India's AI Landscape: Thriving in Applications, Lagging in Compute Infrastructure

VCs bet big on AI application startups

India's AI landscape is rapidly evolving, particularly in application development, but remains behind in AI chip and GPU development due to high capital requirements and market dominance by companies like Nvidia. Venture capitalists are focusing more on software development and infrastructure tooling, recognising India's strong talent in these areas.The artificial intelligence (AI) landscape in India is witnessing a flurry of activities, but mostly within the application layer as development in the crucial compute side, specifically in AI chips and graphic processing units (GPUs), remains nascent, top venture capital firms told ET. "In the compute layer, in GPUs, we're not seeing any companies yet," said Rajan Anandan, managing director at Peak XV Partners that has made investments in more than 30 companies in the AI space. "We have invested in semiconductors, but not AI chips. In the US, we are seeing a fair bit happening even in that layer." Specialised chips and GPUs are key to process huge amounts of data and carry out complex computing involved in AI workloads. While India has a lot of catching up to do here, it is a highly capital-intensive area, limiting the scope for startups that are instead focusing on the application side of how AI tools interact with users. Abhishek Prasad, managing director of Cornerstone Ventures, said GPU is the hardest layer in the AI ecosystem to crack. "It's also a tough one for startups to be able to make a dent simply given the capital requirements," he said. "The market leader (almost unchallenged) is Nvidia, and it's near impossible to challenge them in this space." It is a different story in the application space. "The application layer dominates the AI landscape in India, as it also plays to the historically natural strength of our ecosystem," said Prashanth Prakash, founding partner of Accel India. "The lack of established domestic hardware ecosystems and high capital intensity of chip manufacturing make it less appealing for VCs compared to other layers of the AI stack. Indian startups are more aligned with software development and IT services," added Brijesh Damodaran, cofounder and partner at Auxano Capital. "Indian startups often provide services that enable the adoption of AI for global markets," he said. Infrastructure & tooling Infrastructure and tooling comprise technologies, platforms and tools involved in developing and deploying AI models and scaling them up, an area that Anandan says is seeing a lot of activity. However, according to Prakash, there have only been a handful of globally competitive infrastructure and tooling layer startups from India, "even from the pre-AI era". These areas (newer types of AI chips, data centre infrastructure, tools and software) need very specialised and experienced talent, hence the number of such companies will not be very large, said Arjun Rao, general partner at Speciale Invest which is making investments in this space. A company that is involved in the segment is fabless semiconductor startup Morphing Machines, developing runtime reconfigurable dataflow processors that deliver high performance and power efficiency at a low cost. It has raised $2.76 million in its seed round led by Speciale Invest. Despite the application-centric focus, several players are recognising the importance of infrastructure and tooling. "We at Speciale are strongly focused on the infrastructure and tooling layer in GenAI, mainly due to the technology differentiation and talent advantages there," Rao said. Speciale's portfolio includes companies like Protecto AI focused on secure and trustworthy AI and Tursio that transforms enterprise databases into generative AI machines. "We see infrastructure and tooling companies as the picks and shovels to provide the underlying foundation to build robust applications," said Poorvi Vijay, vice president of Elevation Capital. But broadly, Rao said, VCs are investing heavily in the application layer, given the breadth of opportunities in horizontal (which are designed for a wide range of uses across fields) and vertical (developed for specific industries and purposes) applications. Challenges and opportunities Development of AI chips in India faces significant hurdles. "It's not an easy game, requiring significant capital and, most importantly, close collaboration with pioneering system customers who will use these chips," explained Somshubhro Choudhury, cofounder and partner at Bharat Innovation Fund. "Beyond designing the chip, it's also important to create an ecosystem of software infrastructure on top for easy adoption of the chips in electronic and data centres." These challenges also present unique opportunities. "We're seeing the most vibrant startup activity in India at the infrastructure tooling and application layers," said Sharvi Dubey of the investment research team at Good Capital. "However, it's important to understand this isn't a limitation but rather a reflection of India's unique market opportunities." Dubey believes that Indian companies can gain a competitive edge by focusing on solving India-specific challenges through AI-powered applications. While the development of AI chips may still be at a nascent stage, the future of AI in India appears bright, with a strong emphasis on application development and a growing focus on critical infrastructure and tooling. As the industry continues to evolve, it is crucial for Indian players to navigate the challenges and capitalise on the unique opportunities. By fostering a strong ecosystem of innovation and collaboration, India can emerge as a global leader in the AI revolution, experts say.

