Accordion

• Transcript

  Melissa Harrison (00:07): 

  Welcome to another special Inside Innovation episode of CES Tech Talk, where we'll dive into semiconductors, the small chips that power the world's most advanced AI and tech, and how the latest innovations in analog and embedded processing are making things faster, smarter, safer and more efficient. I'm Melissa Harrison, and I have to say, this conversation feels a little full circle for me. Back in high school, I spent countless hours with my TI85 graphing calculator, never imagining that one day I'd interview the Chief Technology Officer of Texas Instruments. 

  (00:41): 

  Joining me today is Dr. Ahmad Bahai, Senior Vice President and Chief Technology Officer of Texas Instruments, and Professor of the Practice at MIT. Before that, Dr. Bahai was a consulting professor at Stanford for more than two decades and a Chief Technology Officer at National Semiconductor. He also holds a master of science from Imperial College London, and a doctorate in electrical and electronic engineering from the University of California, Berkeley. Dr. Bahai, welcome to the podcast. 

  Dr. Ahmad Bahai (01:10): 

  Thank you. 

  Melissa Harrison (01:12): 

  I am so excited to talk with you today because you have called semiconductors the infrastructure of innovation, yet most consumers and probably our listeners never think about them. Your team frames this perfectly around three themes. I really love this, how we move, live and work. From your perspective as the CTO, how do you explain the fundamental role that chips play in enabling the innovations that dominate all of today's headlines? 

  Dr. Ahmad Bahai (01:40): 

  Yes. Thank you very much, Melissa, for a very good set of questions and of course for your introduction. As you mentioned, I've always referred to innovation in semiconductor as a foundational to almost every aspect of our lives and every aspect of the industry in the sense that we have come a long way since transistor was invented several decades ago. It's been amazingly impactful in the sense that a lot of innovation has gone through this to make the performance so important and so impactful that now you have the power of a supercomputer in your phone. The supercomputer that used to be millions of dollars, you can have it at your disposal and your phone. And on everything around you, when you get into your car, all the electronics and the features that you see on your car. At your work, not only the laptop and computers, but the behind the scene intelligence that is driving all these advanced machine learning algorithms and search and interconnection of the networks and the way that we are interfacing with our machines, displays and everything around us are based on innovation in semiconductor. 

  (02:57): 

  I would say that what's amazing about semiconductor is almost in a way managed to go against the inflation. The technology has gone so fast and so efficient that we see the semiconductor is so affordable today after many decades of innovation at the age of the technology. We don't have time to get to the details of how semiconductor devices and chips are made, but there are massive amount of technologies at a very small scale. We are talking about micro and nanometers of designing these devices and making sure that billions of these devices work together. These innovations manifest itself onto almost every aspect of our lives and things that we are using every day. 

  Melissa Harrison (03:46): 

  It really is truly amazing to think about the fact that they are in almost everything we're using. When we go to CES, we see these dazzling 8K displays, we see autonomous cars, we're going to see humanoid robots, but you're really building the infrastructure that enables all of that to be possible. What is the one technology headline that you're looking for at CES or that you've seen at CES where you think to yourself, wow, people have no idea just how complex the analog and power management innovation was to actually make this one thing work? 

  Dr. Ahmad Bahai (04:22): 

  Yes. It's hard to highlight just one technology because many of them are equally exciting. As we talked about, these smartphones are putting the power of supercomputer in your pocket or capabilities of amazing image recognition and video recognition and transfer of data from one point to the other in small phone. That is as exciting as you see a large high performance display with the high capabilities, quantum dots and many other innovations in semiconductors. But one thing that is amazingly common behind all of these things is the fact that there is a lot of intelligence in these data centers to provide the data, upload and download data and analyze data to offer a more efficient and more intelligent capabilities across these devices. And data centers are pushing every aspect of technology to the limit. I would say behind this scene excitement of the data analytics and infrastructure that supports the data analytics is the most amazing things that's happening today, our time and so critical for everything we talk about. 

