Dharmesh Ba (@dharmeshba) header
Dharmesh Ba (@dharmeshba) profile picture

Tech • Design • Ethnography Weekly Indian consumer insights - The India Notes 🇮🇳 (19,000+ subs) 👇

Bangalore, India
Joined November 2011
Photo by Dharmesh Ba (@dharmeshba)Photo by Dharmesh Ba (@dharmeshba)Photo by Dharmesh Ba (@dharmeshba)Photo by Dharmesh Ba (@dharmeshba)Photo by Dharmesh Ba (@dharmeshba)Photo by Dharmesh Ba (@dharmeshba)Photo by Dharmesh Ba (@dharmeshba)Photo by Dharmesh Ba (@dharmeshba)Photo by Dharmesh Ba (@dharmeshba)Photo by Dharmesh Ba (@dharmeshba)
♥️
Some clarifications on various themes being discussed about @SarvamAI in the last few days: 1. Government Support: The IndiaAI Mission under MeitY has played a pivotal role in bootstrapping India's AI ecosystem. It gave ~20 startups subsidised access (40% discount) to GPUs for training models. We used that access to train Sarvam 105B from scratch, which we completed and open-sourced this February. The government did not pay Sarvam directly, instead like with every supported startup, the subsidy went to GPU cloud providers, helping seed AI infra in the country. 2. Sarvam Token Factory: Open-weight models saw massive adoption globally this year. They offer a cost advantage over closed models and let enterprises own their AI stack. But this choice is not available to all Indian enterprises, which often need data to remain within the country and low latency for interactive use cases. Sarvam Token Factory is one of the first large inference offerings in India. We serve multiple open-weight models on GPUs we have privately procured, without any government support. We have optimised datacenter scale inference to make our token cost comparable to or lower than established providers such as Fireworks, Baseten and Together. Today, India consumes a few trillion tokens a day and imports the majority of them. Scaling this consumption up and building sovereign capability requires investing in GPU infra, running efficient token factories, and guaranteeing advantages in cost, latency, and data residency. This is the key objective of Sarvam Token Factory. We will continue to reduce costs further and widen access for inferencing in India. 3. Our own models: Building inference capability does not conflict with our core mission: training our own models. Over the last few months, alongside putting together the token factory and scaling up our enterprise offerings, we optimised our training stack for Blackwell GPUs and achieved significantly higher performance. We used this optimised infra to train an improved checkpoint: Sarvam 105B-1711, which we are open-sourcing today. It improves on reasoning, tool-calling, and co-work tasks, while excelling in Indian languages. It is one of the best Flash-Lite class models, open or closed, achieving high accuracy at a low cost of ₹25 per million output tokens. We also optimised it to run on a single RTX Pro GPU with our agentic product Indus. Along with the weights, we release inference and training recipes so others can build upon it. In the coming two weeks we will release multiple other models across vision, speech, and edge. And crucially, we are working with partners to get the latest training grade compute into India. This will set us up for training our next generation of much larger models. 4. The Sarvam Journey. To put things in perspective, most AI labs invest 100s of millions of dollars into infra to train from scratch, scaling up to billions for frontier capabilities. We are determined to climb that ladder from India, while building a sustainable commercial engine across population scale and enterprise use cases. Irrespective of the different narratives, some of which appear manufactured, we firmly believe that India can and must win across the full-stack of AI - we have the talent, the demand, and positivity to adopt AI. At Sarvam we will stay focused and build towards this outcome.
4
399
What do you meen?
4
2
50
5,749
Meet Kyoto. Our new addition to the family.
She was found unwell in a Bengaluru park. Now she is Kyoto, and she is home with a couple who chose her. Her rescuer Paawan Mukker kept looking until the right family came along. Thank you, Paawan. Adopt, don't shop.
26
7
234
4,749
Thanks @FernFly for great work!
5
327
Manjunatha Travel Agents, Kadubeesanahalli Works with WhatsApp and COD.
