Executive Interview: James Thompson, Executive VP of Engineering and CTO at Qualcomm
Qualcomm has a long history of innovation, beginning with your founders’ vision for CDMA technology. Describe the company’s culture and the process that sustains your innovation.
CDMA was a complete reimagining of the cellular system where, instead of assigning each individual user a frequency, all users shared the same frequency and were assigned a code. This increased the “spectral efficiency” of the system by about a factor of 100, which ultimately led to dramatic reductions in the cost of service. The cell phone went from a symbol of the rich and powerful to a transformational device that about 70 percent of the world's population benefits from today.
The early culture of the company highly valued deep technical competence, cross-functional teams with a wide range of technical skills and a system-level collaborative approach to solving complex problems.
Another example of the power of our culture was our invention of the mobile internet. During the time of our CDMA development, we were deeply involved in the internet and recognized how important data would become. With that vision, we created EV-DO, an all IP wireless network that benefited from many of the CDMA concepts and learnings, along with our significant internet expertise. At first it was met with skepticism, but ultimately it was embraced and became the first big step toward the mobile broadband we enjoy today.
As we have grown into a global company, it has required more effort to maintain the culture that led to that level of innovation. We have expanded well beyond communications systems into many other technology areas; in every case, our objective is to be the leader.
An interesting example of that is digital image processing. We started off with the simple idea of integrating a camera into a cell phone, so you didn’t have to carry two devices. We also believed that a camera integrated into a cell phone SoC would benefit disproportionately from Moore’s law, giving it a cost, power and processing advantage. We started off with two smart and motivated engineers from our CDMA team working on the problem. Today, we have a large team working on camera technology and are very proud to have the best image processor in the world.
We’ve followed this model of building technologies organically over and over again with great success. In each of these teams, Qualcomm’s innovative culture continues.
What are Qualcomm’s core competencies today?
The roots of the company are in communication theory and RF design. We led the mobile industry early on in the 2G digital transition and, today, continue to lead in 5G. We are modem people first and foremost; along the way, we have developed a wide range of additional technical competencies.
For instance, early on, we created an email client called Eudora, which became the most popular email client in the world. Having this type of capability within the company led to our initial pursuit of wireless data. And as wireless data gained traction in the market, we pursued all manner of digital processing so that we could integrate many consumer products into the smartphone.
Most of what you love about your smartphone, Qualcomm developed. As a result, we have deep expertise in all types of multimedia, graphics, display processing and computing, as well as expertise in developing very complex leading-node heterogeneous SoCs and, perhaps most importantly, the software that ties it all together.
There are very few companies that have the breadth of leading technologies that Qualcomm has, but I think it is our ability to apply those technologies in a complex system that is most unusual and is our biggest advantage.
5G has been a tremendous success for Qualcomm: you have achieved leadership in mobile chipsets and pushed accelerated standards development and operator deployment. Share Qualcomm’s strategy and the key milestones that led to this success.
At Qualcomm, we take a system approach to inventions, from early investments in advanced R&D, system prototyping and simulations, to working with industry leaders to push these new technologies into standards and conduct interoperability testing. Finally, we drive commercialization of these new inventions by building industry-leading products.
For 5G, this all culminated in the 5G moment at Mobile World Congress 2019, when we, along with our partners, announced to the world that 5G had arrived. We brought in a number of key players across the ecosystem to help us get to that moment, with our strategy to invest early in advanced technologies and to work with partners that can help contribute to the success of 5G.
Qualcomm has been a strong advocate for using mmWave spectrum to increase 5G data rates, and you have released three generations of a pioneering beamforming antenna module for the smartphone. As you watch the initial use of mmWave by operators and phone manufacturers, what are you learning about the end-to-end performance, both pro and con?
Initially, there was a lot of skepticism in the industry around mmWave — even at Qualcomm, we knew it was a huge challenge. We also knew mmWave could deliver extreme capacity and multi-Gbps throughput, but there were significant roadblocks to making it work in the mobile environment with smartphones. Mobilizing mmWave was widely regarded as an impossible challenge throughout the wireless industry. We started over a decade ago and spent a lot of time designing the end-to-end mmWave system to overcome the roadblocks. Challenges include phased array beam management and beam tracking, taking advantage of multipath, managing maximum human exposure limits, power efficiency and cost.
This is an area that is very open for innovation, and performance will depend a lot on the quality of the implementation. Operators are just starting to commercialize mobile mmWave, proving our early vision that mmWave can indeed be used for mobile broadband. Although we are only at the beginning of commercial deployments, we are already seeing success with impressive Gbps speeds.
