The Same Snapdragon That Flies on Mars Powers a Galaxy Phone

Lex Fridman starts with the object in nearly everyone's pocket. A modern smartphone holds on the order of ten billion transistors, roughly a million times more compute than the best machines of fifty years ago, and billions of them talk to one another. Does that still blow the mind of someone who has spent a career inside the industry? Cristiano Amon, CEO of Qualcomm since June 2021 after more than 26 years at the company, says yes. One reason he loves the firm, he tells Fridman, is the knowledge that the technology it builds can change the world.

The part he wants to add, though, is not only raw processing power or the fact that much of the planet now has broadband in hand. The smartphone is also mankind's largest development platform, he says, and there is nothing else like it. People treat the phone as a consumer gadget for video, photos, and messages. Amon wants both the hardware and the software taken seriously, because this is the first time billions of people have held a programmable computer at once.

Fridman tries an alien thought experiment: if someone had been watching Earth for the last fifty to seventy years, what would they notice? Amon does not stay on the technical spectacle. He sees a contrast between the last century's extraordinary development and events that, in his view, do not belong in the 21st century, including the war in Ukraine. That, he thinks, would puzzle the visitors more than the gadgets. Fridman replies that it may not be puzzling at all; it is human nature repeating itself. Amon agrees.

Qualcomm was central to 2G, 3G, 4G, and 5G, and Amon has been at the company since the CDMA years of the 1990s. He narrates the generations as a sequence of problems, not as marketing labels. 2G and CDMA asked whether every person on Earth could have a cell phone. 3G asked whether that phone could reach the internet. 4G was broadband: a computer in the palm of your hand. 5G, he says, is a different kind of brief. It is how you build technology for a society that will be 100 percent connected to the cloud, and how you make wireless the last-mile fabric for everything.

He lists what that design had to do. It had to kill data congestion, including the soccer-stadium case where tens of thousands of people all want broadband at once. It had to connect not only billions of people but billions of things. For the first time in wireless, it had to deliver mission-critical service: up through 4G the radio was best-effort, while 5G can guarantee a link to the cloud. And it had to create a fabric such that, as a society, an unconnected object starts to look like the exception.

The analogy he has used since the early days of 5G is electricity. Nobody sits around debating the use cases for power from the wall; they assume it is there. Qualcomm wants 5G in the same place. Fridman notes that "everything" keeps expanding, and that the refrigerator in the room is probably still offline. Amon's answer is the company's other half: communications and advanced computers for low-power devices, from the robot on the floor to cameras, factory machines, and retail. He mentions work with companies such as Walmart on electronic shelf labels and smart cameras that can see a shelf that needs restocking. Processor, connectivity, and AI, applied to a market you cannot actually finish defining.

Coverage is not one 5G. There is sub-6 GHz and there is millimeter wave, and they are not the same on speed or on how you build them. When Amon meets regulators around the world, he says he tells them "resistance is futile" and to allocate every remaining slice of spectrum to wireless. The industry extracted a great deal of bits per hertz moving from CDMA to OFDMA, he argues, but the capacity and speed still demanded are too large to squeeze from efficiency alone. You simply need more spectrum.

He does the arithmetic from the operators' side. Since the 2G era, carriers have bought licenses in auction after auction. Add up everything they accumulated across 2G, 3G, and 4G, and you may get one or two sub-6 channels at most, each about 100 or 200 megahertz. That is why 5G was designed to run across essentially every band, from roughly 450, 600, or 700 megahertz all the way to 42 gigahertz, where millimeter wave lives. Sub-6 deployments cluster around 3.5 GHz not because the band is magic, he says, but because it was the spectrum still available; the rest was already in 4G. You can hang new radios on existing towers and get coverage fast, and that is what the world, including the United States, has been doing.

Millimeter wave is slower to build because it behaves like Wi-Fi: dense access points, permits, fiber, city by city. Verizon has led in the United States, with places such as Chicago and Manhattan starting to see coverage, and Amon says it will take years. It is still inevitable. There is not enough spectrum, so every 5G operator will get there. Fridman asks who the real obstacle is: politicians, regulators, consumers, or money. Amon takes regulators off the list. He has not met one who thought it important to be late on 5G. The real constraint is the time it takes to build infrastructure, which he compares to roads and ports. Operators' number-one complaint is site permits, sometimes negotiated municipality by municipality.

