As the world leaves the Compute Era and enters the Diffusion Era, India – its companies, its institutions and governments – must ask three questions

The world is competing over AI. But this will not remain merely a race towards a smaller chip, a larger data centre, or a faster model. The winner will be determined by those who turn innovation into impact and convert technological advances into product value, institutional capacity and community trust. Power in this era will flow not only to those who own algorithms or command fields of silicon, but also to those who find ways to weave AI into daily life at the national or continental scale.

Three distinct processes and evolutions compose this AI age. The first is the Compute Era, in which chip supply chains and hyperscale data centres determine market share. It tilts towards those with capital, energy and traditional resources such as land and water. The second is the Diffusion Era, where nations with population-scale digital platforms and deep real-economy deployment shape how and where value is created, captured, and compounded. The third is the Governance Era, in which sovereign power and political legitimacy — as well as geopolitical leverage — pick winners among AI systems.

As talk of overinvestment in data centres and supporting facilities grows, we may see the peaking of the Compute Era. Computational resources will become a utility — invested in like legacy utilities, priced like them, and regulated like them. Such a path has been followed many times: The introduction of a new technology, a boom period of sensation and overinvestment, followed by an overhang in which the physical, corporate and human-capital remnants of the boom are transformed into core infrastructure for broader innovation and growth.

What you do with compute infrastructure is what will create value in the Diffusion Era. Nations like India — experienced at adopting and adapting technology at scale — will have a natural advantage. The benefits of compute as a utility will flow to those who can build applications that create value for the greatest number. Trust will be the key. Success will depend upon how effectively kinship and trust can be nurtured, and how much value a community sees in a product. LLMs will have to localise. Large models are trained on the Common Crawl, which over-represents Western data and misrepresents the cultural context of countries like India. The Diffusion Era will seek to correct that. New winners will emerge from within communities. Beyond the horizon lies the Governance Era. Governments will respond to the Diffusion Era by guarding data sovereignty more zealously than they have hitherto. They will seek to control the utilities left over from the Compute Era, and intervene in the competitive dynamics of the Diffusion Era. Such a return to sovereign power is almost inevitable for a boom-bust-diffusion technology.

As the world leaves the Compute Era and enters the Diffusion Era, India — its companies, its institutions and governments — must ask three questions.

First, can Indian companies take the necessary risks? Do they have the capability to bet on an unknown future? Smaller enterprises have demonstrated they have the entrepreneurial nous to do so. But will the broader private sector be able to seize the benefits of the Diffusion Era? It will require a change in mindset, in corporate governance, and risk appetite.

Second, can Indian institutions and finance find, energise and incentivise capital to create systems that can bet on unknowns and undiscovered innovations? Can we ensure that patient capital emerges from India and the world that can support and develop infrastructure at a utility scale?

And can the government create a world-beating regulatory structure that blends private profit and public good — like, for example, the Digital Public Infrastructure model? Can it protect individual rights and digital sovereignty without stifling entrepreneurial fever? Can we build an India-specific AI architecture that sits comfortably within and atop a global AI system?

If India can answer these three questions, it will dominate the threefold era of AI. Let’s build compute infrastructure, unleash entrepreneurial communities, and create supportive regulation.

Source: Originally Appeared in Indian Express