Challenges for widespread deployment of mmWave will be coverage and cost of an individual cell. With regards to coverage, in many cases, base stations are mounted to light poles, but not every light pole has a fiber backhaul. Therefore, there is an effort in the 3GPP standards to create a standard for a repeater-like base station design that simultaneously serves phones and connects to an adjacent base station that is directly connected to backhaul. From a cost perspective, a mmWave base station should be less expensive and less complex than a phone.
Last summer, Qualcomm acquired full ownership of the RF360 joint venture with TDK, which gave the company all the RF technologies used in the mobile front-end. What is your strategy for RF?
With this acquisition, Qualcomm Technologies is able to provide customers a complete end-to-end solution from modem to antenna — the Qualcomm Snapdragon 5G modem-RF system — as well as the world’s first commercial 5G NR sub-6 GHz and mmWave solutions, integrating power amplifiers (PA), filters, multiplexers, antenna tuning, LNAs, switching and envelope tracking products.
Now that we can give our customers an end-to-end solution, we believe we have a competitive advantage, because we can make improvements at the system level that improve performance beyond what can be done by just optimizing components.
For example, because we know the envelope of the modulated signal hitting the PA, we can adjust the bias of the amplifier to keep it in compression as the envelope of the signal changes. We can also dynamically compensate for the nonlinearity of the PA digitally in the baseband.
That’s just one example of many. As the complexity of RF goes up, combining it with digital processing offers more and more opportunities to improve.
With in-house RFFE (RF front-end) R&D, we’re also able to achieve breakthroughs at a more fundamental component level. For example, with our thin film technology, the new Qualcomm ultraSAW filters achieve performance previously thought impossible in surface acoustic wave products, and it’s now surpassing BAW performance.
The breadth and depth of expertise in RF and baseband puts us in a strong position to develop comprehensive, global modem-to-antenna 5G/4G solutions for all tiers, while also being a reliable supplier for non-Snapdragon platforms.
Is there an engineering solution to ensure the security of 5G networks, regardless of the equipment supplier?
Security is designed into the 5G standard, which increases the security of all devices on the 5G network and, indeed, the network itself. We have been working on fine-tuning this over the past three to five years, and 5G has been the most secure out of all the iterations of networks.
When we look to the future of 5G as it transforms adjacent industries, it will be critical to continue to ensure the security of all devices on all networks. Although 5G offers enhanced security over 4G, there is no substitute for endpoint security in devices and the cloud. Both are required.
You talk about the marriage of 5G with AI and cloud computing. Paint a portrait of what impact that combination will have on society.
We are going through a digital transformation of almost every aspect in our lives, from conveniences in the home and car to optimizations of business processes to efficient management of the infrastructure. These changes are happening because we can sense the world around us and represent it digitally as data. AI is a powerful tool to help us make sense of all this data.
Where we process the data depends in large part on what is important about the data and how much is generated. For example, an autonomous vehicle will require almost all computations be done by the vehicle, for reasons of immediacy, safety and the massive amount of data generated every second. The cloud will manage changes to precision maps and traffic information, as well as collect high-entropy sensor data from the cars to improve the AI models. So, AI will be used both in the cloud and in the device, with 5G providing the data connectivity required to make the system work.
For the case of city infrastructure, we will see some low bandwidth sensor data processing, consolidated directly in the cloud, and some high data rate, low latency processing — like at a traffic intersection — happen in what we call the “edge cloud.”
In other words, servers close to the application for latency and/or security reasons.
In all of these digital transformation examples there are two key elements that make them possible:
- AI’s ability to consume and make sense of data and the data connectivity capacity.
- The low latency that 5G networks will provide.
I believe this will have a huge impact across many industries as we make our world more efficient, convenient, safe and secure.
The Qualcomm name is the conjunction of “quality communications.” Historically, you have been a communications technology company, although you are applying your capabilities into adjacent applications, such as automotive and AR/VR. What is Qualcomm’s evolving mission?
The 20-year evolution of our smartphone roadmap has given us a wide range of internally developed technologies, ranging from 4G/5G, Wi-Fi, Bluetooth, audio, video, camera and computer vision to displays, graphics and DSP processors and AI accelerators. Mobile is still the primary driver of these technologies and affords us many opportunities for investment.
It turns out that these technologies also apply very well to other markets. In automotive, more than 50 percent of cars in the developed world are wirelessly connected, offering real-time traffic information, software upgrades, streaming content and emergency services. Rich infotainment and digital instrument clusters are becoming common in vehicles.
