|
The Collection Wed, 13 Aug 25 |
Multiple stories, multiple perspectives, one theme worth your time—every week. |
Why India missed the bus on laying its AI foundations. And how it is now scrambling to catch up
| 4 min read |
Good Morning [%first_name |Dear Reader%],
You are on a free plan. Your subscription has expired. Upgrade now to unlock premium newsletters, top feature stories, exclusive podcasts, and more.
Welcome to this week’s edition of The Collection.
For a country that likes to see itself as an information-technology powerhouse, or at least well on the way to becoming one, India’s track record in AI so far has been nothing to brag about.
Sure, Indian Linkedin is awash with a neverending stream of AI hacks, tips, tools, use cases, and updates from scores of teams all working on some AI application or the other. But almost all of them are being built on foundations laid outside its borders.
Claude, ChatGPT, Deepseek, Llama—the list of the world’s most applied foundational AI models does not yet feature an Indian LLM (large language model), though there’s certainly been more than enough noise on the subject over the past couple of years.
All of this effort, though, to be blunt, has been an exercise in trying to catch up.
Just this week, the CEO of the IndiaAI mission—the entity tasked with bringing about our very own homegrown AI revolution—announced that the country’s big LLM project will try to differentiate itself from the rest of the pack by focusing on two specific areas: voice-first AI and Indian languages.
A few weeks before that, the government made another big announcement: India’s LLM project would now decisively shift toward the open-source arena.
If you don’t know why this is a big deal, you should read my colleague Sumit’s incisive newsletter edition this Monday on why open source could be India’s best chance of gaining real ground in the LLM race.
Free code, it turns out, isn’t just a nice academic gesture, or only about economics. It’s a geopolitical tool.
China understands this. Models like Deepseek and Qwen are instruments of soft power. These models, open in weights and widely adoptable, win global developer mindshare, define the AI stack, and make it harder for closed systems to charge rent.
Which is why the choice of licence in India is not some obscure legalism. In the nation’s increasingly politicised AI discourse, it’s the difference between models you can build on, fine-tune, commercialise, and reuse—and models that are walled gardens, with maybe a nice public park out front.
Even here, though, India’s initiative comes late, and key aspects such as licensing norms remain unoptimised.
India risks missing that moment if permissive licensing doesn’t become the default. Without formal licensing norms, many startups may opt for restrictive models, fragmenting the ecosystem. Opportunities to capture developer mindshare, shape standards, and scale fast could be lost.
“Public money, public code” is becoming policy logic globally. If India continues to treat open licensing as optional, it may find itself outpaced…
Free code can build India's AI fortune
The narrative that India must build the next GPT‑5 or Gemini‑X to compete in AI misses the real lever
For a more contextual look into the different open-source strategies at play in India, China, and the US, and how India has failed to capitalise on early open-source projects, you should also give a listen to this Two by Two episode from last month.
Two By Two • 50 |
|
India only fully committed to the LLM race in early 2025, by which time most of the major LLMs in play right now had launched at least one commercial version.
But it did set aside quite a decent war chest—Rs 10,000 crore—to build its own AI ecosystem, including its own foundational models.
The government has since given funding approval under the IndiaAI Mission to a select number of firms working on foundational models, and is even offering a 100% compute subsidy to such projects.
But initially, its strategy was a lot narrower. In fact, for a while, it put all its AI eggs in just one basket.
India’s AI Mission needs many heroes. But it’s settled for one—Sarvam
India’s big bet on its sovereign AI model is off to a familiar start: one winner, many confused contestants
And its overall plan read more like a spray-and-pray playbook.
Your guide to India’s Rs 10,000-cr AI mission that’s running on FOMO
Taxpayers' money is spent on looking for 'one' breakthrough moment, aka Deepseek. Building competitive advantage and solving large problems could be surprise by-products
This is not to say there is nothing interesting going on now, though.
Some companies such as Zoho are taking a contrarian approach to their LLM projects—mostly to avoid the massive costs associated with infusing AI into large SaaS platforms.
Zoho’s LLM isn’t trying to win the AI race. It’s trying to survive the SaaS war
Keeping costs low is a priority for the bootstrapped company—now, even more so, when cloud and AI costs are inflating SaaS prices
And a large number of LLMopps firms are also deploying, optimising, and gluing foundational models together for a number of use cases.
Deepseek cracked open AI. India’s AI plumbers are loving it
Until recently, some of the best-funded Indian GenAI companies used the tech to build new apps. Now, the tide is turning in favour of startups that help other companies build with GenAI
But without a foundational model to call its own, India and its companies risk both geopolitical vulnerability and the heavy costs that come with leasing massive amounts of foreign AI firepower.
You can, as always, find this week’s entire collection below. Do write to [email protected] or leave a comment on our website and app if you’d like to share your thoughts and suggestions.
Get a premium subscription to The Ken
Unrivaled analysis and powerful stories about businesses from award-winning journalists. Read by 5,00,000+ subscribers globally who want to be prepared for what comes next.
Trusted by 5,00,000+ executives & leaders from the world's most successful organisations & students at top post-graduate campuses
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Rahul leads the promotion and distribution efforts for the journalism produced by The Ken. He works at the intersection of the publication's editorial and business operations.
|
|
Share this edition
You can use the easy share buttons below
|
|
|
|
|
© 2025 The Ken
Do you know anyone else who would like to read this newsletter?
Share this edition with them.