Nvidia is shifting its focus from just chips to a more integrated approach, identifying a data centre as the new unit of computing, Hasan Aijaz, business head (cloud and consumer internet), Nvidia South Asia, said on Thursday.
“You don’t have to think chips. You don’t have to think systems. You have to think how chips and systems work together along with the network, along with the software. And the applications which come at the top of this entire stack, how do you accelerate them?” Aijaz said at the Bengaluru Tech Summit. “Accelerated computing, we realise, is the future,” he added.
Meanwhile, Qualcomm said it is focused on edge AI, particularly in making on-device AI more accessible and efficient for consumers. Srini Maddali, SVP, engineering at Qualcomm, said: “AI on the edge is not an option; it is a necessity.”
He highlighted that Qualcomm has already integrated on-device AI into several products such as laptops, smartphones, IoT devices, and smart assistants. “The recent one we launched, the Snapdragon 8B, can not only dynamically manage AI work flows but also multi-modal gen-AI real-time,” he said.
At the panel discussion, Aijaz and Maddali also shared their perspectives on how their companies are shaping India’s AI landscape.
“India is a powerhouse of skills and intelligence for us. If we train the engineers, if we help them with the right products and tools, India and Indian engineers are going to take the world by a storm,” Aijaz said.
Looking ahead, Nvidia foresees a collaborative AI future where ecosystems rather than individual companies drive innovation.
“It will not be the one company which will win; it will be the ecosystem which will come together and which will help build AI applications for the various uses that humanity is used to,” Aijaz added.
Qualcomm’s strategy for India involves tailoring technologies to suit the cost-sensitive market while leveraging its rapidly growing digital ecosystem. “India is probably an exponentially increasing number of people getting onto the smart devices. As Qualcomm, we work with our partners to bring these technologies to the end-users,” Maddali said.
The company has developed a proprietary Hindi language model capable of translating across 22 languages with 800+ combinations. “This technology…will enable us to translate language from different dialectics,” Maddali added.
