Reflecting Back on 2016: Markets

How last year’s predictions panned out over the past 12 months.


Anyone can make a prediction, and sometimes the more outlandish they are the more they get noticed. But at the end of the year some people hit the mark while others may have been way off.

Many people simply make projections based on the current trajectory of trends, while others look for the potential discontinuities that may lie ahead. Semiconductor Engineering examines the projections made at this time last year to determine who got it right and who didn’t. All participants have the option to comment on their own predictions. Of 17 companies that made predictions, seven of them chose to judge themselves.

Looking at the semiconductor industry in general, Gartner predicted that semiconductor revenue was expected to increase 1.9% to $344 billion in 2016. It also forecasted an oversupply in DRAM in 2016, with DRAM revenue expected to decline 12% in due to weak pricing.

In October, Gartner revised its numbers to reflect a decline of 0.9% for the year, with revenues topping out at $332 Billion. As for DRAM, prices, they did decline about 20% early in the year, but they have since stabilized. In addition, the oversupply situation appears to have been resolved. With that, prices have recovered somewhat.

With all of the merger activity in 2015, many were looking at R&D spending levels. Graham Bell, then-vice president of marketing at Real Intent, believed “there is a source of concern for EDA companies because their revenue growth typically lags R&D revenue growth by six to nine months. With the expected consolidation, there will be fewer companies purchasing EDA tools. This will benefit the Big Three companies, and will put additional pressure on Tier 2 and startup companies to differentiate their offerings and maintain product pricing.”

, CEO of Breker, responded. “Consolidation continued unabated through 2016, and it is becoming very clear that this is driven by the need to reduce costs. There has been a realization that there is a lot of wasted effort in verification and validation, and this is one area in which semiconductor companies are actively looking for, and driving solutions. We are seeing this in the adoption of solutions and in the energy levels that users are putting into the Accellera standards group.”

Hamid observes several other changes underway in the semiconductor and EDA industries. “These are both maturing industries, and that means that reducing costs tends to be a focus. This is different than in the past when engineers defined the ways in which things would get done. Today, it is less about that and more about addressing the need to get it done, out of the door, and finding the cheapest, fastest way to do it. Another problem being faced by a maturing industry is the ability to attract talent. This is seen in its most stark form within EDA.”

He said the EDA engineering talent pool is aging, with top technology graduates migrating to companies like Google and Facebook. While there are still pockets of innovation and entrepreneurship, attracting good engineers is tougher.

The automotive category received the most submissions for 2016. “The traditional industry is under a lot of pressure from companies such as Tesla,” said Chi-Ping Hsu, senior vice president, chief strategy officer for Cadence. “The gas engine versus electric represents a fundamental change. A pure electric car is simplified, is a lot more software driven and uses more electronics.”

Marc Serughetti, director of business development for System Level Solutions group at Synopsys, pointed to one area in which he expected to see a lot of progress in 2016. “ADAS is a prime example where fast-paced development is taking place. But this development is not coming without development challenges. In particular, functional safety is challenging the traditional development approach.”

Serughetti responded: “There are some clear proof point in the market, and there is a race towards ADAS in automotive. Semiconductor, software companies, automotive companies are all deploying significant efforts to get there.” He points to several examples of progress companies such as NVIDIA, NXP and Qualcomm, which have announced new chips or directions. On the software side, machine learning is driving several companies in this area (including some of the semiconductors), he says, noting that OEMs, including Tesla and Ford, have been pushing hard.

Dave Kelf, vice president of marketing for OneSpin Solutions, added that “ISO 26262 and other regulations will become important for many companies not usually associated with automotive as they all get in on this market. In turn, these regulations will drive a new awareness of high-reliability design and verification, creating a renaissance in fault injection techniques as part of the verification flow. This in turn will drive a verification strategy for these devices that places formal verification ahead of simulation for the first time.”

Kelf responds, saying “This was spot on. In 2016, the automotive semiconductor market became a primary target for many companies, both traditional automotive and new players alike. An awareness of the verification requirements, driven by regulations such as ISO 26262, swept through these companies, most of which had to step up their game in this area. The verification of random fault handling mechanisms, using fault injection by leveraging either traditional fault simulation or more modern formal verification, has emerged as a key new requirement.”

The impact on EDA also was discussed by Serughetti. “The industry is realizing the need to evolve their electronic development process, and several leading automotive companies from semiconductor to Tier 1 and OEM, are leveraging the power of prototyping internally and through the supply chain. The trend will be to increasingly depend on virtual prototyping from semiconductor to Tier 1 and OEM to collaboratively design and develop automotive systems from SoCs to ECUs to vehicles.”

Serughetti responds saying, “prototyping use has been growing, although this is less visible in the press because there is less talk about design and development methodologies. However, we have this visibility from Synopsys and it is clear that this is a path forward.” Serughetti notes that within Synopsys there are several examples of new virtual prototypes becoming available for vision SoCs, control MCUs and others.

