India should just use the existing opensource model and upgrade it's for it's own use by additional traning instead of wasting millions of dollars by doing from scratch... That's best as of current scenario they should focus on how to improve semiconductor and cloud computing infrastructure for india ...for benefit the mass jobs opportunities 🙂
2 months ago | 0
Thillai The Techie
Recently DeepSeek-R1 has gained significant attention, which was trained with:
Just 2,000 Nvidia H800 GPUs (vs. 25,000+ used by OpenAI/Google).
A cost of $5.6M (vs. the $100M+ spent by competitors).
An open-source model, making AI innovation accessible without huge licensing fees.
🔴 Trump’s AI Warning
Even Donald Trump has called China’s AI advancements a "wake-up call" for the U.S. But honestly, it should be a wake-up call for India, too.
India has repeatedly proven that cost efficiency and innovation are possible.
ISRO’s Chandrayaan mission cost just $74M, cheaper than Hollywood’s Oppenheimer ($100M) or Mission Impossible 7 ($291M)!
Why should we believe AI models are out of reach if we can send rockets to the Moon for less than a movie budget?
💡 The Real Question:
Should India focus only on AI applications, or should we also invest in training our own AI models to stay globally competitive?
👉 What’s your take? Should India train its own AI models or just leverage existing ones?
2 months ago | [YT] | 9