  Melissa Harrison (05:39): 

  As part of your role as CTO, you also steer all the long-term R&D at Kilby Labs. And for our listeners who may not know, that lab is named after Jack Kilby who actually invented the first integrated circuit. So what a legacy opportunity to lead that R&D, but the universe is so vast. I mean, how do you decide which technologies to pursue when product life cycles may span decades, but market trends, generative AI, everything can shift overnight or at least feel like it's shifting overnight. 

  Dr. Ahmad Bahai (06:13): 

  It's a very good point, and a very interesting challenge for us as researchers. Obviously, nobody can predict the future. Nobody can tell what technology is going to dominate the market several years from now, even a couple of years from now. But for sure, we can see the trends and that's where really we can get inspired by the trends that we see in the technology and market. For example, we know that power density is going to be always almost there is an insatiable appetite for more power density out there. We need to deliver more power in the most efficient way. Now, one day that could be powered from a battery to your phone. Next day could be a power from grid to a data center. We know that we need a faster sampling and translating analog signals to digital. These high performance data converters are in demand. 

  (07:10): 

  Now, one day it could be radar, next day it could be in a medical instrumentation. If you look at these trends, you need to push the performance of many of our technologies, almost all of our technologies to the next level to make sure that we can address all the needs and all the technical requirements of the future technologies. And on the other hand, there are some foundational technologies that you need to work on. These are devices that can offer higher power density. These are devices that can offer faster and more efficient RF or high speed links. Now, again, one day it could be for application A, next step for application B. 

  (07:51): 

  A great example of that is in Texas Instrument Research Lab, we started working on a new device called gallium nitride, and the idea was that how this device can offer much higher performance power delivery. Our target was mainly electric vehicles because that was a growing market, but also the same technology now is equally applied for power delivery and data centers. These are foundational enough that you can react to different markets by, of course, with different set of requirements, but the foundations are very critical to build. 

  (08:27): 

  To summarize is that the way that we are doing R&D and research, we are looking at the disruptive opportunities can impact the trends on the power density, on data communication, on sampling of the analog signals and so on and so forth, and then react to the market needs as new markets start ramping up and serving multiple markets, hopefully with these foundational technologies. 

  Melissa Harrison (08:51): 

  Yeah. You mentioned one of the foundational technologies being in EVs, and one of the themes that your team focuses on is move, so I'd like to move into the move theme a little bit. I think everyone in the industry is really thinking about the software defined vehicle. And from a hardware perspective, what does that transformation require and how is TI helping automakers enable features like over the air updates and a more personalized driving experience? 

  Dr. Ahmad Bahai (09:21): 

  The goal in TI is to make the electronics more affordable across all these technologies. And when we hear about software defined, I mean, it used to be software defined radio and still is there, software defined written vehicle, these are really great concepts. But at the end of the day, software defined needs to run on a hardware. I mean, the question is, okay, we are dealing with a processor, we are dealing with a power delivery to that process, or we are dealing with interconnection and connecting the different devices together to make it more intelligent. Software defined devices, in fact, push the hardware requirements even further. 

  (10:03): 

  For example, when people talk about software defined radio, the requirement for data acquisition and data conversion was much more demanding than the traditional way of designing radio. Same thing for vehicles. As you need a mechanical vehicle, no matter how much of the software defined, you still need to have wheels and engine and the doors and seats. Also, you need massive amount of hardware to support that software. And now that software defined radios are relying on massive intelligence of many giant processors, then all the topics that I mentioned about the data center in one way or the other is going to be in a car, how to deliver so much power, how to interact with the devices, what's the human machine interface, how these things talk to each other? 

  (10:50): 

  What's the connection and connectivity link between these things? All of these are massive hardware behind the scene, which goes back to your first question that you don't see all the infrastructure behind the scene, but you see the user interface that is leveraging all those capabilities to offer a very intelligent, a highly interactive platform for a driver or in some cases even driverless machines. 