What is the best agentic flight booking experience currently out there?
14
51
2
1,175
47,448
I was in Coimbatore a few days back. Booked an auto on Rapido. I had gotten used to paying Rs.30 extra than shown amount that week. That night, the app itself showed Rs.143 for 2.8 Kms and I was ready to pay that. Driver shows up, casually demands give Rs. 40 extra. I got pissed and asked him to be reasonable to have some base price for one Km. I asked him if even Rs.50 per Km isnt worth taking a passenger? and told him cab shows lesser fare. He replied 'In TN cab is cheaper than auto thats the truth' and he chose to take me for shown fare on app. What's the point of base price raise in auto if the demands are still the same? and on top of this these thuggery to ban bike taxis.
அண்ணா விட்ருங்கண்ணா.. Rapido ஓட்டுநரை கொடூரமாக அடித்த ஆட்டோ ட்ரைவர்.. பயத்தில் அலறி கத்திய பெண் #Rapido #RapidoDriver #RapidoBikeTaxi #RapidoDriverAttack #AutoDriver #AutoDriverAttack #RapidoVsAuto #BikeTaxi #BikeTaxiIssue #AutoRickshaw #DriverFight #PublicSafety
12
11
161
30,578
When I landed in Bangalore, early morning, I travelled 7 KM in Rapido for Rs.120.
1,019
It doesn’t matter how many voice dictation apps you have your phone. You still need something interesting to say.
3
3
32
2,013
Indian parents after marrying off their daughter.
wtf this looks like a horror movie
4
3
173
8,606
A BNPL product to borrow AI tokens Calling it ‘Token advance’
4
2
47
2,743
short-sighted strategy.
Banshankari 2nd Stage has 2 Lenskart shops just 100 meters apart !! What is the logic behind this ?🤨
30
1,671
From bored times to Time’s board! Congrats @TheProductfolks
1
13
875
♥️♥️♥️
1
1
16
2,212
Dharmesh Ba retweeted
Tax cuts were welcome. But we don’t want more handouts. What industries really need: • Real crackdown on corruption and bribe culture
• Genuine ease of doing business, not just rankings
• Drastic cut in paperwork and inspector raj
• Action against smuggling and black markets that undercut honest factories
• Faster land, power, and logistics clearances
• Predictable, stable rules instead of constant new compliances Give us a clean playing field and we’ll make sure employment and salaries skyrocket. 🚀🇮🇳
#BREAKING PM Modi questions corporate India: Tax cuts given, but where are the jobs and salary hikes? Addressing FICCI, PM Modi reminded business leaders that corporate tax was cut from 30% to 22%, while individuals now pay no income tax on earnings up to ₹12 lakh under the new regime “You always ask for more. But did you hire more employees? Did you increase their salaries?” Tough question to India's biggest businesses!
199
431
45
2,039
47,624
It’s simple. You can never lose a market that you never had.
doesn't really matter how good gemini is the fact that @google actively bans gemini/antigravity subscription accounts for using other coding tools (i.e @opencode, cliproxyapi, pi, etc), forcing end users to pay API prices, is a complete non-starter do better please
1
3
1,230
Rameshwaram Cafe has a gotten a court order that could allow them to send legal notices to places that resemble them. Damn. I don’t know how many temples will get the notice. 😪
28
48
5
1,979
97,125
India's ₹9,000 smartphone has almost disappeared. A year ago, 16 in every 100 phones shipped in India cost under ₹9,400. This quarter it's 2. The average price hit a record of about ₹34,900, but 23% fewer phones shipped and total spend fell 3%. Rising memory costs priced the cheapest phones out. 6 in 10 festive buyers who did buy paid with EMI. Only one price band really grew: ₹38,000–56,000, up 80%. The record price came from fewer buyers, not richer ones. If you've seen this in your city or your family, tell me in the replies. Sources: IDC (Q3 2026), AIMRA, The Economic Times.
3
27
2,032