MEMBER LOGIN

NEWS

Electronics Industry News

28 August 2018

China vendors set to dethrone Korea brands in global smartphone market

Colley Hwang, DIGITIMES, Taipei

 

The second quarter of 2018 marked for the first time that China-based Huawei has outraced Apple to second place in the global smartphone market, according to IDC data. With that performance, Huawei’s consumer business group CEO Richard Yu has boasted that his company aims to overtake Samsung to become as the world’s top smartphone vendor by the fourth quarter of 2019.

Meanwhile, the anticipated rollouts of foldable smartphones by handset vendors in the second half of 2018 is set to make the segment a new battlefield of the industry where China-based handset makers are to show off their strengths, which no can afford to ignore. After all, China-based vendors have become fast movers instead of followers in the past year as far as AI-centric smartphones and iris identification technology are concerned.

Since the development of the smartphone industry has already entered its plateau period, any maker or party which is able to secure a preemptive presence in the foldable smartphone segment is likely to win in the looming battle.

Foldable smartphones

A crucial element of the foldable smartphones lies in the configuration of flexible display. Samsung and Huawei are said to prefer an in-folding design, while Xiaomi is likely to choose a design that will place the foldable display outwardly when folded.

Xiaomi is likely to give priority to promoting the foldable smartphones in the domestic market, while Huawei is trying to trump Samsung for the release of foldable models in the global space. Thus, the year 2019 will be a turning point for the competition between smartphone vendors in China and Korea.

China’s players are also narrowing their combined market share against Korea-based brands in markets spanning Asia, Oceania, Africa and East Europe with the exception in North America.

At the same time, Samsung has been losing ground in the world’s largest single market, China, where handset sales could reach as many as 410 million units a year, accounting for 31% of global sales. However, Samsung currently accounts for only 1% of the handset market in China, according to a Bloomberg report

In the India market, although Samsung managed to recapture the top-vendor ranking in the second quarter of 2018, it was losing out its share in the high-end segment to China-based OnePlus.

OnePlus captured a 40.5% share of the above INR30,000 (US$437) segment of the smartphone market in India in the second quarter of 2018, a significant improvement from the 8.8% recorded a year earlier. The above INR30,000 segment accounts for only 3% of India’s total handset market in terms of unit shipments but 12% in terms of profits generated by the industry.

Meanwhile, Huawei is exerting all-out efforts to promote sales in Europe. Huawei and other China’s brands currently account for a 23.4% share in the Eastern European market, as Korea’s vendors are seeing their share drop to 34.3% from the previous 37.3%.

Huawei also aims to ship over 200 million smartphones in 2018, with its shipments to account for 30% of total smartphone shipments in China in the year and present a 51-fold increase in seven years.

Xiaomi is approaching the brick and mortar stores in Western Europe, using a new Poco Phone brand, as it is trying to get rid of the image of being a second-tier Chinese vendor selling low-priced devices.

With support from China’s government, China’s brands have also managed to lower the gap of market share against Korea’s in Latin America to 6pp currently from the previous 12pp; and in Africa and the Middle East to 18.3pp from 3pp.

Penetrating into Korea market

And to the dismay of Korea’s makers, Huawei and Xiaomi both are taking aggressive marketing strategies to begin selling their smartphones directly in Korea since July 2018.

The handset market in Korea totals over 20 million units a year, with Samsung taking a majority of 55%, followed by Apple with 28.3% and LG Electronics with 15.7%.

But market sentiment in Korea may be changing under the price pressure from Huawei and Xiaomi.

In fact, Samsung saw the operating profits of its handset business unit decline 34.2% on year to KRW2.67 trillion (US$2.358 billion) on revenues of KWR24 trillion in the second quarter of 2018.

Analysts at Korea-based securities investment firms estimated that shipments of the Galaxy S9 devices totaled eight million units in the second quarter, accounting for only 10.26% of the 78 million handsets shipped by Samsung in the quarter.

The smartphone market in Korea is expected to scale down by 5.9% to 20.24 million units in 2019 and to 20.11 million units in 2023, according to an estimate of Euro Monitor. The shrinking market in the domestic front is certainly unfavorable to Samsung.

In sum, China’s smartphone brands are approaching the global market in an echelon form with their devices characterized with innovative designs, competitive pricing, enhanced local presence, in addition to strong government support. Major brands in China include Huawei, Xiaomi, Oppo, Vivo, OnePlus, Lenovo, Coolpad, Hisense, and ZTE. And seven out of the world’s top-10 handset brands are from China currently.