  Melissa Harrison (11:19): 

  Well, I know that beyond everything you just mentioned that TI is also heavily involved in advanced driver assistance systems. How are you using sensing and processing technology like new radar sensors just to make cars safer, more aware of their surroundings? 

  Dr. Ahmad Bahai (11:36): 

  Yeah. I mean, these ADAS or driver assistance system are fascinating in the sense that they need to be intelligent enough to almost emulate a human driver. So that means you need to have the vision to see what's going on around you. And the vision can be supplied through many cameras, radar, which can go even in the dark or can see objects that cameras may not be able to see. And in many cases, even LiDAR. We see that many of these technologies come together to offer this 360 view of what's going on around the vehicle. And on top of that, as you know, when you're driving fast, you need to react very quickly, so you cannot rely on sending the data or all the data to the cloud and get it back. You need to do a lot of local processing. That's what we call it, a great example of edge AI. 

  (12:28): 

  It's a combination of the multi-modality of sensing, including radar and many other visual sensing, but also massive processing, fast links that can really bring data to the processor for quick turnaround time and decision making, and of course, power management and everything around that. So we see that ADAS is an area that really tries to emulate the awareness and the capabilities of a driver, hopefully even at higher level in the sense that you can see that maybe a naked eye cannot see, but then it means that you have to process this massive amount of data in a very quick low latency platform. We think a wide range of technologies in TI from the radar itself, which we have been working on for quite a while and we think we have a very compelling solution to data links between radar and processors or camera and processors to processor itself and power management around processors, we have a lot to offer and we have been working on these technologies for quite a while. 

  Melissa Harrison (13:35): 

  It's truly incredible, and if it can sense faster than I can tell my husband he should be breaking, I would really appreciate that. I think it would help our marriage. I wanted to ask you about EV adoption. There's a lot of talks, especially here in the US around the catchphrase charging anxiety. It seems like it's the current buzzword, but TI is really working in this space to power technology that creates faster, more efficient and reliable charging infrastructure. Can you tell us more about that work? 

  Dr. Ahmad Bahai (14:05): 

  Yeah, absolutely. So new technologies, especially when it comes to technologies like electric vehicle, requires some infrastructure before it fully ramps up. And charging and infrastructure for charging is very important and it's being rolled over last many years and we see more of that in the future. But as you pointed out, the time that you spend at the charging station is very critical because a traditional internal combustion has a couple of minutes of filling your tank and moving on. People are not comfortable spending a lot of time sitting for charging. So how can we deliver more power at the shorter time or more energy at the shorter time to the vehicle? We have to offer higher voltage and higher current and a good understanding of how battery's state of charge is. There are multiple technologies that are critical to make this process much faster and more efficient. 

  (15:06): 

  We need to have devices that can deliver power very efficiently. That's why I mentioned that some of our devices that are designed for very high efficiency, high voltage power delivery are perfect examples of that. Also, you need to have intelligence. Communicate with the battery to make sure that how to charge the battery most efficiently without hurting batteries lifetime or safety. We are working on all aspects of that and we see that this charging time is getting better and better and shorter and shorter because we can deliver a lot more energy in a shorter period of time thanks to all these new semiconductor devices and intelligence that we are building in these charging systems. 

  Melissa Harrison (15:48): 

  All right. Well, you said the word intelligence, and so now I want to move into, we talk about AI at CES and we say AI is everywhere so I want to move into some questions just around edge AI and just to the work that you're doing there. Edge AI promises privacy and instant responsiveness. How is TI making it possible to run sophisticated AI models directly on battery powered devices? And what new experiences could that unlock for our homes or our personal electronics? 

