Computing Crisis Facing India

India is grappling with the issue of compute scarcity, a problem rooted in the high cost of raw materials and specialized equipment for silicon chip manufacturing, the shortage of skilled professionals, and the concentration of these resources in the hands of a few private corporations. This scarcity is particularly concerning as access to compute is considered a strategic geopolitical asset, given the promise of technological progress.

The Indian government has recognised compute as an essential component of its AI strategy since 2018, but it is only recently that specific steps have been outlined to enhance its compute capacity. To address the issue, a comprehensive survey is proposed to measure existing compute capacity and projected needs over the next few decades.

India's compute infrastructure is optimised for specific applications using smaller models, a strategy that aligns with the country's use-case-driven AI strategy. The value of large, general-purpose models is limited in this context.

Compute, in technical terms, refers to the number of calculations that can be performed by a processor, usually measured in floating-point operations per second (FLOPS). The world's most powerful supercomputer clocks nearly 1.2 exaflops, which is more than a billion-billion calculations per second.

India has chosen to partner with US companies such as Google, Meta, and OpenAI for collaborative AI through strategic cooperation and infrastructure projects. Reliance Industries is working with Google and Meta to build a national AI cloud, while OpenAI engages with India as a partner for infrastructure, education, and innovation in AI.

However, the debate continues on whether India's national AI objectives can be achieved through small, custom, and open-source models designed for specific use cases, which will require less compute, instead of large, compute-intensive models that are designed for general use. Studies have shown that the performance of a large language model improves with an increase in the compute power made available to it.

To democratise access to its compute resources, India is considering strategies such as a global repository similar to a "CERN for AI" or the AI-on-demand platform (AIoD) in the EU, and a digital public infrastructure (DPI) for compute. This would provide equal opportunities for academics and startups to leverage compute capabilities.

Scaling up compute capacities is a strategy India is considering, in contrast to China's approach. However, with the factors of production being concentrated in the developed world, there is a concern that the digital divide between the Global North and South will continue to widen.

India has set ambitious targets to achieve net-zero emissions, but the desire for more compute may run counter to these environmental goals. Therefore, a deeper analysis is needed on how India's compute capacity should be enhanced and whether it is the right goal to pursue.

In the spirit of "collaborative AI," India is also looking to partner with like-minded countries on compute, cybersecurity, data sharing, and research. This collaboration will not only help bridge the digital divide but also contribute to technological progress on a global scale.