By S Ramadorai

In the rapidly evolving landscape of artificial intelligence (AI), hardware forms the bedrock upon which cutting-edge advancements are built. The future of AI hardware and high-performance computing is set to redefine the boundaries of innovation and industry. As semiconductor technology undergoes transformative shifts, the strategies that India adopts will determine how well we are able to capitalise on our AI ambitions. As highlighted in Chris Miller’s book Chip War, achieving technological self-reliance in advanced chip manufacturing is now a strategic imperative for countries across the world.

The semiconductor industry is often described as the most complex and precise manufacturing endeavour known to humanity. From the early days of vertical integration to the rise of fabless chipmakers and foundries, the landscape has evolved dramatically. Today, cutting-edge semiconductor technologies, such as those found in Nvidia’s H100 graphic processing units (GPUs), Google’s tensor processing units (TPUs) with the latest Trillium, and emerging neural processing units (NPUs), are enabling breakthroughs across domains, from AI and machine learning to autonomous systems and beyond. These advanced accelerators exemplify how the evolution from 7 nanometers (nm) to 5 nm, and now to 3 nm fabrication technologies, is enhancing computational power while managing power consumption and heat dissipation. GPUs excel in parallel processing, crucial for training deep learning models with large data sets, while TPUs are application-specific integrated circuits (ASICs) optimised for a high volume of low-precision computations. NPUs are another type of ASICs designed for accelerating neural network computations for specific, power-efficient tasks in mobile and edge computing devices. These advancements are unlocking new frontiers of growth and innovation.

India’s entry into the semiconductor space with Tata Electronics’ initiative to build a state-of-the-art semiconductor fab facility in Dholera, Gujarat, in partnership with Taiwan’s Powerchip Semiconductor Manufacturing Corporation, marks a landmark in our journey. The upcoming fab supported by an investment of `91,000 crore (~$11 billion) will provide an unprecedented platform for cutting-edge research and technological development. It is expected to create over 20,000 skilled jobs, underscoring our commitment to enhance our capabilities in becoming a key player in the semiconductor ecosystem.

These developments resonate deeply with my own hardware experiences at TCS five decades ago, when we worked with the then most advanced mainframes to establish our trademark computer-aided software engineering. In 2007, with the support of Tata Sons, we embarked on a historic journey to create Eka, India’s first more-than-100-teraflop supercomputer within a limited budget, which featured among the top four supercomputers. Eka was not just a technological marvel; it was a testament to India’s growing prowess in high-performance computing and ability to achieve the extraordinary. Eka saw several important applications subsequently — from the launch of India’s moon vehicle Chandrayaan by the Indian Space Research Organisation to developing new nanofluids to creating India’s first fully-animated 3D feature film.

The future of AI hardware

The trajectory of AI hardware, like that of supercomputers, is propelled by continuous advancements in semiconductor technology. The global demand for AI-specific hardware — ranging from training accelerators to inference chips — is soaring. Specialised chips designed for specific applications such as power management, telecommunications, digital signal processing, cryptographic acceleration, and so on are becoming increasingly critical. Advancements in ASICs, material science, and potential quantum computing integrations signify a paradigm shift. The industry’s focus on integrating heterogeneous computing elements — combining GPUs and NPUs on a single chip — presents opportunities towards maximising efficiency and performance for diverse AI workloads. AI-optimised field-programmable gate arrays are providing programming flexibility and adaptability to meet the needs of evolving AI systems in edge computing models. Moreover, neuromorphic chips that mimic the human brain’s neural structures are enabling autonomous adaptation in applications such as robotics and complex sensor networks. With such endless possibilities, the success of innovation hinges on the availability of skilled talent who can drive these advancements towards greater good.

Education and entrepreneurship

India boasts a significant portion of the global science, technology, engineering, and mathematics talent pool, which we must leverage to maintain our competitive edge. By focusing on developing specialised skills and learning ecosystems for very large-scale integration design, ASICs, and semiconductor technology, we can ensure our workforce is well-equipped to drive the next wave of AI advancements. Moreover, the integration of AI into diverse sectors will require a skilled workforce capable of leveraging AI hardware to address complex challenges and seize opportunities.

The start-up ecosystem, entrepreneurship, and policy support are equally critical. Government initiatives such as the Chips to Startup programme, the Atal Incubation Centre T-Hub Foundation, and the MoU between India and the European Commission on working arrangements on semiconductors ecosystems are commendable steps towards fostering a vibrant environment of research and innovation, and international cooperation. The Semicon India Design Linked Incentive scheme is offering financial incentives and infrastructure support to start-ups at various stages of semiconductor design development for integrated circuits, chipsets, and system on chips. Moreover, the government’s plan to set up a cluster of 25,000 GPUs as part of the India AI programme can provide considerable impetus to AI innovation.

India stands at a pivotal moment in its technological journey. The future of AI hardware holds great promise, and our ability to harness this potential will depend on our collective efforts to invest in cutting-edge technologies and nurture our talent. Our efforts in AI hardware development and deployment can spur a new era of India’s growth, similar to the software and IT revolution. As we forge ahead with new ventures in semiconductor manufacturing, let us remember that the path to greatness is paved with vision, innovation, and an unwavering commitment to excellence. By embracing these principles, India can not only solidify its position as a global contributor to AI and semiconductor technology, but also drive transformative change across industries and economies worldwide.

The author is the former CEO and Managing Director of Tata Consultancy Services (TCS)

Disclaimer: Views expressed are personal and do not reflect the official position or policy of FinancialExpress.com Reproducing this content without permission is prohibited.