Congrats on the raise! Avoiding the mistakes of Krutrim now (what happened to them?), it is critically important for Sarvam to crack GTM globally now, and fast.
Some initial thoughts and opinionated things to do in the next 6 months:
➡️ Drop/phase out "AI for India" slowly. Yes, it is AI company built in India, not just for India but for the world. Every single frontier lab talks about their "frontier" AI on homepage, only Indian labs talk about sovereign and India on the homepage.
➡️ Double down on coding/programming models/AI harnesses. There is no mention of Sarvam for coding anywhere. Quadrillions of tokens are spent on coding every month (OpenRouter does 100T tokens ALONE per month).
➡️ Play the "intelligent" game or "fast" game or both, not the "India" game. People love fast but slightly dumb models, or they love models that are very intelligent. Much of the focus and marketing should be on serving inference on these two classes of models.
➡️ The only price I see right now is INR, why? Put up pricing in USD/EUR/GBP and make it a global platform, at least the platform subscription to begin with, if not the API.
➡️ Ship a desktop app, ship a CLI for programmers, and bring your model to harnesses like OpenCode, your model should be available via OpenRouter, Vercel AI Gateway, Cloudflare AI Gateway, etc.
➡️ Match SOTA Open Source, surpass it next. The DeepSeek/Kimi/MiniMax playbook is open (except data). It'll take time and $ to build a ~2T model, but it is possible. Celebrating 24B and 100B models is great but eyes on the real prize.
➡️ Custom benches, better evals, and system cards that operate like how Anthropic and OpenAI operate. It's fine if metrics are not good, 4 months later when a new model drops, you get to compare your bad bench scores to the great ones.
This needs to be a global play. Please do not get stuck just serving and boasting the India angle. We need to win on the merit and ecosystem, not on sovereign AI gameplay. I say this as a fan. All the best.
@ETNOWlive@MotilalOswalLtd Please share what is the domestic demand slowdown on a % basis of tvs, hero, hmsi and then compare the % drop to Bajaj. On a % basis, Bajaj is the least and this is as per the latest data on vahan. Now out of all these who has the largest exports?
@wegro_app Motilal Oswal has the worst analysis. BA is the ONLY player of its profitability growing at the rate it does which is 25-30% yoy for q end June. The slowdown in domestic has hit tvs, hmsi, hero far worse. Ba is the ONLY one amongst these with higher exports than domestic. Idiots
@kkktwits Not doing well on the market. On the quarter it will close up by 30%+. Forward pe: extremely attractive. It’s a minimum 15k stock trading at. 50% discount. Growth rate of Bajaj and TVS this quarter is the same - now compare the pe multiples and decide which one needs correcting.
@ku1deep The govt is needed more as a power user than an investor. Once it does that, it will pick up the stake itself re l&t x e2e. That, in itself, would blitzkrieg sarvam or any other player.
@pranav_m28 Exactly. He’s another project like yoro. There’s potential for sure but why are people willing to pay Mateus level money for a guy like him?
Everyone saying the same dumb shit not realising people sell in secondaries all the time. This is not just dilution. Comparisons to Elon are faulty, he never sells.
@Aurielus80HD@AcademyScoop I did and my opinion holds. He’s not good enough to justify his price tag. Hes certainly talented but we’re done paying end product price tags for emerging talents. He costs the same as Mateus
@padelreviewes Salty prick ! I wish Augsburger / LeBron could actually do better and beat them but they can’t seem to get over their own the main hurdles. Too nervous when they play big points. They need to be calmer and talk less and simply expect to win.
@MohapatraHemant True but only one number really matters ie what is it worth to India as a country. Forget $50-60b even if you double, triple it, it still comes back positive.