Written by Lucio Lanza | Managing Partner | Lanza techVentures
Many industry pundits contend the Internet of Things (IoT) has all the makings of becoming the next driver of semiconductor industry growth fueled by trillions (yes, trillions!) of smart, connected sensors. These sensors will be able to analyze data at the point of collection, moving artificial intelligence (AI) from the cloud to the edge, also known as the periphery.
This is a grand vision but difficult to achieve given the shortage of experienced engineers. It may become a reality by the open-source movement, however. Through this lens, it’s possible to envision that achieving millions of new designs can be a reality by leveraging open source as an enabler. Fundamentally, this can happen if designs tools are free and easily accessible.
The open-source movement is huge. Design tools are a small but key piece of the movement.
With free tools, we will see thousands of designs being created in the schools and the colleges as a start. Thousands if not hundreds of thousands of individuals will be able to produce a sensor or some other kind of design targeting an IoT application. While these designs may not be unique or revolutionary at first, at least of a few will become a real hit and move to more sophisticated and complex designs for different applications maybe with a more sophisticated design methodology. The pie will get larger even for traditional tools and methodologies.
It’s beginning to happen. The semiconductor industry is at the cusp of experiencing an explosion and democratization of design largely due to the success of open-source silicon. It’s no longer an elite, closed community. Custom silicon is not being eliminated. Instead, the industry is moving from specialized engineering to creative enablement where anyone with inherent skills can get their chip designs into silicon.
The best analogy I can draw from is the move from custom ICs to semi-custom ICs in the 1980s. Despite intense criticism, it didn’t harm the custom IC space at all. In fact, the move to semi-custom design broadened the range of product offerings and engineering expertise. By broadening the design space to include semi-custom, a new community of engineers emerged who weren’t cutting-edge specialists like custom IC engineers. They showed us that they didn’t need to be experts to be creative and successful. It also produced a new market segment.
In today’s compute world, processing has moved to the periphery. Consistent with the fall of regional design clusters, anyone can design from anywhere in the world and centralized design of processing units no longer leads the penetration of electronics into society. These chips often support computing at the edge, changing the dynamics of the industry by removing the CPU from being the center of the processing universe.
And so, the target data moves from being centralized to being on the periphery. Processing is now done where the data is collected and where things are happening. Optimized processors are created for the periphery instead of being centralized on the cloud driven by limitations in power bandwidth to communicate with the outside world. That’s not to suggest there isn’t plenty of work to do to create what is needed at the periphery to make processing happen there. Challenges range from real-time processing and improved data processing to orders of magnitude of lower power, among others. Of course, security is of critical importance with specialized considerations of its own.
Multitudes of clever engineers or just creative individuals will consider applying AI for intelligent intelligence at the edge and will turn to an open-source environment, perhaps the most obvious place to start. With not enough engineers, freely available tools for anyone to use can create all the billions of designs needed for periphery or edge IoT applications. Multiple numbers of individuals can design easily to quickly proliferate peripheral entities. Chips in the periphery don’t need to be overly complex and those creators using open-source tools don’t need to know deeply all the tools and tricks to design that traditional expert engineers know. Or, perhaps another scenario is that an expert engineer uses an open-source environment as a playground for low-cost investigation or experimentation to prove a concept.
One immediate application for the emerging creator segment became obvious through the pandemic ––health care, subjected to unprecedented pressure and challenges. Computer machines that we used previously for tracking medical records and so forth should be significantly upgraded. We must come to rely on things that are thinking as the number of connected users and assets increase.
A complex computer that understands human needs for medical care is at the intersection of computing and medical. Distance is dead and the computing power that AI allows at the periphery will make the concept of health care evolve significantly. A remote AI computational engine at the periphery will ensure an individual is empowered and an illustration of the way medical/health care is moving. It is no longer centralized and information on what we do will be at the periphery but constantly visible to the center. The device will be intelligent and low power with a functioning human interface. Just imagine what a difference it will make for a pregnant woman to have information on her baby constantly visible to her health care professional.
While centralized computers kicked off the medical technology movement (MedTech) and the PC had a significant impact on society, the era is now making things intelligent entities and constant health care a reality.
Of course, these breakthroughs won’t happen overnight and need the system design ecosystem and creative engineers to be involved to peripherize what’s currently centralized. Once that happens, changes should happen fast. When it does, expect significant progress in the next 10 years.
Embracing Open Source
The change may become a reality by open source, a way to harness an enormous amount of creativity that will enable the semiconductor industry to think differently and move optimized, lower power, energy-efficient AI chips to the periphery. Creative and smart individuals, not necessarily engineers, who don’t have millions of dollars or backing from outside investment will come up with millions or billions of designs to power the next generation of intelligent devices. They won’t pay for sophisticated tools to meet the coming trend of low-power IoT peripheral applications and they may do things in unexpected ways. They may, as an example, advance the notion that computation can be done in memory without transferring from analog to digital to processor.
Open source could become the top of a funnel through which designs filter through. Not all will be successful. Some will succeed or just evolve and be redesigned using more sophisticated tools for complex, highly optimized designs with high-end applications and massive computing needs. These will, in turn, create increased revenue growth for the semiconductor industry.
The open-source movement has the potential to create an enormous number of designs made possible by many, many individuals, though it is not without doubters. And yet, no one disputes the need for more designers. The open-source movement will generate more creators (players), a huge number who will slide through the funnel and become engineers (winners).
While my position on free and accessible design tools is firm, I’m not taking a stand against traditional and successful EDA companies and their business models. My goal is to ensure that we as an industry support the next big growth opportunity for chip design. Successful companies may have momentum now and view the open-source movement as a distraction. When a new momentum happens and in a different direction, ignoring it may, at best, result in a missed opportunity or worse.
Semiconductor companies doing new design starts should welcome the opportunity to minimize the cost of the design and dependence solely on expert engineers. With the prospect of the number of IC design starts growing by a factor of a thousand, the industry needs to find a way to multiply the number of engineers. Open source may be the answer. That’s how a non-technical engineer will create the next breakthrough open-source chips. In fact, many may not know they designed a chip that could one day be the basis for a chip in every consumer electronics product. To do so, it will take millions or billions of designs, not hundreds to get the exponential growth necessary for smart connected IoT devices to operate autonomously and sustainably. This will move the intelligence to the periphery, where it belongs to be effective in significant improvements to society.
About Lucio Lanza
Lucio Lanza is the Managing Director of Lanza techVentures, an early-stage venture capital and investment firm, and the 2014 recipient of the Phil Kaufman Award for Distinguished Contributions to EDA. After executive positions at Olivetti, Intel, Daisy Systems and Cadence Design Systems, Lucio joined the venture capital industry in 1990. He holds a doctorate in electronic engineering from Politecnico in Milan, Italy.