It’s no secret that top tech companies are breaking the bank in looking to build their AI infrastructure, but it can still be stunning to realize how big these budgets can be.
Microsoft is spending $80 billion (Rs. 6.8 lakh crore) to build AI datacenters in fiscal year 2025, which will end in June 2025. “In FY 2025, Microsoft is on track to invest approximately $80 billion to build out AI-enabled datacenters to train AI models and deploy AI and cloud-based applications around the world. More than half of this total investment will be in the United States, reflecting our commitment to this country and our confidence in the American economy,” Microsoft wrote in a blogpost.
“Today, the United States leads the global AI race thanks to the investment of private capital and innovations by American companies of all sizes, from dynamic start-ups to well-established enterprises. At Microsoft, we’ve seen this firsthand through our partnership with OpenAI, from rising firms such as Anthropic and xAI, and our own AI-enabled software platforms and applications. Across the nation, a new generation of AI firms is emerging, each capitalizing on rapid advances in AI models and chips, moving now from Graphics Processing Units (GPUs) to AI Accelerators with Tensors. And across the economy, software programs are being redesigned to operate as AI-enabled applications,” it added.
Microsoft isn’t the only company that’s spending a massive sum of money to build AI data centers. Google has reportedly spent $3-4 billion in building AI infrastructure, and is spending $60 billion on training AI models. Meta’s capital expenditure for AI initiatives in 2024 was projected to be between $38 billion and $40 billion. OpenAI has spent around $8.5 billion in building server capacity and training AI models, and Elon Musk’s xAI has spent around $3-4 billion on building its own AI infrastructure. Microsoft and Google’s spends likely look much bigger than those of other companies because they also have cloud operations, and will likely rent out these GPUs for use to other smaller companies.
These are obviously staggering numbers, especially in the Indian context. Microsoft’s $80 billion spend on AI infrastructure in FY25 is five times the total funding of $14 billion raised by Indian startups in 2024. It is also more than twice the valuation of India’s most valuable tech company, Flipkart, which was last valued at $38 billion. And it completely dwarfs the AI budget of the Indian government, which is looking to spend around $1.2 billion to help build a datacenter in India with NVIDIA GPUs.
While it would be extremely hard for India to compete with the spends of top tech companies in AI, it doesn’t even seem to be actively trying. Experts like Nandan Nilekani have said that India doesn’t need to build its own LLM, and would be better served by finding use cases for deploying existing AI. But given how much importance the US tech industry is giving to create the best AI, one wonders if this is a wise move. AI will likely soon underpin all manner of knowledge work, and even critical sectors like defence and space, and being completely reliant on LLMs built in other nations could create significant dependencies at a time when India is looking to build an Atmanirbhar Bharat.