To meet the performance requirements of automotive, we only have to make minor changes to our mobile designs, allowing us to provide incredible capabilities for a relatively low cost, because of our mobile investment. This is also true for AR and VR. We make relatively minor changes to our mobile processor so it can be used in a low power head-mounted display for VR. Both of these businesses are expanding rapidly at Qualcomm.
I’m also excited about other leveraged products like our cloud AI inference engine. We designed our mobile inference engine to be scalable, so it could support cloud workloads. Our mobile design focus is low power and that’s exactly what is needed in the cloud for AI inference.
We also have many other leveraged products in the IoT space that take advantage of both our compute and communication assets. More fundamentally, we believe the world is going through a digital transformation where a combination of sensors and communications technologies like 5G, plus computation capabilities like AI, will lead to massive efficiencies in many industries.
Driving these changes is our mission.
If not the largest employer of RF engineers, Qualcomm must be one of the top two or three. What’s your view of the global pipeline of RF engineers, and how does that shape your efforts to recruit and develop a diverse workforce?
Qualcomm’s RF engineering team is spread out over several continents, developing leading edge wireless transceiver and RF front-end technologies for cellular and connectivity. We actively recruit from top universities worldwide at the master’s and doctorate levels and have a strong intern program in which we engage early in students’ academic development.
We believe a diverse workforce makes for a better work environment. This positive work environment, in turn, helps us attract the best talent into an RF engineering career and into Qualcomm. It is no secret that RF engineering suffers generally from a lack of diversity in students, especially female and minority students. We try to address this challenge at the source through our Thinkabit Labs across the country. Our objective is to inspire those underrepresented groups to get engaged in STEM early. Over 65,000 kids have participated. One may consider this a small effort, but many small efforts can lead to big changes.
What led you into science and engineering? Tell us about your journey to this role as CTO.
I grew up in Wisconsin, practically on the campus of the University of Wisconsin-Madison, because my dad was a finance professor there.
He was always really into technology and computers. He was a mathematician in the early days of vacuum tube computers before pivoting to finance. The irony is, at that time, he figured he had learned everything about computers and wanted to move onto something else.
His interest sparked my passion for computers, as well as math and physics. So when I went to college, I started taking math, physics and computer science. I really enjoyed physics, in particular. Eventually, physics led me to electrical engineering, because there are parts of it that are very physics oriented, and I was really attracted to that.
I eventually took a job in the Bay Area as an RFIC designer. I learned a lot, but had always known I had wanted to get a PhD, because I had either wanted to be a professor like my dad or at least have that option. After earning my PhD at the University of Wisconsin, I ultimately decided I didn’t want to become a professor, so I joined Qualcomm when it was a very small company and started working on the CDMA project.
It was an amazing experience seeing this small group of people transform an industry. After the CDMA project, I joined the Globalstar program and eventually became the engineering lead. This exposed me to almost every technology and function Qualcomm had: RF, digital, software, communications, networking, budgets, HR. After Globalstar, I moved to our semiconductor group to lead hardware design and systems engineering and, shortly after, took over leadership of the semiconductor engineering group. It was relatively small at the time, but it grew dramatically, becoming much of what Qualcomm is today. Going through all the successes and the occasional failures of the semiconductor business gave me a breadth of experience that is pretty unusual.
About three years ago, I was given responsibility for the rest of engineering at Qualcomm, which includes our wireless research and standards teams and our engineering support groups. Since then, I’ve put a lot of emphasis on making a strong connection between our research teams, which focus on 5 to 10 years out, and our technology teams, which have more of a product focus. I’ve also built up our AI technology presence, connecting every technology team in the company with a centralized AI team that focuses on energy-efficient computation and algorithms. This is making many things in technology we thought were impossible, possible.
Any career advice for engineers to sustain their technical careers?
A few thoughts:
- Most things worth doing take more than a few people to do. Teamwork and collective brainpower can make the good great. Follow the Golden Rule in all your interactions and collaborate openly with others. Always assume others have good intentions.
- Don’t ask permission to do things. Just do them if it is the right thing to do. The best people don’t need to be told what to do.
- It took a long time for me to realize that the world is very uncertain, and it is okay to not know everything. In fact, the people who are most successful deal with uncertainty the best. Embrace it.
- Don’t over plan. If you are doing something complex, things always go wrong. Get started, learn, course correct and repeat.
- Be optimistic and persistent. People want to be led by someone who is positive and won’t give up.
- The trust of your coworkers is hard to earn and easy to lose. The best way to gain trust is to work in your “enlightened self-interest.” Do what is best for your project and your company first and yourself last. The most effective people in a team environment follow this approach.
- Learn the jobs around you so can you do your job better.
- Never stop learning.