Drones are reshaping traditional battlefield tactics big time. No military can ignore drones and anti drones going forward. Juxtapose 1 world war and 2 nd world war . While 1 St world war was predominantly fought using traditional weapons, the allied and axis forces had dedicated airforce in the 2 nd ,which decided the fate of the war . Similarly drones and anti drones could be a game changer in warfare .
2 Likes
Fascinating story of Radico Khaitan
100% agree. Once we are done with AI/Power, robots will be the talk of the town. Have you identified which companies sit in the value chain? Actuator and bearings are one way to play it based on my limited search.
1 Like
My research too is limited. Sona Comstar seems to be one company which has done some work / talks about meaningful contribution to humanoids . Below is from their concall transcript and ppt regarding humanoid robots :
Most of the Chinese robotics companies and US companies like Tesla (Optimus) have evolved from an EV pedigree. So Indian auto ancillary companies within the EV domain could naturally extend to Humanoid Robotics albeit being a tier 2 / tier 3 supplier rather than an OEM.
Another company which talks about humanoid robots in their concall is OBSC Perfection. It is like a pick and shovel play .
Having said this , I believe the proper OEMs for humanoid robots out of India could come from the start up ecosystem . But there could be several auto ancillary and precision engineering companies who could supply the nuts and bolts and can be critical to Actuators, QDD, Dexterous hands and Sensory systems along with fabrication, wire harnesses etc etc. I believe similar to the AI ancillary play that made money for investors for the past 6-12 months in India , once humanoids become mainstream there could be some pick and shovel plays that could emerge .
P.S: I have drawn heavily from twitter account of Sekhar who has done excellent work on humanoid robots.
3 Likes
An excellent company in robotics was ABB which had earlier intended to demerge its robotics division in 2025 and I was keen to have it in my core portfolio either via ABB demerger or later once it lists…unfortunately they decided to sell it to softbank lead group and we lost opportunity to have some stake in one of the leading tech company in robotics and that too in India via its indian listed entity…
I am still on lookout since then to invest in a robotics company with equally good technical moat…this parts supplier angle methodology to invest in an ecosystem is good but difficult to identify long term moats here and understand the cycle…so still on lookout…but i agree this is a very very long term theme…
3 Likes
With backlashes across US on the unprecedented Data Center build out and inability of the grid to supply , the hypersclers , neoclouds and other DC operators would look else where. The theme hopefully has more steam left .
It’s been a while since I posted my portfolio update.
The last 5 months have been excellent in terms of returns since April. The portfolio is currently at all time highs and there about. I have 17 stocks which are considered as core holdings and another 3 stocks which I have started allocating. Hold around 14% cash.
The aim is to become more concentrated , cut to 10 to start with and progressively move to 6-7 from the current 17 . The philosophy shared by @Rokrdude [(Investing Decision Making; Template1 - #14 by Rokrdude)
Have been trying for a while but not been successful so far. I strongly believe in bull markets if you are concentrated enough then you can build substantial alpha . The challenge is to build bullet proof conviction.
Some of the highest % gains (Neuland, Navin) are from businesses held over 4-5 years. Rest of the ones are fairly new : TD, E2E, Car trade etc are 1-2 years old and rest of the ones were picked up post January of 2026. The stocks from #14 onwards are 1-2 months old or below.
The interesting aspect is that I have bought and sold almost 20 odd stocks in the last 9 months some of them at loss and some with gains 
KPIT ,SWPE, Taril, Kaynes, Connplex, Genesys, Waaree , Ceinsys, Bluejet, Websol, Goldbees, Cohance, Dishman, Fisher Medical, Stallion Fluoro, HZL, Yatra, United heat , Sterlite , Inox, Dee Dev etc to name a few. Some have been cut to reduce the AI/DC exposure as the portfolio was heavily tilted to the DC ancillary theme, some were sold due to a notional stop loss, some for making way to a more interesting theme.
I have taken ideas shared in this forum by the likes of @phreakv6 (but have tried to do my own research) and am truly grateful. I have also found the quarterly updates by Xponent tribe to be of great value.
I am currently researching/studying Orchid Pharma, Standard Engineering , Gaja, Apollo Micro and few others.
Happy to discuss about any of the portfolio stocks.
P.S: Pls do your own research. No buy/sell recommendation. I have made abundant mistakes and paid more than my share of tuition fees 
19 Likes
Congratulations Mithun on ATH, Neuland & Navin are stars here. Just curious, Navin had consolidated in a range from 2021 to 2025, you already had a 5X in two years from 2019-21. How did you hold it patiently as it was a raging bull market in 2021-2024?
It pains when you see everything running and your stock is sleeping, salutes to you for that kind of patience. There were numerous switch available at that time as different sectors were evolving and catching up the rally.