Serughetti also responds to the larger trend with ADAS. “There is a significant investment in electronics for automotive, and ADAS is one of the drivers. This is now a proven technology and is ongoing although automotive has a longer development and deployment cycle than more traditional mobile/consumer markets. Because of this, development methodologies have to evolve, and prototyping is used to address the changes needed.”

There were also some fatalities in 2016 related to ADAS and automated driving. Charlie Janac, president and CEO for Arteris, comments that “as predicted, 2016 showed large investment in automotive semiconductor technologies driven by development of the automated/self-driving car, the electric car, the over-the-air automotive software updates, and the consolidation of automotive microcontrollers into SoCs. In automotive, functional safety features such as resilience are supported by ISO 26262 qualification. The first automated driving fatalities showed that such a qualification is nothing to fool with, and is necessary to save lives and provide reasonable liability protection for the automotive supply chain.”

Frank Schirrmeister, senior group director of product management in the System & Verification group of Cadence agrees, saying “from a verification perspective, automotive has become a prime market. 2016 marked the year in which we rolled out the first comprehensive TCL1 documentation to support the ISO 26262 standard. This really means that we document how to use our tools for automotive and the ‘tool confidence level’ is confirmed.”

The Internet of Things
Another area in which a lot was expected for 2016 was the Internet of Things (IoT). Most of the industry was expecting the IoT bubble to burst, given the level of hype attached to this term.

Brian Derrick, vice president of marketing for Mentor Graphics, expected that “new opportunities will emerge in four areas of the IoT architecture: sensing and actuators (the nodes), networking and data communications (gateways), embedded software for the nodes and gateways, powering the nodes and gateways, and cloud-based applications harvesting the vast amounts of new data.”

Schirrmeister responds that “the IoT, for which some predicted a bursting bubble, is certainly here to stay and has been refined as an overlay to application domains. IoT in industrial and medical applications, as well as consumerized-medical applications addressing fitness issues, are driving IoT applications. Two sub-trends have become very interesting in 2016. First, game-ification of IoT apps is a driver. For example, at Cadence we competed with each other on step counts for 100 days for health benefit cost savings in 2017. Second, there is a value shift from pure semiconductor value to systemic value in IoT applications. The edge-node sensor itself may not have made a great contribution to profits. However, the systemic value of combining the edge node with a hub accumulating data, and sending that data through networks to cloud servers where machine learning and big data analysis happens, allows for cross-monetization. The value definitely is in the system.”

Breker’s Hamid says that some aspects of the IoT are not that clear yet. “It could go one of two ways. The first is that there will be lots of little chips, each with a different set of sensors. To enable this, it will be necessary to build a hundred tiny chips and pack them all onto a wafer. Alternatively, there will be just a few large semiconductor companies that create IoT superchips, which can have the necessary features enabled. The only thing that really differentiates these chips are the sensors. We don’t want to do a lot of compute on them because that will consume too much battery power. However, by far the largest and most profitable part of the IoT will remain in the software that uses, aggregates, and distributes the content provided by the sensors.”

Hamid also looks at the impact that IoT may have on tool development. “The IoT will have two impacts on the EDA industry. The first is that these devices are more about software than hardware. This means that Portable Stimulus will have to move up to help verify the software as well as hardware. It will also have to have the ability to simplify and migrate software tests down to something that can be handled in simulation. The second impact is that these chips are small enough that they will fit into an FPGA, and so will not produce demand for emulation.”

Ron Lowman, strategic marketing manager for IoT in Synopsys, added a couple of technologies that could fuel many expanding markets, including automotive and IoT. “Vision systems will become increasingly adopted to add value for robotics, safety and security applications. In addition, voice recognition will improve significantly and be viable in small smart home products.”

Lowman responds that “this has been seen with the continued popularity of Amazon Echo, Dot and other like products. Amazon expects over 40 Million units to be sold by 2020, so voice recognition is here to stay but now will be implemented on more smart objects.”

While nobody made predictions directly about smart cities and the Industrial IoT, it is another area gaining traction. Lowman says “the benefits in operational efficiencies are clear for investments made in the industrial internet, but the numbers shown for smart city investments are rather large. These investments for Smart Cities have been a welcome surprise.” An example of this is the announcement of an addition $80 million of new Federal investment announced by the White House in September.

Part two of this series, which will be published in a couple of weeks, will look at predictions made in Semiconductors, Design Methodologies and Tools.

Related Stories
New Wave Of Consolidation
What’s driving it, what’s next, and who should be concerned?
Sensors Enable ADAS
Advanced driver assistance requires a careful and complex balance of hardware, software and security.
Tech Talk: ADAS
Why the Advanced Driver Assistance Systems standard is so important and where the potential pitfalls are.
Tech Talk: ISO 26262
What can go wrong in designing to this automotive standard.
Smart Cities, Challenging Issues
Research points to massive savings and better quality of life, but getting there isn’t so simple.

Leave a Reply

(Note: This name will be displayed publicly)