  Dr. Ahmad Bahai (16:21): 

  Yes, of course the infrastructure for AI and intelligence are these massive data centers with amazing processing power, but also on the device side, we need to have a lot of intelligence because as we move forward to more AI enabled devices, we need a hierarchy of the AI and capabilities across the cloud all the way to the device. And these are for many reasons, for many examples that you need to make a decision right on the spot. As I said, in vehicles, you cannot just send the data out to the cloud and get back because you need very low latency. In some cases, privacy and security are important so we cannot just send data by transfer too often. In some cases, the power consumption or the connectivity bandwidth requirements are limiting factors for these massive data back and forth transfer of data back and forth. We see edge AI is a very critical part of this AI deployment in devices. 

  (17:23): 

  A lot of intelligence need to be local and the device needs to be able to handle a lot of AI capabilities with its own processors. I split it in two areas, we call it extreme edge and edge. I mean, edge AI is things like your car, as I said, needs to make a lot of decisions. I mean, intelligent driving and smart cars or your phone or your laptop. These are all intelligent devices which can have massive amount of AI processing in it, but also there are some extreme edges, which is like a sensor. A good example of that is a medical device that you put on your R, like a CGM and others. So all of these have some level of intelligence. And in TI, we have been working on hardware accelerated edge AI processors, which is very important from multiple angles, from power consumption angle, because these devices, most of them, if not all of them, are running off of a battery, your electric vehicle, your laptop in most cases and your phone or your handset or the patch on your arm. 

  (18:29): 

  We need to be very mindful of the energy consumption and hardware accelerator is optimized for very low energy consumption operation. But also, in terms of security, we can offer much higher level of security by having a hardware enabled or AI accelerator in the hardware that provides some unique features to make sure that the data is secured and the processing is done in a way that you don't compromise the privacy and security of the data. And now, we are launching multiple products with these capabilities in our microcontrollers in addition to all other components like connectivity and power management and other analog capabilities that we need around these devices. 

  Melissa Harrison (19:18): 

  And you mentioned CGMs and healthcare wearables, I have one on right now. Most of my friends and family have some type of wearable, and some of them really have gotten to the point where they have clinical accuracy. What advancements and analog sensing technology are allowing us to get to this medical grade level of data from just a simple consumer device? 

  Dr. Ahmad Bahai (19:43): 

  Yeah, that's one of the areas that are very close to my heart in terms of opportunities for innovation, but also impacting people's lives. As you mentioned, CGM, which TI pioneered with some other biotech companies, is a great example that you can offer a clinical quality measurement of the biomarkers in the body in a very small form factor and affordable price. And that can open up an opportunity for how to monitor our health and lifestyle and diet and everything else and see when we need to intervene could be as intervention at a personal diet level or physical activities could be a medical level. 

  (20:27): 

  The key point is that how to make it reliable enough that to avoid false alarms. And that is the opportunity for us to innovate on the device level and the circuit. All of these devices need to do some level of processing locally. They need to be connected to your phone or to the network, and they need to have some power management to run off of a tiny battery. All of these pieces need to come together in addition to biosensing capabilities that really can translate the biomarkers to a reliable signal that is passed to your phone or to your doctor or the clinicians. This can open a huge opportunity for not only the semiconductor innovation in semiconductor, but impacting people's lives and solving a lot of healthcare issues that can be very costly if we just leave it all the way until it becomes symptomatic. 

  Melissa Harrison (21:25): 

  And earlier, you mentioned privacy, which is something that I think is top of mind for consumers, not just from the healthcare wearables standpoint, but also from smart home devices. But how does on chip processing help solve that? And can you give an example of a smart home feature that this enables? 

  Dr. Ahmad Bahai (21:43): 

  Yeah. Privacy is very critical. Of course, when you talk about some healthcare related, it's even more critical to talk about it when it comes to private data and things like that, or even intelligent devices around your home, privacy and security are very important. Of course, when you do processing locally and you are not sending it over the air or communicating it back and forth too often, obviously you have more control over data, and that gives you one level of privacy and less chance for compromising your security of the data. But also, as you know, there are concerns that even for local data, you might be able to... I mean, there are technologies that can compromise privacy, and that's why we need to have more secure processors in these smart devices. 

