Absolutely fascinating that Indian manufacturers are securing long-term contracts with US tech giants while still unable to lock in their own raw material supply, leaving them completely exposed to Chinese export controls.
It's an interesting perspective to see India's AI boom is more than building better models. But also about building it's infrastructure to support them. Great article!
Something feels off in this sentence - "Interestingly, Indian cable makers are also turning customers away simply because they can’t handle this overflow of demand without adequate demand."
The maize section raises a useful measurement question. The November 2025 mandi average and September 2026 poultry procurement quote describe different markets at different times, so they cannot alone show who captured the ethanol premium. Is there district-level evidence on distillery procurement prices and volumes alongside farmgate maize prices during the same harvest window? That would help test whether the policy benefit reached growers.
Hey thanks for reading! We did find mandi level price data and some company reported maize procurement prices, but not matching district level purchase prices and volumes for the same harvest period. So we can't fully say how much of the ethanol incentive actually reached farmers.
Yes, they (Hindustan Zinc) just announced a 10 mtpa recycling plant to recover zinc, including germanium! And I agree on the HFCL point. Well thought of :)
Considering that Total Internal Reflection dictates optical fibre viability, manufacturers must look into alternatives like Chalcogenide Glass over Germanium in order to reduce geopolitical shocks..
This is thoughtful! Chalcogenide glass is interesting as an alternative optical material, especially in specialised applications. For conventional telecom and data-centre fibre though, it isn’t quite a like-for-like substitute for germanium-doped silica (yet!). Currently, they are used in areas adjacent, like mid-infrared optics.
I think manufacturers should reduce geopolitical exposure through alternative dopants and diversifying supply. your broader point, that the industry (in general) has an incentive to look beyond germanium dependence, is a good poiht. Thanks for sharing.
AI RACE HEATS UP: DEEPSEEK AND MOONSHOT REPORTEDLY USED OPENAI MODEL OUTPUTS
Chinese AI developers DeepSeek and Moonshot AI reportedly used outputs from OpenAI's GPT-6 Astra to help train their latest models, according to The Information, citing people familiar with the work. The practice, called knowledge distillation, can help AI developers build capable models at lower cost. OpenAI, DeepSeek and Moonshot did not immediately respond to Reuters' requests for comment.
Why it matters: The dispute raises questions about AI competition, intellectual property and the cost of developing advanced models.
Veritasium channel recently released a nice explanatory video on creation process of these fiber optic cable.
People who got interested from the first article can check this out.
https://youtu.be/NiCgcUTL9_Q?si=MbZEngrFApl29gix
Thanks for sharing, this is interesting!
Absolutely fascinating that Indian manufacturers are securing long-term contracts with US tech giants while still unable to lock in their own raw material supply, leaving them completely exposed to Chinese export controls.
I agree! Thanks for reading!
It's an interesting perspective to see India's AI boom is more than building better models. But also about building it's infrastructure to support them. Great article!
Thanks for reading, and it is definitely interesting to see how it all plays out.
Something feels off in this sentence - "Interestingly, Indian cable makers are also turning customers away simply because they can’t handle this overflow of demand without adequate demand."
Hi Gunjan, thanks for flagging the typo, we've rectified it now
The maize section raises a useful measurement question. The November 2025 mandi average and September 2026 poultry procurement quote describe different markets at different times, so they cannot alone show who captured the ethanol premium. Is there district-level evidence on distillery procurement prices and volumes alongside farmgate maize prices during the same harvest window? That would help test whether the policy benefit reached growers.
Hey thanks for reading! We did find mandi level price data and some company reported maize procurement prices, but not matching district level purchase prices and volumes for the same harvest period. So we can't fully say how much of the ethanol incentive actually reached farmers.
makes you think about how much goes on behind the scenes of the AI boom. Really enjoyed reading this!
Thanks for reading!
hindustan zinc, india's biggest zinc producer, is still doing r&d on whether the germanium in its ore can even be extracted
hfcl doesnt need germanium today only because its preform comes from japan, and its new plant making core and clad in-house changes that
Yes, they (Hindustan Zinc) just announced a 10 mtpa recycling plant to recover zinc, including germanium! And I agree on the HFCL point. Well thought of :)
in one wire there is so much value good explanation … thanks zerodha
Thanks for reading!
Considering that Total Internal Reflection dictates optical fibre viability, manufacturers must look into alternatives like Chalcogenide Glass over Germanium in order to reduce geopolitical shocks..
This is thoughtful! Chalcogenide glass is interesting as an alternative optical material, especially in specialised applications. For conventional telecom and data-centre fibre though, it isn’t quite a like-for-like substitute for germanium-doped silica (yet!). Currently, they are used in areas adjacent, like mid-infrared optics.
I think manufacturers should reduce geopolitical exposure through alternative dopants and diversifying supply. your broader point, that the industry (in general) has an incentive to look beyond germanium dependence, is a good poiht. Thanks for sharing.
AI RACE HEATS UP: DEEPSEEK AND MOONSHOT REPORTEDLY USED OPENAI MODEL OUTPUTS
Chinese AI developers DeepSeek and Moonshot AI reportedly used outputs from OpenAI's GPT-6 Astra to help train their latest models, according to The Information, citing people familiar with the work. The practice, called knowledge distillation, can help AI developers build capable models at lower cost. OpenAI, DeepSeek and Moonshot did not immediately respond to Reuters' requests for comment.
Why it matters: The dispute raises questions about AI competition, intellectual property and the cost of developing advanced models.