VCs bullish on AI startups in India despite a fall in funding

Indian AI startups saw a funding dip in 2024, but deal volume increased, with a focus on application-layer ventures. Investors are optimistic about this trend, particularly for applications targeting the US market, mirroring the enterprise software sector. While foundational models remain a US stronghold, India sees opportunities in infrastructure and applications, despite challenges in talent acquisition and rapidly evolving technology.Venture capital investors are bullish on application-layer artificial intelligence (AI) startups in India despite an overall fall in funding amount for companies in the sector this year. According to data from Venture Intelligence, Indian AI startups raised $747 million in 2024, down 26% from last year but the number of deals in the space increased to 121 from 78 in 2023. "This year, we saw many companies building vertical AI applications by leveraging more mature platforms to build on top of that," said Ravi Srivastava, partner at early-stage investment firm Leo Capital. There are broadly three layers in the AI stack -- the foundational models, horizontal layers called enablers, and vertical AI applications. Leo Capital closed multiple deals in the AI and enterprise space this year, including Zeplyn, an AI copilot platform for financial advisors, Arch0, a cloud infrastructure security startup, and DecoverAI--a platform built for the legal community. Notably, several AI startups being backed by investors in India are building in the US. Peak XV Partners managing director Harshjit Sethi, who focuses on AI investments for the VC firm, said that AI startups are expected to follow the trend set by enterprise software companies, which mostly build for US customers. "What we should expect to see, and my hope for next year, is that there will be many more interesting application layer opportunities," said Sethi. "Taking the capabilities of AI and translating them into what it means for each of our roles, I think that gap will be something companies will look to address, particularly those in India." According to Tracxn, Peak XV Partners, through its various programmes and the main fund, backed 13 startups building in AI space this year, including Enterpret, RapidCanvas, OrbitShift and Brainfish. Peak XV also participated in the $105 million round for SaaS startup Atlan, which is aiming to capitalise on the opportunity from large corporations looking to make their databases suitable for artificial intelligence. Big-ticket AI deals In addition to a series of small-cheque deals in the AI space this year, some of the larger deals included Mukesh Bansal-founded enterprise AI assistant platform Nurix AI, AI cloud and platform-as-a-service startup Neysa, and Bhavish Aggarwal-led Krutrim AI, which became a unicorn, a privately-held company valued at $1 billion or more. Industry executives said in addition to the application layer, the opportunity for India is in the infrastructure layer, having missed the chance to compete on foundational models that are largely being developed in the US by large companies such as OpenAI, Meta, Google and Anthropic, which have raised billions of dollars in funding. Neysa, which operates in the infrastructure layer, offers generative AI platforms and services to assist clients in managing AI projects. It raised $50 million this year across two rounds. Founder Sharad Sanghi pointed out that while the process was easier for them due to his experience with data centre services provider Netmagic, building the infrastructure layer is challenging for new startups. As a result, most new AI startups are focusing on consumer and enterprise applications. "Because of our relationship with NTT and my experience founding Netmagic, I had access to data center space and power, which not everyone has. Access to capital is also crucial, as well as access to the right talent. You need highly skilled engineers on both the infrastructure and platform side. This talent is available, but it's expensive and hard to attract," he told ET. Computing infrastructure key The Indian government took a major step by approving a Rs 10,372-crore allocation for the India AI Mission for the next five years, focusing on building computing infrastructure through public-private partnerships. However, VCs are of the view that the hype over AI has created a glut at the inception stage with regard to the applications layer. "Technology is changing so rapidly that the notion of 'problem' and 'value' is getting redefined every quarter," said Alok Goyal, partner, Stellaris Venture Partners. "It is not clear if what is differentiated today will remain so in just a few months. Secondly, entry barriers have become much lower, and hence in every space, there is a clutter at the inception stage of the category itself. This is particularly so for AI applications."

A dichotomy that's playing out in GenAI investments in India

Despite the palpable enthusiasm around GenAI in India, investments in the space declined significantly in 2024. If you were a regular in India's tech circles, you would have noticed one thing - most of the conversations start and end with artificial intelligence (AI). And understandably so. Since the launch of ChatGPT in 2022, technology is changing the way companies and workplaces function. This has caught the venture capital (VC) circle buzzing. Peak XV Partners have invested in over 25 AI companies this year and the sector will continue to be in focus,