akt (@showdinger) profile picture

A physicist, currently at Indian Institute of Science

IISc, Bangalore
Joined April 2011
Photo by akt (@showdinger)Photo by akt (@showdinger)
They wont be able to make foundation model even if they get 1 trillion dollar investment. They need something which is more than money which they lack.. There is a proper term in economics for this mindset called "Scarcity Mindset".Read about it from Chinese mdel hosted at sarvam
If @SarvamAI intends to compete for global capital and become a frontier lab, they need access to large compute capacity and captive data pools. They also require permanent tax holiday if specific technical thresholds/capabilities are met. @GoI_MeitY can guarantee all of these. Getting turned down by top American PEs and VCs can be largely expected. Some of it is down to the US’ unwillingness to financially back a sovereign Indian AI venture. The rest will be down to the financial opportunity simply not being sufficiently appealing. That said, getting rejected by SoftBank should set off alarm bells.
3
Will RTI tell which code was used and for what purpose. Kon sa code kiske liye use hua hai.. who will tell. This fact establish nothing !!
Someone please file an RTI to verify the GPU given via IndiaAI Mission.
19
Galgotias showed Chinese robots at India AI Summit, poor professor lost her job.Nowsame summit's poster boy hosts Chinese models and the PR agency calls badge of honour. Next: ship Chinese models to space data centres, or ship joke tokens to band-baja-baraat India AI Missn
Congratulations to @SarvamAI on launching Token Factory for everyone, with Sarvam’s own models and leading open-weight models hosted on GPUs in India. Data residency requirements and a lack of inference in India have kept many teams here away from the best open models. Token Factory serves them from India, with performance and reliability that match global providers. For enterprises, it means more models to choose from and an AI stack they can own without lock-ins. @pratykumar | @vivekrag | @MohapatraHemant | @BaggaRohil
20
After so much PR what next gimmick after token factory.. astrology services or Baba ji ka thullu aka data centre in space hosting Chinese models !
Huge congrats to @pratykumar on making the inaugural TIME100 Next India list, and to @vivekrag on making TIME100 AI earlier this year! 💪🏼💪🏼 So proud to back you both and the whole @SarvamAI team! cc @vkhosla @khoslaventures time.com/collection/time100-…
23
Most important AI company from India.. hahahaha.. what a joke.. TIme magazine is known for not knowing about India and now some Deepseek trained experts too !!
unavailable
23
For guys like these, only profit matters whether selling Chinese models or selling Astrology services or selling country ! What happened to your data centre in space gimmick !
unavailable
13
To save India now digital Bhagat Singhs will host Chinese AI
🚨 Sarvam Co-Founder Warns 🇮🇳 India Against Becoming a 'Digital Colony' India’s position as one of the largest users of tools like ChatGPT should not be seen as a measure of AI success, said Sarvam co-founder Vivek Raghavan. “We are exporting our data and importing intelligence,” he warned at the Bengaluru Tech Summit, urging the country to build its own AI capabilities to avoid becoming a digital colony.
36
One thing which certainly Sarvam has mastered is making fools at scale .. right from government to India AI mission to some flatterers
I've been bullish on sarvam since 2024, when they had SLMs for basic translation on huggingface (made this basic ahh assignment to propose a career advice platform for rural students to get an internship lol). Even if the moat is that all Indian user data stays here instead of going back to somewhere else, it's a big enough reason to be supportive of what they're doing. and the most depressing part is all of the hate is coming from Indians, when we have so many ai trainers who are annotating the data for most of the frontier labs. Sarvam does have a moat thanks to India's diverse culture and languages, and I don't think there's any reason to be negative. there are sooooo may use cases to setup models that are good at indic context (most government contracts, procurement data etc is still in Indian languages) Crab attitude never comes from people who know what they're doing lol.
18
And these jokers judge AI founders on live TV !!
Indian billionaire entrepreneur Aman Gupta launched a face wash after betting his money on R&D with handsome man.
12
Even Chinese models have got confused and have gone into inference loop how to make a post emotional with tadka of Nationalism !!
sarvam is yet to post about pratyush kumar being in times 100 but the VC that invested in sarvam already did :p
2
21
Scratch ??? Really.. what Sarvam has built from scratch ? Transformer architecture, State Space Architecture, Graphics Card, CUDA, Pytorch .. what is from scratch.. Oh got it .. equation on how to lobby for funds and block other genuine startups !
Sarvam is building Sovereign AI, most of it's models are built from scratch, so I think similar firms should get govt funding (definitely not ones fine tuning only) and who better to run these other than someone who guides/ signs off on phds.
41
Really surpised how people get on time magazine for selling chinese pizza !! sab kuchh bikau hai
Congratulations @pratykumar on this well deserved recognition from @TIME. We are incredible excited about the company that you, @vivekrag and team @SarvamAI are building! @peakxvpartners
19
Congrats on your enlightenment. Finally you have got ultimate "flattery nirvana" via deepseek.
All these dudes crying "whys sarvam hosting Chinese models" need to understand it's a good way to utilize capacity, serving models people will actually use.
1
131
What next @SarvamAI @pratykumar .. Astrology services ? That is also legal and perfect business model too !!
65
likh ke rakh lo and save this for the future "beta tumse na ho payega" even if they get 1 trillion dollars !!
India's most prominent builder of sovereign AI models, Sarvam, will offer DeepSeek and other Chinese AI models on its platform. Its India sovereign AI mission remains intact, founders say in an interview. bloomberg.com/news/articles/… #AI
29
Incompetency and Impotency is much bigger problem than hosting Chinese AI models. This clearly shows they have diluted the vision and never had the substance to challenge the big techs. And worst of all they lobbied, blocked other small companies from getting to India AI mission
Training a LLM model is hard, very few companies in the world has done this. It needs courage to even attempt one. Sarvam, Gnani and others from India are trying it. I think we should encourage them. By nature, success of startups is very hard, 90% plus failure rate and that increases when you want to try harder things. Is serving LLM's profitably easier, I don't think so. If Indian companies do not succeed in building great products that produce tokens efficiently and economically, then all our companies should import them, which has much larger implications. Few of us are making an effort, in trying to change that. Will we succeed, may be yes/no. Startups are famous for doing things against the odds. We all will do our best to challenge the status quo. How could you serve Chinese models: Because China was able to produce some of the best open weights models. Customers wants to solve problems, with whatever tools work. If we look around, we all can notice a lot of things that are built in China, US and other countries. As new models come up, which are better they replace. The origin of the models just plays a part in wether they are served are not. As a country, should we not celebrate Entrepreneurs and support them when they fail.
2
77
What it’s like to bleed over an 80-page proposal, tear through a thousand research papers, only to be completely shut out because some connected lobbyist blocked the door? Sarvam has suffocated countless small startups & they can't even deliver a competent product. @SarvamAI
29