Anand Agarwal: India's IT Talent and the IP Ownership Dilemma
For 30 years, India has produced exceptional tech talent, but unfortunately, the platforms built by these engineers are not owned by India. The intellectual property often belongs to clients in San Francisco, London, or Zurich. Companies like Wipro, Infosys, and TCS employ over 1.3 million engineers, yet the value they create accrues elsewhere. This historical model, driven by cost arbitrage, needs a fundamental shift towards owning what is built.
The Evolution from Cost Arbitrage to Value Creation
The initial advantage for foreign companies setting up in India was primarily cost arbitrage. However, as Indian IT moved up the value chain from support to engineering, architecture, design, and product innovation, the model shifted. Companies like Unilever, Cisco, and Tesco demonstrated that designing and delivering entire tech stacks from India offered far greater returns and efficiency than mere cost savings. This evolution has led to a global functional alignment where reporting structures are integrated across geographies.
The Challenge of Building Indigenous Tech Infrastructure
Establishing India's own foundational tech infrastructure, such as mobile phones, laptops, cloud services, and operating systems, is a capital-intensive endeavor with a long gestation period (3-5 years). While capital is now available in India, investors often prefer quicker returns, leading to the acquisition of promising Indian companies by foreign entities. This trend, exemplified by acquisitions like Entity Data buying a data center company and Microsoft acquiring an HR systems firm, hinders the growth of independent Indian tech giants.
Sam Altman's assertion that India cannot compete in AI was a catalyst for Socket AI Labs, which was founded in 2019. The company aims to build sovereign frontier models for India's critical infrastructure, including defense, cybersecurity, AI assistive coding, and banking, to ensure national control and trust over essential systems.
The Data Deficit: A Multilingual Hurdle
Building AI for India faces a significant data challenge, as global datasets like MC4 contain less than 0.1% of India's 22 scheduled languages. Socket AI Labs recognized this gap and focused on curating data for Indian languages, understanding that data variety is critical for model performance and accessibility across the nation.
Socket AI: Beyond Languages to Agentic Capabilities
While Socket AI started with a focus on 22 Indian languages and the Global South, its evolution is towards building agentic systems with strong mathematical, coding, and reasoning capabilities. This shift recognizes that AI's future lies in autonomous action, not just conversational interfaces, and is crucial for applications in cybersecurity, defense, and banking.
Mohandas Pai argues that India needs a 50,000 crore rupees AI and deep tech fund over the next three years to compete globally. He emphasizes that current funding levels are insufficient compared to China and the US, and that this investment is crucial for scaling up Indian AI companies and preventing talent exodus. Therefore, substantial public money is required to foster AI innovation and ensure India becomes a significant AI player.
Government Inaction Hinders AI Progress
Pai criticizes the Indian government for its slow execution and bureaucratic processes in allocating and disbursing AI funds. He points out that allocated funds often go unspent due to procedural delays and a lack of private sector involvement in decision-making. Thus, Pai suggests that the government needs to streamline its processes and involve private sector experts to accelerate AI development.
Pai's Action Plan for AI Funding
Mohandas Pai outlines a concrete action plan for effectively utilizing AI funds, including forming an investment committee with private sector experts, inviting deep tech funds and companies to pitch for capital, and investing in hyper-cloud capacity. He emphasizes the need for quick disbursement and strategic allocation of funds to drive innovation and create a vibrant AI ecosystem. Therefore, Pai's plan aims to create excitement and transform India's AI landscape.
