 (3).png)
8 hours ago
2
ARTICLE AD BOX
Summary
Big Tech is investing billions to revolutionize AI by drastically reducing training costs and enhancing deployment efficiency. As Google and Microsoft shift their focus to selling AI at scale, how is India positioning itself in this rapidly evolving landscape?
Big Tech is spending billions of dollars to lower the cost of training and running AI models. For these giants, controlling all parts of the AI ecosystem gives them a significant cost advantage. Can this help them sell AI at scale globally? Where does India stand in this journey? Mint explains.
Why is Big Tech fixated on making AI cheaper?
When ChatGPT took over the world on 30 November 2022, generative AI was too expensive to run at scale—about $20 for a query of 1 million tokens (units of data that are processed and generated by AI models). The cost was simply too prohibitive to deploy at scale. As a result, companies conducted pilot projects with limited budgets over the past four years to see proof of the technology.
After three full fiscals, tech firms realized they cannot monetize AI unless they can run the tech at scale for a fraction of the cost. Therefore, both Google and Microsoft’s annual events this year focused on selling AI rather than fundamental innovation.
Does this mean AI will become more affordable for all?
Yes. Brokerage BMO Capital Markets said on 10 June AI will continue to become cheaper. It said token prices have fallen by 75%, which means companies can use 3-4x more AI workloads than before.
Graphics processing unit-maker Nvidia expects custom chips to crash AI costs further. AI is projected to become significantly more affordable in the next three years. This is key for AI companies to sustain their work.
Anthropic expects its first profitable quarter this June—after having raised $133 billion ( ₹12.6 trillion) to date. OpenAI, having raised $190 billion ( ₹18.2 trillion), is yet to show profits.
Which companies are leading the AI ecosystem—in India and abroad?
Globally, Google is ranked by most, including BMO, as the top ‘full-stack’ AI company, meaning that it develops everything from chips and AI models to data centres and applications for both enterprises and consumers. Amazon comes a close second, with greater focus on businesses. Microsoft, through Windows, is trying to catch up fast. In pure-play AI models, OpenAI and Anthropic are undisputed.
India, though, is far behind—government-funded startup Sarvam showed early AI models, while Sequoia-backed Avataar only just launched its video generator app and model. Progress in India, compared with the US, is lagging and sporadic.
What do falling AI costs mean for India's IT companies?
Cheaper AI can be both good and bad for India’s IT companies. So far, top IT companies have seen limited client projects at scale from generative AI. With costs falling, IT companies may be able to sell large AI deployment and transformation deals to their top clients—akin to how the cloud migration and digital transition phases played out.
However, cheaper AI also means it may be easier to replace process-based human jobs, which could be a threat to India’s IT workforce and equity returns. The job replacement may not be immediate, Bernstein said in a note on 11 June.
Does cheaper AI mean India's IT stock sell-off will reverse?
It is difficult to immediately predict that. Tata Sons chairman N Chandrasekaran expects a better run once businesses start to spend on AI and AI itself shows clear returns.
In India, the BSE IT index has fallen 8.45% since ChatGPT was introduced in November 2022, marking a CAGR of -2.46%. In comparison, the tech-heavy Nasdaq index in the US returned 26.5% in CAGR in the same time as US companies invented the foundational tech used worldwide.
Indian IT’s reversal of fortunes will depend on whether service integrators can prove themselves necessary. Bernstein argues that large IT players will be needed as implementation partners as companies adopt AI.
English (US) ·