Fridman asks the question that makes CEOs nervous: some people worry there are not enough studies of 5G's effects on the body. Is it safe? Amon answers from physics. As networks densify, he says, power at the tower and at the phone goes down, because a handset closer to a cell needs less energy to close the link. Moving from 4G to 5G, in his account, reduces the power required for the radio link. The FCC runs rigorous device-safety tests that other countries have used as a model. Cellular has been around for decades, the smartphone is the most beloved device people own, and he argues that 3G and 4G data did not bear out the early fears. 5G is newer, but it is also lower power.

Fridman splits the worry in two: biology versus what happens when you scale any technology exponentially through a civilization. Love, he notes, can hurt. Amon treats that second fear as the healthier one. A fully interconnected society is already building digital twins of people and things, and it is natural to ask what rights, obligations, and protections from the physical world should follow you into the digital one. He thinks Europe is ahead of the United States here, from privacy law through the Digital Markets Act, which he describes as an attempt to keep digital markets open to innovation and competition so that no single company holds all the data and writes the rules of the digital realm.

Fridman then vents about regulators who follow technology clumsily, meaning well while doing unintended damage, especially when they do not understand machine learning or recommender systems. Amon does not argue. He still likes the European framework, then puts the other side of the ledger: Qualcomm depends on the ability to innovate, on free markets, and on the ability to ship disruptive technology. At the same time, he says, tech companies should spend time helping governments understand these systems ahead of time so they can be better prepared.

Snapdragon is the system-on-a-chip behind most premium Android phones, iterating from last year's Snapdragon 888 on 5 nanometers to this year's Snapdragon 8 Gen 1 under a new naming scheme. The idea Amon is proudest of is not a single block. It is the discipline of putting every computing capability the phone needs onto one chip: cellular, Wi-Fi, Bluetooth, every satellite constellation used for positioning, a power-efficient single-thread and multi-thread CPU, a GPU for graphics and the fastest-growing slice of gaming, an AI block Qualcomm calls the neural processor unit, plus video and multimedia engines for audio and imaging.

The constraints are the part that sounds like a shop floor. The phone is small enough to hold. You will put it against your face, so it cannot run hot, and thermals have to be managed. Users expect a full day of battery. They also want gigabits, not tens or hundreds of megabits, eight cores, and a GPU with ray tracing, the sort of list that used to live only on a desktop PC. Doing all of that in a handset, generation after generation, is what he calls an incredible thing, and one Qualcomm is proud of. Camera and computer vision, he adds, now often advance in the smartphone first, because it is the highest-volume, most competitive, fastest-iterating computing platform on Earth.

Fridman piles on the rest of the brief: every radio has to be good, the chip has to get faster every year, new applications arrive as if they were free, and the NPU, GPU, and CPU have to share the machine without killing the battery. Qualcomm cannot pick a lane and call itself a CPU company, a GPU company, or an AI company. It has to be good at each of those technologies and then make them dance together.

Asked what it actually takes to design a chip like that, Amon, more than 26 years in, names two ingredients. The first is a collaboration culture strong enough for technologies that evolve on different clocks. You have to be excellent in each one, then make them work together. His image is a train: "The train is going to leave the station. Every car needs to be there right when it leaves, and it needs to leave on time." In phones you cannot move Christmas, you cannot move Black Friday, and you cannot move the selling season. The launch date is fixed, and every technology has to be on the train.

The second ingredient is discipline. These are complex systems, and in many cases Qualcomm is ramping production before the silicon is even back from the fab. The company has to trust simulation models and has to design from the start to commercial quality. There is almost no room for mistakes. Fridman notices, as if for the first time, that the *qual* in Qualcomm is quality. "No wonder our name is quality communications," Amon says. For an engineer, hard deadlines and no slack are stressful and, both men agree, exciting.

Google now ships its own Tensor silicon in Pixel 6. Apple has long designed its own chips. How, Fridman asks, does Qualcomm beat them? Amon rejects the premise. The company does not have to out-compete Google. When he became CEO he wanted the mobile story unambiguous: Snapdragon should be synonymous with a premium Android experience. Qualcomm still builds chips across price tiers, and every year it still puts out a flagship, but the point of the stack is to enable Android, not to pick a fight with the operating system's owner. He calls Qualcomm the number-one enabler of Google's Android ecosystem.