Premium smartphone models rolled out by China-based brands normally feature the same high-end hardware specifications as those found in the flagship models from Apple or Samsung but are available at two-thirds of their prices.

Aside from hardware specs and related technologies, China’s brands are offering a complete line-up of products in high quality and are readily available through established channels at affordable prices. Smartphone vendors from Japan and Taiwan have already been defeated, and how much longer can the Korean brands hold on to their lead?

Which explains why

Seeking new growth drivers, Samsung plans $22 billion spending on new technology

SEOUL (Reuters) – Samsung Group will invest 25 trillion won ($22 billion) in artificial intelligence, 5G mobile technology, electronic components for autos, and the biopharmaceutical business in pursuit of new growth areas.

In flagging specific areas of investment, the plan is the first of its kind for the conglomerate and suggests more emphasis than usual on expanding into new segments as its core semiconductor and smartphone businesses show signs of weakness.

The investment is also part of a bigger 180 trillion won three-year plan to create jobs and secure fresh revenue streams, providing the first hint of direction for Samsung since group heir Jay Y. Lee was released from jail in February on a suspended sentence following a conviction for bribery. He is appealing the conviction.

“The total investment size is not surprising. But Samsung can be more active in M&As to do better at AI or 5G, when Lee is fully back to his business,” Greg Roh, an analyst at Hyundai Motor Securities, said.

Samsung Electronics said the announcement followed “many months of deliberations and review” by the group. Samsung Electronics, the world’s top maker of memory chips and smartphones and the group’s flagship company, did not provide a breakdown of the figures but said it will provide the funds for most of the investment.

It said the group plans to significantly expand artificial intelligence (AI) research capability, increasing the number of advanced AI researchers to 1,000 across its global AI centers, and heavily invest in biopharmaceuticals.

“Samsung has seen strong growth from both its contract manufacturing and biosimilar businesses. It will continue to invest heavily in the businesses, including developing and manufacturing biosimilars,” the statement said.

The announcement sent shares in contract drug maker Samsung BioLogics (207940.KS) up 7 percent. Shares in the group’s information technology services affiliate Samsung SDS (018260.KS) rose nearly 6 percent.

 

Samsung Electronics shares were little changed.

The 180 trillion won plan includes capital spending as well as research and development in chips and displays and marks a 6 percent increase from its spending over the past three years.

Samsung Electronics Co Ltd

 

Of the total, 130 trillion won will be spent in South Korea, and is expected to create 40,000 jobs, or 20,000 more than its previous hiring plans, Samsung said.

The announcement comes after South Korea’s finance minister met Jay Y. Lee on Monday and called for help in job creation, a priority for the country’s President Moon Jae-in.

Lee has sought to expand Samsung’s business into automotive electronics components. He signed off on an $8 billion purchase of U.S.-based Harman International in 2016.

But the expansion drive has since stalled after the political scandal that led to Lee’s conviction and a shake-up of top management.

Samsung Group has 62 affiliates, with total assets of 399 trillion won, according to the country’s Fair Trade Commission.

Reporting by Ju-min Park; additional reporting by Heekyong Yang; Editing by Miyoung Kim, Himani Sarkar and Neil Fullick

The Thomson Reuters Trust Principles.

 

Electronic Shortages – Bad, and getting worse

EBN – Hailey Lynne McKeefry, Editor in Chief

For weeks and months, it’s been a buyer’s nightmare: all sorts of electronic parts are on allocation and prices are on the rise. Everything from capacitors to hard disk drives is in short supply. There’s no end in sight.

The recently released Fusion Worldwide’s Greensheet traces some of the trends and there is little good news to be found.

MLCCs

Multilayered ceramic chip capacitors (MLCCs), a basic staple in many electronic products especially in the telecommunication and automotive markets, are in short supply. That’s a reality that is pushing pricing. The Greensheet explained:

The cost of MLCCs continues to rise and there is a growing consensus that the high prices will remain long after the shortage ends. Passive component makers have long felt that their prices have been too low and undervalued relative to the manufacturing cost.  The rising labor costs in China and the stricter environmental control restrictions will cause prices to rise further, while preventing them from reducing to pre-2016 costs even after the shortage has ended.

Even as component makers are rushing to get more capacity online with new factories focusing on these high-demand passives. However, for MLCCs limitations on materials and production equipment as well as demanding technology hurdles, makes it unlikely that these passives will be part of the shift. However, a number of suppliers are moving their capacity capabilities away from larger case size legacy products to increase volume on smaller case size technology. These shifts take time, though, so capacity likely won’t bounce back any time in the near future.

