analog

China’s semiconductor sector has exploded to become one of the largest in the world and match the region’s prominence as the biggest consumer of chips. From 2015 to 2020 alone, revenue of China’s integrated circuit (IC) industry expanded at a rapid clip of 20% CAGR (RMB) to $128 billion.

Back in 2012, China ranked fifth among seven regions worldwide in IC wafer capacity but surged past the Americas and Japan in 2018 and 2019 to claim the number three position (figure 1). That’s a big deal given that ICs account for the largest share of wafer capacity excluding discrete, opto, MEMS and sensors.China’s IC wafer capacity growth accelerated to tune of 14% in 2019 and 21% in 2020 and is expected to grow at least 17% this year, as we report in the latest update of the World Fab Forecast, published December 3rd by SEMI. Of all regions, Taiwan boasts the second strongest growth rate over the same period at 3% to 4%.Figure 1: Total IC installed wafer capacity for top five regions The report shows that from 2019 through the end of 2021 China will have increased wafer capacity for memory by 95%, foundry by 47% and analog by 29%. Foundry will represent the largest portion of those gains, reaching 2 million wpm (200mm equivalents). Memory will follow at about 1.5 million wpm and then analog at over 120,000 wpm.But Chinese companies aren’t pulling off this feat singlehandedly. Many international companies are contributing to the wafer capacity increases in China (figure 2). Figure 2: IC wafer capacity in China by company origin The share of capacity contributed by Chinese-owned companies and international companies has changed little since 2012, though Chinese-owned companies saw a slight dip in their slice of the pie from 60% to 57%From 2019 through 2021, Chinese-owned companies will add almost 60% capacity for foundries, the most of all sectors. Companies including SMIC, Hua Hong Semiconductor, Nexchip, XMC and Hua Li Microelectronics are driving the increases.During the same period, Chinese-owned companies will ramp up memory capacity from basically zero to 300,000 wpm. Companies such as Yangtze Memory Technology and ChangXin Memory Technologies (CXMT), also known as Innotron, are powering the quick rise with aggressive ramps of 3D NAND and DRAM capacity.Among international-owned companies, TSMC and UMC are driving the largest share of foundry growth, while Samsung, SK Hynix and Intel are powering gains in memory capacity.More information is available in the World Fab Forecast report. The report currently collects information for fab equipment and construction investment, capacities, technologies and product types for over 280 fabs and lines in China alone, including 40 facilities that either began operation in 2020 or will from 2021 through 2024.Christian G. Dieseldorff is senior principal analyst in the Industry Research and Analysis group at SEMI in Milpitas, California.

The seemingly simple act of commanding consumer devices by voice is a choice that nearly 118 million Americans now make every day, according to a recent report from eMarketer, the digital marketing research firm.While the voice interface is convenient for users, its implementation comes at the potential loss of individual privacy. The reason? Always-on, always-connected voice-first devices such as Amazon Alexa and Google Home require a wall plug and an internet connection to powerful cloud processors, making it possible for cloud companies — however benignly — to collect data on personal habits, location and conversation that were never intended for sharing. Move processing to the edgeTo address concerns over user privacy, device designers are attempting to do more of the audio processing within the consumer device, rather than sending users’ voices into the cloud. Moving more processing to the edge is a trend across the Internet of Things (IoT) industry, and not just for voice data but for other types of sensitive or proprietary data as well.Yet designers have realized limited success because the conventional approach to always-listening edge processing is notoriously inefficient: It digitizes and processes 100% of incoming sound data even though up to 90% of the data is irrelevant noise. This digitize-first approach wastes vast amounts of system power digitizing and analyzing the audio signal as it searches for a wake word when there isn’t even speech present, making it impractical for use in small, battery-operated devices.Workarounds don’t workTackling this power issue is critical to keeping private data secure. Unfortunately, it’s also exceptionally difficult. Design engineers have tried workarounds to decrease power consumption in an always-listening system, including duty cycling and reducing the power of each individual component in the audio signal chain that handles the data. The reality is that these kinds of approaches don’t address the root cause of the problem: too much data.To truly tackle the problem, we need to change our approach to a system solution, not a component solution. By moving to a more efficient edge architecture that intelligently minimizes the amount of data that moves through the system, we can focus the system’s energy resources on analyzing voice and not on searching for a wake word in irrelevant noise. Analyze, THEN digitize It’s time to move away from the digitize-first approach that has dominated voice wake-up device architecture since the invention of voice-first applications.Inspired by the way the human brain efficiently filters incoming information, differentiating, for example, a dog bark from a baby’s cry, an ultra-low-power analog machine learning technology is changing this paradigm. For the first time, device designers can use low-power analog machine learning to detect which data are important for further processing and analysis prior to data digitization.Leveraging an analyze-first architecture, a new analog neuromorphic semiconductor platform allows the higher-power-processing components in the system to stay asleep until voice has actually been detected, and only then does it wake them to listen for a possible wake word.Delivering a post-microphone audio chain that draws as little as 25µA of current when always-listening and collecting preroll data, this analyze-first architecture allows designers to extend battery lifetime significantly. That’s the difference between smart earbuds that run for weeks instead of hours or a battery-powered smart speaker that runs for months instead of weeks.More importantly, it’s the difference between the current always-listening devices that indiscriminately record and send all sound data to the cloud, and one that has the localized intelligence to select and send only the relevant data, reducing the user’s vulnerability to the loss of private data.Balance convenience with privacyThe trade-off between making our lives easier and keeping our personal information private is a choice that we are asked to make throughout our day in a hundred different ways. Bringing more audio processing capability to the mobile device without draining the battery is the first step toward delivering more secure voice-first solutions. But to succeed in this effort, we must shift to a bio-inspired architecture that determines which data are important and requires further processing at the earliest point in the signal chain. Once we move to the analyze-first approach, only a small fraction of the tens of zettabytes of data collected by the forthcoming generation of always-on IoT devices will require further processing in the device and in the cloud.A better balance between cloud and edge processing is a better balance between convenience and privacy, and that’s a win for everyone.About the AuthorTom Doyle is CEO and founder of Aspinity. He brings over 30 years of experience in operational excellence and executive leadership in analog and mixed-signal semiconductor technology to Aspinity. Prior to Aspinity, Tom was group director of Cadence Design Systems’ analog and mixed-signal IC business unit, where he managed the deployment of the company’s technology to the world’s foremost semiconductor companies. Previously, Tom was founder and president of the analog/mixed-signal software firm, Paragon IC solutions, where he was responsible for all operational facets of the company including sales and marketing, global partners/distributors, and engineering teams in the US and Asia. Tom holds a B.S. in Electrical Engineering from West Virginia University and an MBA from California State University, Long Beach. For more information, please visit https://www.aspinity.com/Technology.Aspinity is a member of MEMS Sensors Industry Group (MSIG), a SEMI technology community, that enables the MEMS and sensor industry to address common challenges, innovate and accelerate business results.

