New Delhi: Qualcomm is taking aim at Nvidia’s leadership in data centre chips designed for artificial intelligence (AI) workloads, targeting a significant market share in the category in the coming year.
A top executive told ET the US-based fabless chipmaker has created several technologies that it believes have tremendous relevance for data centres. “We have to find the right time to go into data centre business with something that’s completely unique,” Akash Palkhiwala, chief financial officer and chief operating officer at Qualcomm, said in an interview.
Competition is intensifying in the data centre AI chip market, which has shifted from a near monopoly of Nvidia to a fragmented landscape. Hyperscalers and AI labs such as Amazon, Anthropic and Google are increasingly relying on in-house custom chips to train and run large models.
“The data centre market is so big that there is room for multiple players,” said Palkhiwala. “Qualcomm’s broad technology portfolio has extreme relevance in data centres, and we are very focused on bringing... something unique (for) a competitive advantage and differentiation.”
The San Diego-headquartered company has prepared three major areas or product lines for the data centre market. “The first is making a CPU chip for agentic AI workloads in the data centre,” said Palkhiwala. “The second product line is the AI accelerator, where Nvidia is the leader.”
Incredibly exciting
“The third thing we are doing is building custom chips for hyperscalers that start shipping by end of this calendar year,” said Palkhiwala. The CFO said the AI accelerator chip market is changing, and rather than having the same solution for varied workloads, one can build solutions for specific types of workloads and optimise, based on the needs. “We are building something that is totally unique and a new architecture versus what the industry has,” he said. Apart from data centres, Qualcomm — best known for smartphone chips — is diversifying into AI-led wearables such as smart glasses, pendants and rings, as well as automotive, telecom infrastructure and laptop computing platforms.
Terming as “incredibly exciting” the developments in the field of personal AI devices, Palkhiwala said companies are trying a variety that has the capability to become an individual’s personal agent, and while it’s not clear now which device will win, most are using Qualcomm chips. “We have partnerships with Google, Meta and OpenAI, among others,” he said.
On the memory chip crisis and its impact on Qualcomm’s financials, he said the company expects the June quarter to be the bottom after which revenues are expected to grow. Smartphone prices have been rising due to higher memory chip prices amid severe supply constraints.
On the semiconductor manufacturing ecosystem in India, Palkhiwala said Qualcomm has a very close relationship with Tata group, spanning several areas. “They do packaging for us; we are partnering with them for a system-in-package,” he said. “We also do certain manufacturing technology for power management chips with them and, eventually, when manufacturing happens here, we will bring our orders to them.”
Qualcomm also has partnerships with other Indian firms such as Dixon Technologies for smartphones and wearables. “We recognise India as a major manufacturing hub, and because we use a lot of manufacturing from around the world, we can be one of the key customers for manufacturers in India,” he said. “We are very excited with the initiatives that the government has put in place and the way the companies are responding to it.”
A top executive told ET the US-based fabless chipmaker has created several technologies that it believes have tremendous relevance for data centres. “We have to find the right time to go into data centre business with something that’s completely unique,” Akash Palkhiwala, chief financial officer and chief operating officer at Qualcomm, said in an interview.
Competition is intensifying in the data centre AI chip market, which has shifted from a near monopoly of Nvidia to a fragmented landscape. Hyperscalers and AI labs such as Amazon, Anthropic and Google are increasingly relying on in-house custom chips to train and run large models.
“The data centre market is so big that there is room for multiple players,” said Palkhiwala. “Qualcomm’s broad technology portfolio has extreme relevance in data centres, and we are very focused on bringing... something unique (for) a competitive advantage and differentiation.”
The San Diego-headquartered company has prepared three major areas or product lines for the data centre market. “The first is making a CPU chip for agentic AI workloads in the data centre,” said Palkhiwala. “The second product line is the AI accelerator, where Nvidia is the leader.”
Incredibly exciting
“The third thing we are doing is building custom chips for hyperscalers that start shipping by end of this calendar year,” said Palkhiwala. The CFO said the AI accelerator chip market is changing, and rather than having the same solution for varied workloads, one can build solutions for specific types of workloads and optimise, based on the needs. “We are building something that is totally unique and a new architecture versus what the industry has,” he said. Apart from data centres, Qualcomm — best known for smartphone chips — is diversifying into AI-led wearables such as smart glasses, pendants and rings, as well as automotive, telecom infrastructure and laptop computing platforms.
Terming as “incredibly exciting” the developments in the field of personal AI devices, Palkhiwala said companies are trying a variety that has the capability to become an individual’s personal agent, and while it’s not clear now which device will win, most are using Qualcomm chips. “We have partnerships with Google, Meta and OpenAI, among others,” he said.
On the memory chip crisis and its impact on Qualcomm’s financials, he said the company expects the June quarter to be the bottom after which revenues are expected to grow. Smartphone prices have been rising due to higher memory chip prices amid severe supply constraints.
On the semiconductor manufacturing ecosystem in India, Palkhiwala said Qualcomm has a very close relationship with Tata group, spanning several areas. “They do packaging for us; we are partnering with them for a system-in-package,” he said. “We also do certain manufacturing technology for power management chips with them and, eventually, when manufacturing happens here, we will bring our orders to them.”
Qualcomm also has partnerships with other Indian firms such as Dixon Technologies for smartphones and wearables. “We recognise India as a major manufacturing hub, and because we use a lot of manufacturing from around the world, we can be one of the key customers for manufacturers in India,” he said. “We are very excited with the initiatives that the government has put in place and the way the companies are responding to it.”
