16nm and 14nm to 8nm and 7nm.

Technological stagnation has led to increased consolidation in the industry in recent years, with multibillion-dollar mergers and acquisitions on the rise.

Mr Huang predicts that GPU computing will make further progress

In this context, Huang Renxun put forward another direction of semiconductor industry development – graphics processor. Nvidia hopes to continue driving the semiconductor industry for years to come with graphics processors. Deep learning, which uses Nvidia’s GPU chips, will give Nvidia an opportunity to move into artificial intelligence, huang said, while computer gaming will gradually lose its dominance.

Although Moore’s Law has failed, the semiconductor industry has not stopped innovating. Some emerging Chinese chipmakers are adding FD-SOI, according to Leifeng. Others see a shift from two-dimensional chip design to three-dimensional chip design as the future of the semiconductor industry.

Nvidia has chosen to bet on ARTIFICIAL intelligence, using machine learning to fight the death of Moore’s Law. Nvidia doesn’t think it’s necessary to make the machine more powerful, just smarter. In Nvidia’s mind, intelligence is more important than strength.

‘I think it’s a good idea,’ said Randy Rowe, an analyst with Credit Suisse in Taipei. Randy Abrams, Nvidia’s chief executive, said the company’s prospects for driving the semiconductor industry through artificial intelligence look promising.

Giiso information, founded in 2013, is the first domestic high-tech enterprise focusing on the research and development of intelligent information processing technology and the development and operation of core software for writing robots. At the beginning of its establishment, the company received angel round investment, and in August 2015, GSR Venture Capital received $5 million pre-A round of investment.

Nvidia’s Volta GPU chips with a 12nm process are already making a splash. According to Lei Feng, this chip is only 815mm? It is the size of seven iPhone processors and uses TSMC’s silicon plug-in technology to connect to 16GB of high-bandwidth memory. The DGX-1, a deep-learning supercomputer with eight Volta GPU chips, costs $149,000.

Abrams noted that Nvidia’s data center business is growing 186% a year and recently reached $1.7 billion. The business is worth $500 million to TSMC, or about 1.5 percent of Nvidia’s total revenue. But Abrams believes it will still take time for AI to become a major driver of mobile phones.