This year, SEMI ISS covered it all – from a high-level semiconductor market and global geopolitical overview down to the neuro morphic and quantum level. Here are key takeaways from the Day 1 keynote and Economic Trends and Market Perspectives presentations.In the opening keynote, Anne Kelleher from Intel pointed to the huge growth of data, with fabs collecting more than 5 billion sensor data points each day. The challenge, Kelleher noted, is to turn massive amounts of data into valuable information. Moore’s law is not dead. New models of computing benefit still from Moore’s law and advances in Si/CMOS technologies for conventional, deep learning, neuro morphic and quantum computing.With customers expecting continual improvements in applications, the question is whether the chip industry is moving fast enough to meet these expectations, Kelleher said. A broad supply chain, equipment and materials innovations, and attracting the “best of the best” college graduates to fuel innovation is key, she said.In the economic trends session, Nicholas Burns (ambassador ret.) from Harvard University pointed out that we will see a major shift in power. The U.S. will remain the major world power over the next 10 years, but we will see a major shift in power in the next coming decades as the gap with countries like China, Russia and India continues to narrow.Duncan Meldrum from Hilltop Economics said that we are passing the peak growth of economic cycle. He warns that a more likely outlook is that a global growth recession is developing. Although semiconductor MSI growth will see a noticeable slowdown in 2019 and 2020, the semiconductor industry is still healthy over the longer term.Bob Johnson from Gartner sees demand shifting from consumer to commercial applications with higher ROIs and budgets. AI, IoT and 5D are the major enablers. He sees structural changes in the semiconductor industry especially for memory but also for Moore’s law with increasing costs and fewer players.The DRAM markets shows volatility and NAND market may be negative in 2019 but non-memory are expected to accelerate mainly because of increasing content and some price hikes.Overall Gartner expects good long-term growth with a CAGR (2017 to 2022) of 5.1%, outpacing 2011 to 2016 CAGR of 2.6%. After a strong 2018 with 13.4% revenue, he forecasts a slower 2019 with 2.6% growth followed by a 8% growth in 2020 and negative growth rate in 2021.Andrea Lati of VLSI went “Back to fundamentals” in his presentation about the industry. VLSI sees a downside bias due to slowing global economy, tariffs, and trade wars. Future drivers are data economy, cloud, AI and automotive.As memory leads the 2019 slowdown, analog, power, logic and other sectors remain in positive territory. VLSI lowered its semiconductor equipment forecast for 2018 from 20% (Jan. 2018) to 14% (Dec. 2018) but increased its sales outlook from 8% to 15% in 2018. VLSI expects revenue to slow into the first half of 2019 but increase to over 4% in the second half of the year, resulting in total 2019 drop of 2.7%. Semiconductor equipment sales are expected to drop from 14% in 2018 to -10% in 2019.Michael Corbett of Linz Consulting, covering wafer fab materials in the years of 3D scaling, sees these as good times for the industry. His outlook for wafer fab materials is bullish based on strong MSI and because wafer fab materials suppliers are getting bigger because of M As.In the Market Perspective session, Sujeet Chand of Rockwell Automation pointed out that as more and more data is generated, the problem is how to get value of all the data collected. There is a need to create the right architecture for machine learning and AI and big data is increasingly being replaced by contextual/structured data. He expects Industry 4.0 to drive foundries to become smaller, more flexible and more productive.In the Technology and Manufacturing session, Aki Sekiguchi of TEL addressed process challenges in the age of co-optimization. The semiconductor industry continues to expand, driven by massive growth of interconnected devices, with heavy demand for processing power and storage. He expects an exponential increase of data from about 40ZB in 2018 to 50ZB in 2020 to 163 ZB in 2026.Major technologies such as DRAM, 3D NAND and logic are dealing with scaling challenges. The density of DRAM (Mb/chip) is plateauing according to 2015 to 2020 trend data, with DRAM is in need of EUV. Memory capacity demand is leading to increasing layers and higher aspect ratios that is concern for 3D NAND and mainly for plasma etch. With Logic already implementing 3D structures, it appears to be in a solid position. Buddy Nicoson of Micron talked about his 50 years in the industry and looked ahead to the next 50. The anchors – quality, cost, scale and speed – won’t change. It has been a great journey so far with unprecedented opportunities and challenges ahead of us. We are getting into a convergence (specialization, integration) and solution-based phase. We will see some inflection points in the coming years, with the best yet to come.Christian G. Dieseldorff is senior principal analyst in the Industry Research and Analysis group at SEMI in Milpitas, California.