What is Qualcomm’s game plan to ensure that the next generation of chips is not only AI-ready but also affordable and relevant for large markets such as India?
Our two solutions — the 4-series and 6-series — were launched globally in India, which indicates that they are meant for this market. We continue to view affordability as a key consideration in the Indian market, and that was one of the main reasons these two chips were launched here. The market has become a bit complicated because of memory solutions and memory pricing. That is not something we can control.
As AI moves on to devices and more agents are launched or adopted, I don’t believe there will be much of a lag between India and the US or any other market. Historically, we have adopted technology very quickly — 5G and data are examples.
At this stage, the good thing about AI is that it can run in the Cloud. An agent running on a device that lacks the computing power can rely more on the Cloud, while a higher-tier phone can run the model on the device. I see this as an equaliser. AI can run in hybrid mode, either on the device or in the Cloud. As AI adoption grows and we see this next big transition in the market, India will not be left behind because we have options for running it across devices.
Do you see smart glasses becoming a mass-market product?
If you look at our published road map for glasses, you will see that we are already moving towards lighter, slimmer, and more compact designs. There is a chip coming up called AR (augmented reality) Light. We have showcased very compact form factors that fit into the temples of glasses.
Any technology starts somewhere and eventually has to move towards mass adoption. I think smart glasses will follow that path, too. It is a form factor that several of us are already wearing. The question is how to package the technology in a slimmer, lighter form. That is the path we are working on.
I see smart glasses being adopted, for sure.
Our launch with Lenskart is a classic example of how India is moving towards mass adoption. India is an important market for Qualcomm. The 19,000 engineers here are designing solutions for the global market. One of India’s advantages is its scale: 700 million people are connected to the internet. Given that scale, the market becomes a natural focus for us. When you look at this market, you have to adapt to it, which means making products more affordable.
One thing that surprises me is that the “waterfall”, as we call it, has become shorter and shorter. That reflects consumers’ desire, in a positive sense. The timeline (of adoption) will shorten, for sure.
What does the data centre opportunity look like for Qualcomm in India?
There are three categories: storage, training, and inference. India produces 20 per cent of global data but stores only 3 per cent. That helps explain why we are seeing more data centres being established here, even before considering the impact of agentic AI. The global figures need to be validated, but 20-30 per cent of the workload today may be inference, with the remainder being training. We expect that balance to reverse, with inference accounting for 70-80 per cent and training for around 20 per cent.
This shift changes data centre requirements. Inference calls for a different design approach, with power consumption, cost, and efficiency becoming central considerations. It is also more decentralised. That creates an opportunity for Qualcomm, which we have already begun discussing with the investment community through our announcements.
Inference requires high performance and low power consumption, capabilities that are deeply rooted in our mobile computing expertise. We can extend that expertise to data centres, at a different scale, to support inference workloads.
As our chief executive officer has said, we believe the timing is right because this next phase of inference is beginning now. Agentic AI will change the balance between training and inference and, with it, the requirements for computing infrastructure. That is where we see the opportunity for Qualcomm.
Does Qualcomm have a role in satellite technology?
Our infrastructure products, including Qualcomm distributed unit and radio unit solutions, are RAN products. We are working closely with Tejas Networks, an indigenous 5G player that is building 5G networks using Qualcomm’s RAN solutions.
We see opportunities for partnerships like this to expand and serve both domestic and global markets, strengthening India’s network infrastructure capabilities.
Is D2D connectivity an opportunity?
One hundred per cent. D2D, or consumer, connectivity has two parts. One is stationary or nomadic connectivity, like Starlink, where you stay in one place. The second is connectivity directly to a mobile device. This is where NTN comes in.
I think we will introduce that feature in the next release, towards the end of 2027, but the timeline needs to be confirmed. Devices will have that capability. You also need satellites with the same capability.
In India, satellite connectivity is generally used to fill gaps in the network. We are fortunate that operators have provided excellent coverage. I am not saying satellites won’t come. Countries with thousands of islands need those kinds of solutions. India will need them, too, but fortunately, our network coverage is quite ubiquitous.