  (22:34): 

  We have been working on this security protocol so that every device or every microcontroller has this level of security and secure hardware built into that to avoid any side channel attack, any chance of losing data to this kind of side channel interaction with our devices. This secure data core is something that we have optimized, not only in terms of the level of the security, but also power consumption and hardware size to make it very affordable and making sure that every device can benefit from this secure core hardware that we are building into our processors. 

  Melissa Harrison (23:14): 

  That's great. As we continue our conversation around AI, I want to transition to physical AI, better known as robots. We think it's going to be a big conversation at CES 2026, and we're seeing robots working shoulder to shoulder with humans on factory floors, they're in operating rooms with physicians. With your new modulators aiming for ultra precise robotics, what does that new level of precision unlock for collaborative robots and factories and even across the medical scene? 

  Dr. Ahmad Bahai (23:48): 

  Yeah, very interesting question, Melissa. A robot like AI, it's been a technology in progress for many decades. And we see that thanks to a lot of advances in semiconductor and in algorithms and in connectivity, a robot is becoming more and more efficient and coming to the point that unlike the past that where robots were in a cage doing a repetitive work of manufacturing lines and not working with humans because of safety issues, now robots can work side by side. That's what we call it, cobots, right? I mean, collaborative robots. And that makes it very challenging because now you need to have a lot more intelligence to make sure that the safety is not compromised when robots are working next to humans and going, walking or moving on the floor with humans around them, that means a lot more sensitivity and a lot more capability of checking and sensing what's going on around robots. 

  (24:49): 

  And that translates to massive number of sensor capabilities around robot. That, again, could be cameras, could be radars, could be pressure sensors, could be... I mean, humans have an amazing capability of when you touch something, you can detect the level of hardness of the object so you can act accordingly. You don't hit it hard, you don't let it slip through your fingers. So robot needs to have all of these capabilities or sensing capabilities that humans are equipped with and making sure that the safety of the robot working next to a human is already taken care of in the sense that not only they see what's around them, but they know how to interact with things around them with the right level of pressure, force and understanding what objects need to be picked up, what kind of force and grip you need to apply. There is massive amount of, I call it edge AI or intelligence and sensing comes to the picture. 

  (25:49): 

  And then you need to locally process all this data to translate it to actions. That means massive amount of data processing and intelligence, or we call it edge AI capabilities. And on top of that, these robots are mobile, so they are running off of a battery. So you need to do all of these at a very high level of energy efficiency and power management and balance management. So you can see again, being autonomous vehicle or a robot or other smart devices, there are some foundational advances in semiconductor that are essential to make these dreams a reality. And that's why we see that because of these capabilities now, robots are a lot more intelligent, a lot more capable, autonomous vehicle is a reality and data centers are growing in a staggering rate. 

  Melissa Harrison (26:40): 

  I'm really excited. I've not heard the term cobots before, so I hope it's okay if I start using that. I'll attribute it to you. You just mentioned a little bit about power demand. Will you just share a little bit with us about how TI is innovating in power management to make data centers more efficient and sustainable? 

  Dr. Ahmad Bahai (27:00): 

  Yeah. As I mentioned earlier, I think what's exciting about some of these power management opportunities in AI is that we are getting to a performance level that we never even dealt with just a few years ago. For example, having processor that requires multi-thousands of amps delivered to these devices at point of load, that's a very big challenge for a power management chip. In order to make that happen at a very high level of efficiency, we need new devices, new circuits and new system architectures. We are looking at all of these from device level, as I mentioned, we are working on new devices that can offer a much higher efficiency and much higher power density for power delivery. On the circuits, we optimize the circuit to make sure that the circuits are responding to the needs of the processor. And on the system, we design it in a way that this power management is happening at different stages and they're all working together very harmoniously. 