The largest customer is Samsung, which has always had its own silicon. For years the mix was roughly even: the most advanced markets, the United States, China, Japan, Korea, went to Qualcomm, while emerging markets and channels Samsung controlled with appliances and other goods used Samsung's own chips. On the Galaxy S22 that had just launched, Amon says about 75 percent is Qualcomm. Behind Samsung sit the Chinese Android OEMs, Xiaomi, OPPO, OnePlus, and vivo, among Qualcomm's biggest customers and a force pushing Android upmarket. Xiaomi was number one in Europe at one point the previous year. After Huawei shrank under sanctions, those brands took share, especially in China, and they are doing the premium work on Snapdragon.

Apple is a different relationship. Qualcomm supplies modem technology under a multi-year deal, and Apple has been public about building its own modem. Amon's line is that Qualcomm's mobile strategy is not defined by selling Apple a modem. The strategy is the integrated Snapdragon SoC and the Android ecosystem. Asked which Snapdragon phone he actually uses, he admits to several and then names the Galaxy S22.

Fridman asks whether Qualcomm can take on Intel and Apple in laptops and desktops as the PC starts to look like a large smartphone. Amon first wants the frame corrected. Qualcomm is already changing. 5G was designed to connect everything, and the connectivity and processing that were designed for phones are moving into almost every industry. The PC is one of those markets, and the pandemic, he says, revealed something the telecom industry had been chasing for decades: the number-one use case on a personal computer today is communications.

He walks the graveyard of video telephony. There were ISDN-era AT&T desk phones with tiny screens, before IP was even common. *Back to the Future Part II* staged the same dream. 3G was sold as the data pipe that would finally make video calls real; 4G was sold as the broadband that would. It took a pandemic. Once that happened, buyers started caring about cameras, microphones, the latest Wi-Fi and cellular, battery life, and portability, because work now happens in places you did not plan. A CAD engineer will not drag a workstation home. The job has to run in the cloud, which means 5G as a fat link between two computers. CIOs, watching staff spend days at home, want the data in the cloud as well, on tools such as Microsoft OneDrive, which also need bandwidth.

The product example he offers is from Mobile World Congress: Lenovo announced a ThinkPad on Snapdragon with 5G and 28 hours of battery life. Microsoft, on Windows 11, is using Snapdragon's NPU so that on a Teams call your eyes appear to look at the camera the whole time. Fridman says that is hard, which is why he still insists on in-person interviews. Amon treats the difficulty as the opportunity, and then sketches the next step: better PC calls, then connecting physical and digital spaces, then, in the metaverse work, calling someone as a hologram.

Fridman returns to the aliens, centuries on, writing the history of the 21st century. What would they remember Qualcomm as, a car company, a robotics company, some other pivot? Amon says Qualcomm is already an automotive company. What he wants it remembered for is broader: intelligence and processing at the edge, everything outside the data center. In 4G, broadband and mobile computing grew side by side. In 5G, he says, AI develops the same way. Devices connected to the cloud all the time can decide in real time, send context upstream, and compare notes with the cloud. Cars are the working example. Qualcomm, he says, now works with 26 global brands. You will not put a server in the trunk, but you still need that much compute, and the thermal and power limits look a lot like a phone.

The product name is Snapdragon Digital Chassis: compute and sensors for assisted driving, a digital cockpit, and a connection from the car to the cloud. Will you be able to buy a Qualcomm car? No. Traditional automakers, in his view, must become technology companies, but they are not going away, and it is a mistake not to bet on them even when new electric names have flashier market caps. Qualcomm wants to be the preferred technology partner, the brain without the badge. Fridman mentions Rivian; Amon's point stands either way.

Robotics sits in an IoT business he calls one of Qualcomm's fastest-growing. Amazon's Astro robot, he says, has two Snapdragons in it. Industrial robots, drones, and the factory and warehouse of the future, once 5G is on the floor, are the rest of the list. Then he names the most famous robot on Earth at that moment: NASA's Ingenuity helicopter on Mars. The cameras, the flight control, the whole machine run on Snapdragon, the same chip that goes into a Galaxy, not a special space part. If that computer vision stack were the first thing to see an alien, he jokes, they did not program for it. For an engineer, he adds, the quality bar is the same as the phone: imagine it getting to Mars and not working.

Fridman asks for an expert read on the global semiconductor shortage. Amon separates cause from aggravation. The main driver, he says, is that almost every industry is being digitally transformed, so the amount of silicon required is a different order of magnitude. A new car launching today, in his example, most likely has ten times the chips of the prior model, and the program people are working on next may be ten times that again. Billions of things are becoming smart, more data is going to the cloud, data centers are growing, and the floor on semiconductor consumption has moved up a long way.