Other integrated circuits

Currently, lead times on diodes are stretching out across the board. “For example, current lead times are around 30 weeks for TVS Diodes, 24-36 weeks for Schottky diodes and 30 to 40 weeks for Zener diodes,” the Greensheet said. “In comparison, last year around this time, lead times for TVS diodes were at 20 weeks, and small signal diodes and rectifiers were around 18 to 25 weeks. “

Metal-oxide-semiconductor field-effect transistor (MOSFET) prices are likely to also be forced up by increased demand, with increases of 10% to 25% across the market. These shifts can be traced to increased demand from the automotive market as well as a limited supply of silicon wafers and other raw materials needed to produce power MOSFETs “Most industry professionals are estimating that parts won’t be back in regular supply until 2020,” the Greensheet explained.

CPU & memory

Shortages of central processing unit (CPU) chips is ubiquitous, with rising demand combining with supply limitations to create a pinch. “Virtually all segments including Xeons, desktop, mobile and chipsets are facing varying degrees of turbulence,” the Greensheet said. “The strongest number of PC shipments in years have pinched supply availabilities across all client segments.”

DRAMs offer a glimmer of hope, as current demand is currently in pace with supply. At the same time, pricing is also likely to rise again this year although not substantially. In July, market pricing increased higher than last month at about three to five percent, the report said.

Hailey Lynne McKeefry, Editor in Chief, EBN

 

Algorithms Outpace Moore’s Law for AI

Alexei Andreev and Jeff Peters, Autotech Ventures

Moore’s Law continues to change the world. But algorithmic advances have been every bit as critical for driving electronics.

How confident are we that algorithms of tomorrow are a good fit for existing semiconductor chips or new computational fabrics under development? With algorithmic advances outpacing hardware advances, even the most advanced deep-learning model could be deployed on a chip as small as a $5 Raspberry Pi.

Which solves faster: a top modern algorithm on a 1980s processor or a 1980s algorithm running on a top modern processor? The surprising answer is that often it’s a new algorithm on an old processor.

While Moore’s Law gets a lot of attention as the driver of rapid advance of electronics, it is only one of the drivers. We regularly forget that algorithmic advances beat Moore’s Law in many cases.

Professor Martin Groetschel observed that a linear programming problem that would take 82 years to solve in 1988 could be solved in one minute in 2003. Hardware accounted for 1,000 times speedup, while algorithmic advance accounted for 43,000 times. Similarly, MIT professor Dimitris Bertsimas showed that the algorithm speedup between 1991 and 2013 for mixed integer solvers was 580,000 times, while the hardware speedup of peak supercomputers increased only a meager 320,000 times.  Similar results are rumored to take place in other classes of constrained optimization problems and prime number factorization.

What does that mean for AI?

In the past half-decade there has been an explosion of artificial intelligence (AI) in the academic, industrial and startup world. Probably the greatest inflection point occurred when AlexNet, a team from University of Toronto, destroyed the competition in the 2012 the ImageNet Large Scale Visual Recognition Challenge (ILSVRC) using deep-learning. Deep-learning has since become a key formulation for AI implementations.

Advances in computer vision sprawled into natural language processing and other AI domains. Smart speakers, real-time computer translation, robotic hedge funds, and web reference engines do not surprise us anymore.

AI has also become a driving force in the transportation industry (our investment focus at Autotech Ventures). We’ve observed great promise in advanced driving assistance systems (ADAS), autonomous driving, fleet inspection, manufacturing quality control, in-vehicle human-machine interface, and more. So far, we have made several investments into startups developing various AI solutions in this domain including ADAS and autonomous driving, visual inspection, and edge computing. As we analyze these opportunities, the interplay between algorithm and hardware is a key consideration in our decision-making.

Public attention on AI hardware

The inflection point for deep-learning based AI was followed by ravenous demand for graphical processing units (GPUs). Because of their parallel computation prowess, GPUs happen to be surprisingly efficient for the logic employed by deep-learning algorithms. Nvidia, the major manufacturer of GPUs, “decimated” their competition, and saw their stock price climb 20-fold from 2013 to 2018.

