What is the scope of NXP Semiconductors’ products?
NXP is predominantly a processor company, focused mostly on general-purpose edge and embedded processors. Our full portfolio starts from traditional microcontrollers, which are efficient in terms of cost, power, and runtime. Then we move into crossover MCUs that offer more capability, approaching full applications processors. Finally, we have applications processors - Linux-based, multi-core, heterogeneous compute environments that can drive anything from toasters to industrial robots and automotive systems.
We are putting AI into each of these product lines, specifically by integrating dedicated hardware acceleration for AI workloads. We are also enabling software on those devices so that customers can build AI solutions directly on them.
How are software-defined vehicles (SDVs) changing the automotive segment?
Automotive used to be a conservative domain, with features added slowly and with heavy oversight. But with the shift towards software, we are now seeing much faster delivery of advanced and innovative features directly to the end user. That is the transformation SDVs enable.
SDVs are an area where NXP started investing early. We have a long legacy - from the Philips side of NXP and the Motorola side of Freescale - so automotive semiconductors are part of our pedigree.
What edge and embedded technologies for automotive is NXP Semiconductors working on?
In automotive, we have the CoreRide platform and an SDV approach. I am part of the business line handling automotive from an in-cabin perspective, not the chassis.
That includes infotainment, e-cockpit, driver and passenger assistance – features that enhance the in-vehicle experience. Across industrial and IoT as well, we focus on enabling developers – the ones building the final product or components – to leverage our hardware in ways that help to accelerate deployments and reduce friction.
What new edge computing advancements for industrial and IoT applications can we expect on the market in the next 12 months?
What is possible in mobile robotics now is fascinating -
We are shifting from science fiction into real technology. There is a big move towards humanoid or assistance robots which involve human interaction and can be applied in industrial settings, IoT, smart homes, smart buildings, or consumer-facing environments.
Exciting new deployments in healthcare are about better patient care and practitioner support. Deploying on the edge allows better monitoring of a patient’s health context and service delivery, ultimately improving outcomes.
The hype around ChatGPT and conversational LLMs intersects with embedded platforms through conversational human-machine interfaces. We are demoing that on very small devices, which could be used in factory automation – asking, for example, how machines ran overnight, getting system reports, or even having AI agents that act proactively.
The public is becoming more concerned with data security and safety. How does that affect NXP Semiconductors’ product strategy?
Our devices include fundamental security aspects – things like TrustZone, Zero Trust deployments, and trust provisioning through our flows. These technologies are designed to ensure that the device is secure, because without security, you cannot have safety. If your workload or runtime is compromised, your safety is also compromised. Security is a core part of what we do; it is separate from AI and has to be there even in traditional embedded development. With AI accelerating vulnerabilities or expanding the attack surface, the need for basic security fundamentals remains and needs to evolve and be ready for future threats.
For this we have started adding post-quantum cryptography (PQC) to our devices starting with the i.MX 94 family, to provide security for risks quantum computing poses in cracking previously unbreakable algorithms (most notably, RSA and ECC) and leaving most devices and networks that depend on these algorithms, vulnerable. We have implemented PQC from boot time, so the device has the ability to boot with classic cryptography, and with PQC. We support boot, secure updates and secure debug access. The point is to be ready because it’s very difficult to retrofit devices, especially when it comes to the implementation of secure boots. The secure boot starts from immutable memory, because this is the part of the code you need to guarantee is not modified.
You travel to trade shows around the world. How do focus areas in the industrial market vary between regions?
The US tends to focus more on the bigger picture or large tech movements, especially in AI. That is where you find the enterprise-level players - Google, Meta - and the contributions they are making to the ecosystem. In Germany and Europe, I see a stronger focus on actual deployment, especially in factories and industrial use cases. It is real tech being used in real manufacturing environments. Asia, on the other hand, has a very practical and pragmatic focus on delivering technology.
Today’s approximately 15 billion edge AI devices is expected to double in five years. How is NXP keeping up with the speed of innovation?
Without software, even the best hardware is useless - you cannot extract its full potential. We have to enable it, and do it quickly, because everything is moving fast. It is a constant race. A new model gets published, like DeepSeek, and suddenly it is competing with OpenAI, offering similar performance at a lower cost.
That is good for the edge market. It shows someone is shrinking large-scale models to run on embedded platforms, making that quality and technology more accessible. At NXP, we are able to demonstrate running newer models like DeepSeek on our hardware – less than a few weeks after its release. We optimize previous models while preparing for the next, constantly improving in a feedback loop.
How is NXP Semiconductors using AI to cut its own operating cost, runtime and power burdens?
We are using AI internally to improve our development processes - coding with AI, building solutions with AI. Increasingly, our enablement includes AI building blocks, so you are essentially building AI with AI on the edge.
For us, the software we build to enable hardware is becoming more automated. Tasks that took weeks now take hours. So, the question becomes: what do you do with that extra time? You build better, constantly improving. It is a continuous cycle of enhancement.