The pandemic sat on top of that. Homes became enterprises, with massive broadband and IoT upgrades. Offices changed around collaboration tools and video. People stuck inside bought phones, tablets, computers, and game consoles. Those spikes were real, Amon says, but they were not the main issue. The main issue is long-term digital growth. Fridman restates it: the pandemic was not the cause, it was an aggravation. Amon agrees.

The way out is supply. The industry, he says, is planning to double total semiconductor manufacturing capacity within five years, and Qualcomm's suppliers are already expanding. Policy is adding a second push: the United States CHIPS Act at about $52 billion and a European counterpart at about 43 billion, with a combined goal of making at least 50 percent of what those regions consume. He likes that not only for more capacity but for capacity that is geographically distributed, which is how you actually design a supply chain. If the crisis taught anything, it is that semiconductors matter, semiconductor companies matter, and they need investment.

Fridman asks about the billions Qualcomm has paid and received in settlements and fines, and whether giant lawsuits help or hinder innovation. Amon changes the tense: there *used* to be a lot of lawsuits. Qualcomm has two businesses. One is semiconductors, designing processors and modems, building products, selling them, not licensing that chip technology. The other is licensing, in the company from the beginning. As a leading contributor of essential technology in 2G CDMA, 3G, 4G, and 5G, Qualcomm put those inventions into the standards and licensed them so everyone could produce, collecting royalties on standard-essential patents rather than keeping production to itself.

The fight with Apple is over. "They're my customers," he says. In the process, the licensing model was tested in what he describes as every geography, and in each one it was upheld as pro-competitive and consistent with fair, reasonable, and non-discriminatory terms. He would argue it is the most successful licensing business in the industry, and the most stable, because it has been battle-tested in essentially every jurisdiction. That past, he says, now creates stability. The company's growth, however, is in semiconductors. Licensing is selling an idea that sits in the standard; the chip business is designing Snapdragon, having a foundry build it, and selling a product.

He is not yet a year into the CEO job, with the anniversary due in June. The first lesson is the one many CEOs give, and he means it: he needs more time. Qualcomm is expanding in several directions at once, which means more industries to learn, more relationships to build, more technologies to sit with, and a calendar that reflects the size of the opportunity. The second lesson is personal. Business partnerships are done by people, and long-term trusted relationships matter; Qualcomm does not play short, and new partnerships are expected to last decades. The third is timing. Physical and digital space are merging, the automobile is transforming, computing and mobile are collapsing together, and the home has become an enterprise. Those trends point at Qualcomm's technology, so he is in a hurry. It is exhausting, including the internal work of aligning employees and then convincing investors, but he says he likes doing it.

Fridman asks for advice to people in high school and college, given a path that ran from Brazil through a risky move to Japan and then to the top of Qualcomm. Amon's first answer is to have a plan, even if it only covers the next two years: what you want to do, where you want to go, a dream you actually pursue rather than sleep on, and belief in yourself. He joined Qualcomm as an engineer with no plan to become CEO. He did know, as an engineer, what he wanted to work on and where he wanted to contribute, and then he kept evolving from that point.

The career advice he got early, and still gives, is more specific. Spend time learning what you are good at and what you are not, the real border between competence and incompetence. Once you can see it, you know what to work on, and you can say where you want to go next and what the gap is. It is faster, he says, when you identify that yourself before other people tell you. The next step follows automatically: surround yourself with people who are very good at the things you are not, which requires being radically honest about those gaps. How often do his plans actually work? He tries to keep the hit rate above 50 percent.

His father was an electrical engineer at a utility company and wanted him in traditional power generation and distribution. Amon liked electronics and communications, so he studied both. The first job out of college was at NEC in Brazil; about a year in, they transferred him to headquarters in Tokyo, the first time he had left Brazil, a different culture on the other side of the planet. His father still gives advice, tells him he has not changed, and is proud that few Brazilians have reached a CEO seat of this size. Amon feels that as a burden, especially for the Latino community: he has to be a good example.

On mortality he is direct. He is a devout Christian and believes this life is a transition, but he does not spend much time on that. He tries to live in the present and make this place a little better. The meaning of life, he says, is a question each person has to answer; he believes people are here for a purpose, and his prayers are that he stay on track for whatever that purpose is. Love, in the human condition, is not optional. It comes in the package, and looking at underprivileged communities and homelessness, he thinks everyone needs more of it. Fridman worries that people who build astonishing technology sometimes forget that part. Amon comes back to smartphones as the first time essentially everyone could connect to the internet and to one another, which he still sees as an incredible force for good.