  (28:09): 

  The interesting thing is that we need to pretty much address power management and power delivery to these processes all the way from very high voltage, potentially 800 volts and above, all the way to point of load, which is sub one volt in a very optimized way because there is so much power at stake that we cannot be... We have to be mindful of every step of the way when this power delivery goes through different stage of the power transitions. 

  Melissa Harrison (28:39): 

  But it's really interesting because even though AI does have a power demand, AI is also helping us figure out how to solve for that power demand. So it's really interesting to hear you unpack how this is working in real time. 

  Dr. Ahmad Bahai (28:53): 

  That's the thing about AI. While we are doing all these things to make AI reality, AI is helping us to make all these things more efficient. So it's a two-way street that the more we can offer the intelligence at the device level, at the data center level, the more we can optimize things that we are doing at the circuit and system and device level. It's a very long, interesting conversation that we can have at some time later. 

  Melissa Harrison (29:16): 

  Well, we'll have to set up another podcast just to talk about that. We've talked about semiconductors, we've talked about AI, we've talked about robots and cobots. And at the real heart of the mission of TI though is to make technology more accessible, sustainable, human centered. I'd love to hear from you, what does human centered semiconductors look like to you and how does it influence how you approach research and development? 

  Dr. Ahmad Bahai (29:42): 

  Yeah. I think at the end of the day, as you pointed out, the technology that you're talking about with massive deployment or AI, the massive intelligence and all the semiconductor technologies that enabled them have the mission of making our life more efficient, safer and more sustainable. And all of these are important because let's say when we talk about these medical devices, how to make sure that we stay healthy and have a better quality of life, efficiency, how to use the capabilities of AIs and all these devices around us to have a more efficient and more productive life and how to be safer. So all of these translate to the fact that how to tie a market opportunity, a technology opportunity and the impact on the people's lives. And all of these are part of the whole ecosystem of innovation, and as I said, the foundation of the technology for all of that is semiconductor. 

  Melissa Harrison (30:44): 

  Well, before I let you go today, I want to take a look forward to CES 2026 with you. And I would love to know which area at CES do you believe is poised for the biggest leap forward once your next generation silicon infrastructure is in place? 

  Dr. Ahmad Bahai (31:00): 

  Again, it's such an interesting time for technology development and technology innovation. It's hard to pick one because they all play in symphony that is creating this amazing music for next generation of the intelligent devices. That could be human machine interface, how you can interact with machine more efficiently, display technology, robotics to offer creativity, and intelligence behind all of these things in data centers. I think the most exciting part is that just as we walk around CES, how to tie these pieces together to make this symphony play even nicer and more beautiful music. 

  Melissa Harrison (31:44): 

  I love it. Do you have one trend that you're most excited about to see at CES? 

  Dr. Ahmad Bahai (31:52): 

  I would say smart devices. I mean, the things are getting more and more smarter from basic displays that are getting so smart all the way to your smartphones and all the gadgets that you're carrying around. I would say higher level of intelligence that the device is an amazing trend that makes it more and more critical for all the applications that you are looking at, as I said, from moving to living to working. 

  Melissa Harrison (32:18): 

  Amazing. Thank you so much, Dr. Bahai, for joining us today and for sharing your insights. And I can't wait to see you in Las Vegas in January. 

  Dr. Ahmad Bahai (32:28): 

  Thank you so much. 

  Melissa Harrison (32:30): 

  And thank you to our listeners for tuning in. That's our show for now, but there's always more tech to talk about. Be sure to follow, subscribe, comment and whatever else you need to do to keep those algorithms happy. You can get even more CES and prepare for Vegas at ces.tech. That's C-E-S dot T-E-C-H. Our show is produced by Nicole Vidovich with help from Paige Morris and Doug Weinbaum, recorded by Andrew Linn and edited by Third Spoon. I'm Melissa Harrison on this special Inside Innovation episode of CES Tech Talk.