Indigenous server technology vital for India’s security: Netweb CMD

Sanjay Lodha, CMD of Netweb Technologies, a leading Indian manufacturer of motherboards and server solutions, discusses the importance of indigenous server technology and the impact of recent government initiatives on the industry. In an exclusive interview with Bizz Buzz, Lodha dives into how 'Make in India' servers can benefit both domestic and international players, the role of AI in supercomputing advancements, and the collaborative efforts between Netweb and NVIDIA to establish GPU servers in India

How made in-India servers are more secure and affordable for domestic and international players? How will the recent changes in the PLI scheme for semiconductor manufacturing benefit you as a motherboard manufacturer?

Servers are the basic building blocks for many high-end solutions, such as supercomputing, AI, and enterprise computing, where the processing of data and the underlying algorithms lie. Therefore, any country like in the Indian context must have servers' technology that is indigenously designed and manufactured, making it secure in many dimensions.

Currently, meeting the domestic requirement is the priority. Afterwards, as the entire ecosystem develops, very soon, the focus will shift towards serving the international market with custom-designed products and solutions designed for such markets. By focusing on indigenous manufacturing companies, we will be able to offer competitive pricing while ensuring ambitious standards of quality and reliability, thus providing a more secure and affordable solution for both domestic and international markets.

We would welcome semiconductor industry setting up facilities in India. This will enable us to source the relevant components for our PCB from domestic companies instead of importing them, resulting in an optimized supply chain and reduced import imbalance. Furthermore, the change in the PLI scheme - that will have increased fiscal support to 50 per cent of the project cost for all semiconductor fabs, regardless of the technology, and for display manufacturing- also means reduced capital expenditure, access to more advanced technologies, and an overall boost in competitiveness and development of the ecosystem.

How is artificial intelligence poised to revolutionize supercomputing environments by unlocking new levels of performance and efficiency, and what specific advancements can we expect in the near future?

I would like to emphasize that the artificial intelligence journey has only just begun. Today, what we are witnessing is the adoption of early use cases. However, technology has a profound potential to immensely impact and improve every sector and segment's products, solutions, and services. The AI and HPC workloads have been converging, and a similar infrastructure is required for their use. Netweb also has its own mixed workload architecture to ensure that the seamless uses of supercomputers can be attained between supercomputing and AI. Integrating AI into supercomputing is driving innovations in algorithmic design, hardware optimization, and system-level improvements, leading to a new era of computational efficiency and sustainability.

The AI market is expanding in all aspects, i.e., the training and inferencing market. Earlier, I believed that the inferencing market would be much more significant in times to come. However, that is not the case so. Seeing the new and emerging uses cases, even the training market is not showing signs of slowing down and only accelerating.

With the semiconductor industry becoming increasingly competitive, how does the Production Linked Incentive (PLI) scheme in India aim to boost domestic manufacturing and innovation, particularly in relation to semiconductors, and what role might AI play in optimizing production processes?

As I said earlier, AI will play a pivotal role everywhere including the optimization of the production process. AI can enhance manufacturing efficiency through predictive maintenance, yield optimization, and process control, leading to higher productivity and lower costs. The integration of AI with semiconductor manufacturing could further India's ambitions by enabling more sophisticated and efficient production lines, ultimately contributing to the country's competitive edge in the global semiconductor market.

Coming down to the point of semi-conductor industry becoming competitive, it is just one side of the coin. The important thing to bear in mind is competence rather than the pricing in competitiveness. Setting up a fully functioning semi-conductor industry ecosystem requires expertise and decades along with massive capital expenditure. Fortunately, our policy is focused for the long-term with a vision to establish a domestic ecosystem with the entire stack of technology around it. The goal is to catalyze the entire semiconductor manufacturing chain, from design and fabrication to packaging and testing, thereby promoting a self-reliant semiconductor ecosystem within the country. As soon as domestic manufacturing begins, we will definitely see a competitive edge around the price also.

Can you elaborate on the collaborative efforts between Netweb and NVIDIA to establish GPU servers in India, and how do these initiatives contribute to the broader landscape of AI-powered computing infrastructure?

Sure, we have partnered with Nvidia to manufacture the NVIDIA Grace CPU Superchip and GH200 Grace Hopper Superchip MGX server designs that will help bring out the immense potential of AI in India and APAC. Our AI systems, based on modular reference design of NVIDIA MGX, will target complex workloads of HPC, data science, large language models, edge computing, enterprise AI, and design and simulation. The product range will also support handling a wide range of workloads such as AI training, inference, and 5G on a single system.

Grace CPU Superchip represents a significant advancement in the realm of data center processors, particularly designed to cater to the intensive demands of AI data centers that handle large volumes of data for intelligence processing offering a substantial leap in performance and efficiency for modern data centers.

GH200 Grace Hopper Superchip is engineered for high-performance computing (HPC) and giant-scale AI applications, marking a significant leap in computational power and efficiency. This superchip is particularly tailored to handle demanding AI workloads, including large language models, recommender systems, and vector databases, making it a cornerstone for the next generation of accelerated computing and generative AI.

By manufacturing AI systems based on the NVIDIA MGX platform, this partnership aims to propel advancements in various AI applications, including high-performance computing, data science, large language models, edge computing, enterprise AI, and design and simulation. The flexibility offered by these AI systems for a range of applications, such as generative AI, speech analytics, text analytics, and automation, is expected to address the evolving needs of enterprises effectively.

In what ways do GPU servers manufactured through the collaboration between Netweb and NVIDIA represent a significant step forward in enhancing computational capabilities, particularly in industries reliant on high-performance computing, such as data analytics, scientific research, and machine learning?

It is very significant because only very few OEMs across the world can manufacture such high-performing CPU- GPU based server design for different kinds of AI workloads including the mixed workloads for Nvidia. I feel happy that Netweb is the only company in India to have this distinction. Both Grace CPU Superchip, GH200 Grace Hopper Superchip represent the future of rapidly evolving chip technology which are to be used extensively in AI computing infrastructure. The collaboration between Netweb and NVIDIA to produce GPU servers is poised to significantly enhance the computational capabilities available to industries reliant on high-performance computing.

By providing advanced GPU technology, optimized servers for AI and ML workloads, scalable solutions for HPC applications, enhanced energy efficiency, and support for emerging technologies, this partnership is set to drive innovation and efficiency across various sectors. Together with Nvidia we are looking forward to addressing this market by manufacturing locally.

How do you foresee the integration of AI technologies into semiconductor manufacturing processes, facilitated by collaborations like Netweb's partnership with NVIDIA, shaping the future of the semiconductor industry of servers, both in terms of efficiency gains and technological innovation?

The integration of AI technologies into semiconductor manufacturing processes is poised to significantly shape the future of the semiconductor industry, particularly in the server sector. This integration is expected to bring about both efficiency gains and spur technological innovation in several key areas including enhanced precision and quality control, predictive maintenance, optimization of manufacturing process, supply chain and inventory management, innovation in semi-conductor design, and customization and rapid prototyping.

AI and supercomputing are already playing a crucial role in semi-conductor manufacturing. With the setting up of more new and bigger the magnitude of use is expected to increase further. But what is going to happen is that with the coming up of more semi-conductor unit, and other technologies such as Netweb’s AI servers and solutions, Nvidia AI chips, the overall AI addressable market and the diffusion of AI market will accelerate further with the adopting of many compelling use cases.