Of course, Nvidia competitors are trying to catch up. Qualcomm, Arm, and others have focused attention on AI chip design, while Intel acquired startup AI chip company Nervana Systems. Google, Facebook, Apple, and Amazon have all waded in to build their own AI processing units for their data centers as well as other initiatives. There are also a handful of startups who jumped in when they saw the opportunity (e.g. Graphcore, Mythic, Wave Computing, Cerebras, SambaNova) to build a better-designed Turing machine. A few others like D-wave Systems and IBM are actively exploring post-Turing architectures. For most chip development efforts, the goal is to catch up to or beat Nvidia. However, most of these processors — to the best of our knowledge — are being designed for the AI algorithms of today.

We are going through an ongoing Cambrian explosion of alternative AI chip designs despite their enormous upfront development costs. The end game for AI is so breathtakingly large that industry participants are willing to invest heavily, holistically matching hardware to the underlying mathematical algorithms. But, how confident are we that algorithms of tomorrow are a good fit for existing semiconductor chips or new computational fabrics under development?

Given the rate and the magnitude of algorithm evolution, many of those alternative AI chip designs may become obsolete even before their commercial releases. We speculate that AI algorithms of tomorrow might demand different compute architectures, memory resources, data-transfer capabilities, etc.

ven though deep-learning frameworks have been known for a long time, it has only been recently that they have been put into practice thanks to those Moore’s Law-predicted hardware advancements. The original mathematics were not necessarily designed for engineering practice as early researchers could not dream of computational power that $1,000 can buy today. Today, many implementations, driving toward accuracy, simply add more layers (e.g. making them “deeper”) or add more data using the original mathematics. This simply strains the computational capacity of the GPU. Only a small fraction of researchers is focusing on the hard problem of improving the underlying mathematical and algorithmic frameworks.

There is opportunity to recognize and capitalize on these novel mathematical advances. We have observed methods that trim redundant mathematical operations to reduce computation time, squeeze the convolutions to smaller matrices to reduce memory requirements, or binarize weighting matrices to simplify mathematical operations. These are the first forays into the algorithmic advances that are starting to outpace hardware advances.

For example, DeepScale*, spun-out of UC Berkeley research, “squeezes” AI for advanced driver assistance systems (ADAS) and autonomous driving onto automotive grade-chips (as opposed to GPUs). Their neural network models have demonstrated 30 times speedup compared to leading object detection models using algorithms alone while reducing energy and memory footprint enough to run on existing hardware in just a couple of years.

Another example of such algorithmic leapfrogging came from researchers from the Allen Institute of Artificial Intelligence. Using a novel mathematical approach employing binarization of neural networks, they showed that they can drastically increase speed as well as reduce power and memory requirements. This enables even the most advanced deep-learning model to be deployed on a chip as small as a $5 Raspberry Pi. The researchers recently spun out the algorithms and processing tools as XNOR.ai* to deploy AI on edge devices and drive further algorithmic advances for AI.

The interesting point is that the new binarization frameworks fundamentally change the type of optimal processing logic. Instead of the 32-bit floating point convolutions required to solve neural networks they need only bit counting operations – shifting the balance of power away from GPUs. Further, the computational resource requirements can be further reduced if these algorithms are matched by specifically-designed chip.

And algorithmic advances will not stop. Sometimes it takes years or even decades to invent (or perhaps discover) new algorithms. Those breakthroughs cannot be predicted in the same way as computational advances driven by Moore’s Law. They are non-deterministic by nature. But when they happen, the entire landscape shifts often turning incumbent dominant players into vulnerable prey.

Black Swans

In his bestseller The Black Swan: The Impact of the Highly Improbable, Nassim Nicolas Taleb showed that optimal decision-making depends heavily on whether the analyzed process is unpredictable or uncertain. In other words, are we dealing with “known unknowns” or “unknown unknowns”?  Algorithmic innovations are fundamentally unknown unknowns. Betting on those developments requires constant attention due to their non-deterministic time-to-discovery and unpredictable impact.

However, the field of applied mathematics in general, and domain of artificial intelligence in particular, saw several disruptive algorithmic discoveries in the last two decades that — alongside with GPUs — brought AI from an obscure field of research to the forefront of commercialization.

We recognize and appreciate potential for computational “Black Swans” that would make existing chip architectures obsolete or reshuffle their relative popularity overnight. For us, these Black Swans could finally unleash safe autonomous vehicles as well as many other, as of yet unknown, use cases.

— Alexei Andreev PhD is a managing director and Jeff Peters PhD is a principal at Autotech Ventures, a venture capital firm focusing on transportation-related technology. (Disclosure: DeepScale and XNOR.ai are Autotech Ventures portfolio companies.)

