Cristiano Amon has spent nearly five years at the helm of Qualcomm, one of the companies that helped define the smartphone era and one of the biggest players in the global semiconductor industry. He became CEO in 2021, in the middle of the global chip shortage, and since then has had to lead the company while the business that made it a giant began to change shape.
Now he is convinced that an even deeper transformation is coming. For fifteen years, the phone was the center around which devices, applications and services were organized. Amon believes that role is going to shift to AI agents. And he is moving Qualcomm accordingly: smart glasses, new personal devices, processors for data centers and a direct push into markets where Nvidia has had a head start for years.
It is not a bet he is making from a position of comfort. His way of looking at the industry comes from having grown businesses, managed stable ones and also spent three years running a company on the verge of bankruptcy. Perhaps that is why he tells his teams they are in the gladiator business.
It is not a motivational metaphor. It is a description of the odds.
Amon sees the industry as a fight that never ends. You enter the arena and three things can happen: you win, you lose, or both sides lose. And even if you win, it does not last very long. It only earns you the right to compete again the next day.
In an industry that produced some of the most valuable companies in history but also watched seemingly untouchable giants disappear, that idea works as a rule. For Amon, the fact that something works today guarantees absolutely nothing about tomorrow.
Three years at a bankrupt company
There is an episode in his career that helps explain that way of thinking.
In the 1990s, Amon left Qualcomm to take charge, in Brazil, of a company on the verge of bankruptcy in which Qualcomm held a minority stake. He did it because he felt there was one kind of experience he was missing: he already knew how to grow a business and how to manage a stable one, but he had never gone through a real crisis.
He says those three years taught him what would otherwise have taken a decade to learn. When a company is against the ropes, priorities stop being abstract: salaries have to be paid, taxes have to be paid and the business has to stay open.
That experience also helps explain the way he looks at today’s technological changes. And the one that matters most to him now has to do with where the center of computing will be.
From the phone to the agent
For the past fifteen years, almost everything revolved around the phone. A smartwatch, for example, mainly worked as an extension of the smartphone: it sent sensor data, received notifications and depended on it for many of its functions. That is why it made sense for the same company that built the phone to end up controlling much of the ecosystem around it as well.
Amon believes that center is now going to shift to the agent.
The phone is not disappearing — he says that often, and it is also in his interest to do so — but it will no longer organize everything else.
That changes the logic of the market considerably. When the phone is at the center, one company can control the device, the operating system and much of the surrounding ecosystem. If an agent capable of operating across different devices becomes the center instead, that dependence can weaken. And that creates room for companies that, until now, did not even make electronics.
Amon is convinced that 2026 will be the year agents begin to take that role. He is even willing to put a date on it: before the Northern Hemisphere summer, he says, every major operating system will have announced some kind of orchestrator.
He puts it simply: agents will take the place now occupied by applications, and orchestrators will take the place of app stores.
He uses a banking app as an example. Today, someone decides in advance what the user will see, in what order the information will appear, what colors the interface will use and which data will come first. Amon believes that logic could disappear.
If an agent has authorized access to your credentials, you could ask it to show you your balance, your last three transactions or a chart of your spending. And if you do not like the way it presents the information, you could ask it to change it. The application would stop arriving as a fixed product and instead be built around whatever you need at that moment.
He also imagines the next phone working in two ways at once: it will remain a device you operate directly, but it will also be able to temporarily hand control over to an agent that performs tasks on your behalf.
Glasses
Of all the new device categories, there is one Amon is especially confident about: glasses.
He believes they occupy a privileged position because they sit next to the eyes, ears and mouth and, more importantly, can see what a person sees as they move their head. Tens of millions of glasses are sold every year today. Amon believes that within a few years the figure could rise into the hundreds of millions. For perspective, around 1.2 billion phones are sold worldwide each year.
Beyond glasses, it is still unclear which other formats will take off. Qualcomm is working with around forty different designs: pendants, pins, watches, camera-equipped earbuds and even devices resembling alarm clocks that people can talk to.
