Offerings of novel computational architectures, in the form of machine learning and open hardware, mark the 18th rendition of EE Times' Silicon 60.
EE Times' 18th major revision of its Silicon 60 listing of emerging startups and companies is marked by a key trend — offerings of novel computational architectures, in the form of machine learning and open-hardware.
Startups offering such things had been brought into the Silicon 60 in previous years but there are several representatives this year in a "gold rush" of players trying to gain a foothold in what now looks likely to be the next big thing.
The technologies covered in this year's Silicon 60 include silicon and compound semiconductor manufacturing, analog and digital ICs and SoCs, memory, FPGA fabric, gallium nitride for power and for lighting, energy harvesting, photonic circuits, sub-threshold voltage operation of ICs, signal processing techniques, 5G communications, communications for automobiles and IoT, 60-GHz for short-range communications, wireless power transfer, techniques for gesture recognition and haptic feedback, image sensors, MEMS, environmental sensors, neural networks and other architectures for machine learning, for vision and cognitive processing.
As in other years, about half companies have been retained from the previous list. Some 29 companies have been brought into version 18.0 (see EE Times Silicon 60: 2017's Emerging Companies to Watch). That compares with 25 admitted in 2016 and 30 brought on to the list in 2015.
It also brings the total number of companies admitted to list to 425 since v1.0 came out in April 2004.
It is important to note that, for the purposes of compiling this list, we take a relatively narrow view of the meaning of technology. In the world-at-large the term 'technology' is now used to refer to any product or service that is electronically or digitally enabled. And often that is about how some straightforward software and/or Internet services have been applied to new application areas.
For the purposes of the Silicon 60 we narrow the focus to include on our list companies that could have a major impact on our readership of engineers and managers in their professional lives. And that means we usually seek out companies that have one or both of their feet in the hardware camp. It is true that increasingly hardware companies must be "platform" companies that offer both hardware and software but those companies that are pure software or service plays — with the exception of EDA and hard and soft IP suppliers — are less likely to be included.
It is also important to remind readers that we do not claim to be tipsters or that the companies on our list are inevitably destined to go on to great financial success. We do think these companies may be interesting to watch for a variety of reasons not least because they are pioneers of a technology or market.
And in 2017 the technology changes are plentiful, perhaps driven by the apparent increased cost of further transistor miniaturization and the concomitant slowing of Moore's Law. As the cost of leading-edge production increases it has made sense for entrepreneurs to try and find ways to create value in circuit design. At the same time the complexity enabled by past progress in miniaturisation has provided great scope for alternative computer architectures. Both entrepreneurs and venture capital appear to have recognised this.
To judge by our Silicon 60 'radar' list — an internal compendium of startups from which the Silicon 60 is selected — the years 2013 through 2016 were exceptionally strong in terms of startup formation.
This is perhaps because in the early part of that cycle established companies were not hiring, making it a bit easier for individuals with an entrepreneurial bent to go it alone. In the second part of the period companies were hiring for R&D — markets were just starting to pick up and companies, entrepreneurs and investors recognized it was time to invest for the future.
In contrast, 2017 has been a relatively quiet year so far. One could call it the beginning of the next phase of the cycle — the phase of execution. But then again many of the companies founded in any given year may not send a ping back to our radar until they are ready to announce seed or Series A funding, which can be a couple of years later.
If the Internet of Things was the trend in 2016 then machine learning and open hardware in the form of companies building to the RISC-V standard is the trend for 2017. Previous trends such low-power design, IoT, non-volatile memory, MEMS and sensors, drones and automotive driving assistance systems (ADAS) haven't gone away. They are vibrant and mutually supportive.
In version 18.0 of the Silicon 60 there are six startups with machine learning in their approach to the market and three with a RISC-V approach and a lot of those companies have pocketed good sized tranches of venture capital. Complex digital hardware that can use billions of transistors manufactured in a leading-edge process is expensive. And that exemplifies the return of venture capital to hardware after a decade of curtailed interest.
Although the largest sums of capital seem to have been sent towards some of the solid-state lidar startups on our list. Quanergy Systems Inc. has raised approximately $150 million and Innoviz Technologies Inc. about $74 million. Again, there is an urgency in this financial response. In this case to the perceived inevitability of autonomous vehicles.
Next page: Geographic trends and re-rise of Japan