Deep Learning Next Big Chip Growth Driver

 Count Wally Rhines, semiconductor industry veteran and long-time CEO of Mentor Graphics, among the many who believe that deep-learning hardware will drive the next wave of growth for the semiconductor industry.

Speaking at the GSA European Executive Forum here this week, Rhines added that memory will continue to be a key driver of the chip industry going forward. Despite the volatility of the semiconductor industry, R&D investment continues to be around 14% of revenue as it has been for the last 36 years, Rhines said, dismissing arguments put forth by some that there isn’t enough being ploughed back into R&D to maintain sustained growth.

On growth, we should be watching out for China, and a lot of growth will come from visual processing for AI applications, Rhines said.

He offered his perspective on the future of the semiconductor industry and why revenue forecasts from research and analyst firms have been consistently off target. With numerous historic charts and data, he highlighted the 96% error in the Semiconductor Industry Association’s three-year projection in 2002 and how analysts were off by 17% for the one-year forecast for 2017.

He questioned why it was so difficult to forecast semiconductor industry revenue, especially when IC unit demand has always been predictable at around 8% per year and silicon area shipments can also be predicted and easily measured.

A reason for the discrepancy more recently is because of the rise of system IC companies, whose revenue is only partially counted. “With leading system and digital companies also now becoming the new SoC designers and manufacturers, semiconductor industry forecasts are often deprived of the cost of the system,” said Rhines. “Systems companies comprised 13% of pure-play foundry sales in 2017.”

Rhines asked whether the 2017 boom in semiconductor revenue was a fluke and then went on to answer his question. “When you consider the evolution of memory versus logic, memory now dominates the percentage of transistors manufactured,” he said.

But, he added, when you consider that the next wave of computer architectures for visual processing and AI will require vast amounts of memory for brain-like pattern recognition, this will drive up memory volumes and, hence, semiconductor revenues.

This will be dominated by areas like autonomous vehicles, with their need for visual information capture, storage, and processing with the corresponding flood of data.  

In addition, AI-related companies are also the ones that tend to be funded heavily — Q4 2017 shows some huge investments in this area, with companies like Horizon Robotics raising $100 million, Graphcore raising $50 million, and Vayyar Imaging raising $45 million.

Addressing the wider ecosystem, free trade

Rhines was upbeat about the future prospects of the semiconductor industry, and the conference generally understood that there was a new reality for the industry. Sandro Grigolli, EMEA executive director, said at the opening session that the rise of companies not previously part of the industry, like Google, Amazon and Facebook, meant that the traditional semiconductor industry needed to figure out a way to “play gracefully with these companies.”

“We are part of a larger ecosystem, and the GSA needs to change to accommodate this, which is what we are doing with the introduction of automotive and IoT security interest groups,” Grigolli said.

On the threats of the new companies entering the chip industry, Faraj Aalaei, president and CEO of Aquantia, wasn’t too worried. “These companies are not a long-term threat to the semiconductor companies — over time, they’ll see the value in the learnings and experience [of established semi companies]. If your customer can build a better chip than you, you don’t deserve to be in the chip business — it just means your edge is not as good as you think it is.”

Aalaei also said that the biggest problem with the semiconductor industry is Moore’s law. “By always referring to this law, the customer interprets this as, ‘I’m going to get the same product in 12 months’ time at half the price. So we need to stop talking about Moore’s law.”

The tensions in free trade also came up, with Helmut Gassel, chief marketing officer and board member of Infineon Technologies, commenting, “In San Diego, two weeks ago at the World Semiconductor Council, we agreed that we need to work hard to ensure free trade across the globe and to convince politicians that trade wars are not good for growth.”