He has also discussed the AI pins that appeared a few years ago and ultimately failed. For Amon, the problem was mainly one of timing: they arrived before agents existed that could actually do useful things with all the information those devices collected. Without that, they were little more than a camera and a microphone.
But there is another aspect he considers even more important.
Current models were trained on much of what humanity had already digitized: books, websites, emails, images, videos and posts. If millions of people start wearing glasses capable of recording part of the world they see, the amount of available information could multiply enormously.
For Amon, that is a major part of why AI companies are so interested in controlling devices. More than the hardware itself, what matters is the data it can generate.
What he does not answer
That is also where one of the more uncomfortable parts of his argument appears.
During the conversation, he was asked twice about privacy. The second question was direct: if he envisioned a world in which millions of people carried devices capable of constantly seeing and listening, did that not concern him?
Amon did not answer directly.
He talked about a technology that is still in its early stages, the regulation that will probably emerge, the need for user permission and adoption that will happen case by case.
Earlier, he had offered a more commercial argument: if a device is going to see and hear a large part of what you do, you will probably want it to be made by a company you trust.
It is also the argument one would expect from someone whose business depends on those kinds of devices eventually becoming mainstream.
Memory
When it comes to the memory shortage, by contrast, Amon is much more direct.
Asked about supply problems, he responded with a joke that captures the situation fairly well: “If you have memory, I’ll buy it.”
His view is that installed capacity simply cannot keep up with current demand. While much of the industry expects the situation to normalize only in 2028, Amon believes that could happen in the second half of 2027.
For him, the shortage also exposes a deeper problem: many of today’s devices were not designed for the workloads that are coming. That is why Qualcomm is reviewing virtually its entire roadmap.
There is also a technical detail with major business implications. Orchestrators run mainly on the central processor rather than on graphics processing units. After several years in which almost all the industry’s attention was focused on GPUs, that could restore some of the CPU’s importance.
And Qualcomm happens to make exactly those kinds of processors.
One in the United States, one in China
Amon’s other major bet is on data centers. And there, Qualcomm is arriving late and going head-on against Nvidia.
He openly admits that, if he could go back, he would have started this plan several years earlier. At the time, he had other priorities and Qualcomm did not yet have all the pieces it needed.
Now he wants to offer a complete package, from silicon to software, but without closing the ecosystem. To reinforce that strategy, Qualcomm acquired Modular. His goal, he says, is to build something similar to what Android represented for phones: an open alternative to a closed ecosystem.
When discussing early customers, he identified some and kept others confidential. He mentioned Meta for central processors, Microsoft for a computing technology that avoids the use of high-bandwidth memory, and two major cloud providers whose names he declined to reveal: one in the United States and one in China.
The mention of a Chinese customer opens a more delicate issue: the technological relationship between the United States and China.
Amon accompanied Donald Trump during his meeting with Xi Jinping and argues that maintaining trade between the two countries also helps create stability.
Faced with U.S. restrictions on technology sales to China, he reframes the debate around a broader question: is it better for technology developed by American companies to remain inside the United States, or to be sold around the world?
When the journalist pointed out that the question had been framed in a way that almost assumed the answer, Amon returned to the argument he considers central: the size of the market.
Five years, two shortages
There is a parallel Amon does not explicitly draw, but it helps make sense of his five years at the helm of Qualcomm.
He became CEO in 2021, in the middle of the global chip shortage. Qualcomm was also coming out of a long legal battle and had resisted a hostile takeover attempt.
In that environment, he made a risky decision: to bring the transition to 5G forward by two years, at precisely the moment when semiconductors were harder than ever to obtain.
Five years later, he is facing another shortage, this time in memory, and once again his response is to accelerate: a new business line, a virtually rewritten roadmap and an entry into a market where his main rival has had years to build an advantage.
It is fairly consistent with someone who sees his job in one way: enter the arena, compete and, if he wins, get ready to do it all over again the next day.