3 Likes
Hi @Cshar ji,
Thanks for your kind words. I generally abhor selling , especially if the stock/ business has delivered good returns for me. For eg. I got KPIT foc (or below 100 not sure ) after it got demergered with Birla soft . It went all the way up to around 1800 and came down . I held it all the way upto around 700 and sold finally. Not sure which is a good approach. On the other hand ,in Neuland , I had reduced my exposure at around 6000 around 18 months back to see it go 4x from there. I guess the journey of an investor is laden with regrets of omission and commission 
16 Likes
Can you please help me understand or start a thread for GaudiumIVF
Though I am not a tech guy (actually bordering tech illiterate 
),taking a jab at the Nvidia earnings call for Q2 CY 26. I found out the Nvidia call gives good insights on where the whole AI /DC theme is heading.
Investor Call Notes :
-
Though NVIDIA has other businesses bulk of the revenue comes under their Data center vertical ($89 bn out of $96 bn . ie circa 9 tn INR for 1 qtr !!!)
-
They categorize their DC revenue under 2 verticals : Hyperscalers and ACIE (AI clouds, Industrial and Enterprise including Neo clouds , Sovereign AI etc).
-
Nvidia has given a guidance of 70% growth in sales for next year. In fact this could be much higher if the supply chain constraints were not there.
-
They believe the Capex of the top 5 hyper scalers could reach $800 bn in 2026 and $1.3 tn in 2027.
-
Announced AWS partnership where they are deploying additional 2 mn GPUs starting now going unto 2 years along with Vera CPUs. some integrated with Rubin and others standalone. AWS will serve NVIDIA Nemotron family of open models on Amazon Bedrock and Sagemaker.
-
NVIDIA architecture is running on every model both closed ones like Open AI, Anthropic, Grok, Gemini as well as open models like Mistral, Qwen, KIMI, GLM, Deep Seek, Mini Max and Nemotron (Nvidia own model).
-
Nvidia has unique capability of full stack AI platform that is expanding DC TAM. They have launched Groq 3 LPX (LPU/Agentic AI enabler with lower latency), NVL72 (Worlds first rack scale architecture where 72 GPUs are wired together in a single unified n/w domain using high bandwidth NV link copper cables . Due to their ultra low latency and architecture the DC treats the entire rack as one giant GPU).
-
In each of their generation of GPUs,the revenue opportunity/GW has grown :
Hopper : $18 bn/GW — Blackwell : $25bn/GW ----- Vera Rubin : $40 bn/GW
Vera Rubin for eg: delivers 30x higher throughput /MW and 35x lower token cost relative to Grace Blackwell ultra. -
NVIDIA considers itself as a neutral partner to neoclouds and sovereign clouds. They helped Coreweave, Nebius and Nscale build entire infra business.
-
Nvidia lists the following business world over as top Neoclouds/soverign AI : Firebird in Armenia, Cassava Technologies in Africa, GMI cloud in Taiwan, Yotta and Neysa in India (E2E and ESDS not mentioned), Neotra in Japan.
-
Hudson River and Jane street uses NVIDIA powered AI factories to accelerate quant trading . Samsung electronics uses NVIDIA cuLITHO to achieve 20x performance in computational lithography, Bristol Myers is investing in Vera Rubin while Roche and Lilly are extensively using chips to compress R&D timelines.
-
Jenson Huang believes Open models have finally reached frontier and interesting cyber security companies are being built on top of frontier models.
-
Nvidia sees $20 bn demand from server CPUs and CPU revenue to more than double next year.
"Rising adoption of Agentic AI is driving acceleration in demand for DC CPUs " -
Nearly 20 companies including Cursor, Sigma, Together AI exceed $1 bn annualized revenue run rate.
-
Nvidia sees memory scarcity to persist . Despite having strong relationships with top 3 memory suppliers there would be memory supply crunch.
-
Aim is to put as much as possible compute per GW . Vera is $40bn /GW presently v/s Blackwell at $25 bn/GW. Jensen thinks even $1 tn/GW could be possible in the future.
-
Regarding allegations of circular financing :
Don’t see support extended to frontier AI labs as circular financing. Recently announced partnership with leading infra capital providers: Apollo, Blackrock, Blackstone, Brookfield, KKR & Goldman. This platform to raise $500bn 3rd party capital . -
Regarding Recursive Self Improvement (RSI) and AGI : Jensen said that it’s already taking place. For eg: Nvidia today has 40k employees, tomorrow it will have 4 mn agents working in the back ground and it will keep improving.
In summary , Nvidia is leaving no stone unturned be it GPU, CPU, software , networking, investments in frontier AI labs, financing Neo clouds and establishing alliances world wide. It is getting bigger and bigger and is rewriting the old paradigms across entire AI landscape and is deepening its tentacles in everything that has to do with AI .
8 Likes