    

Electronic Shortages Bad & Getting Worse

Hailey Lynne McKeefry

For weeks and months, it’s been a buyer’s nightmare: all sorts of electronic parts are on allocation and prices are on the rise. Everything from capacitors to hard disk drives is in short supply. There’s no end in sight.

The recently released Fusion Worldwide’s Greensheet traces some of the trends and there is little good news to be found.

MLCCs

Multilayered ceramic chip capacitors (MLCCs), a basic staple in many electronic products especially in the telecommunication and automotive markets, are in short supply. That’s a reality that is pushing pricing. The Greensheet explained:

The cost of MLCCs continues to rise and there is a growing consensus that the high prices will remain long after the shortage ends. Passive component makers have long felt that their prices have been too low and undervalued relative to the manufacturing cost.  The rising labor costs in China and the stricter environmental control restrictions will cause prices to rise further, while preventing them from reducing to pre-2016 costs even after the shortage has ended.

Even as component makers are rushing to get more capacity online with new factories focusing on these high-demand passives. However, for MLCCs limitations on materials and production equipment as well as demanding technology hurdles, makes it unlikely that these passives will be part of the shift. However, a number of suppliers are moving their capacity capabilities away from larger case size legacy products to increase volume on smaller case size technology. These shifts take time, though, so capacity likely won’t bounce back any time in the near future.

Other integrated circuits

Currently, lead times on diodes are stretching out across the board. “For example, current lead times are around 30 weeks for TVS Diodes, 24-36 weeks for Schottky diodes and 30 to 40 weeks for Zener diodes,” the Greensheet said. “In comparison, last year around this time, lead times for TVS diodes were at 20 weeks, and small signal diodes and rectifiers were around 18 to 25 weeks. “

Metal-oxide-semiconductor field-effect transistor (MOSFET) prices are likely to also be forced up by increased demand, with increases of 10% to 25% across the market. These shifts can be traced to increased demand from the automotive market as well as a limited supply of silicon wafers and other raw materials needed to produce power MOSFETs “Most industry professionals are estimating that parts won’t be back in regular supply until 2020,” the Greensheet explained.

CPU & memory

Shortages of central processing unit (CPU) chips is ubiquitous, with rising demand combining with supply limitations to create a pinch. “Virtually all segments including Xeons, desktop, mobile and chipsets are facing varying degrees of turbulence,” the Greensheet said. “The strongest number of PC shipments in years have pinched supply availabilities across all client segments.”

DRAMs offer a glimmer of hope, as current demand is currently in pace with supply. At the same time, pricing is also likely to rise again this year although not substantially. In July, market pricing increased higher than last month at about three to five percent, the report said.

 Hailey Lynne McKeefry, Editor in Chief, EBN

Related Stories

EIPC Events view all events

Publications view all publications

  • Proceedings EIPC 50th Anniversary Conference Dusseldorf

    By KWestenberg@eipc.org

    Proceedings 50th Anniversary conference 2018

  • WECC Report 2016

    By KWestenberg@eipc.org

    WECC Global PCB Production Report For 2016 FREE of charge EIPC membership service A Product of the Partnership among the WECC Member Associations on Behalf of their Members in the Electronic Circuits Industry Worldwide

Industry Directory view directory

  • Cicor Group

    www.cicor.com

    Cicor is a globally active group of leading companies in the electronics industry. The group’s companies provide complete outsourcing services and a broad range of technologies for the manufacture of highly complex PCBs, 3D-MID solutions, hybrids circuits, electronic modules, plastic injection molding and box-building.

  • Isola Group

    www.isola-group.com

    Corporate Responsibility. The responsible stewardship of the environment and sustainable business practices is important to long-term success. Committed to operating our business in a manner that creates better social, economic and environmental outcomes for all those involved in the electronics industry.

SITEMAP FEEDBACK

Please provide some more details

FIND OUT MORE

Please login or register

Layer-70

EIPC membership has many advantages, both financially and in services. Such as:

REGISTER NOW

You must be a member to view this content

Layer-70

EIPC membership has many advantages, both financially and in services. Such as:

REGISTER NOW

Please Login / Register

IM NOT REGISTERD TO EIPC
IM REGISTERD TO EIPC

Please provide some more details

Please provide some more details

Thank you!

An EIPC staff member will contact you soon to arrange payment.

Please fill in the form

EIPC will send an invoice to you through email and discuss your yearly fees.