Normal Guy

This Week of Idealist is Now live :) This time, I discussed a business undergoing restructuring. What the business was in 2021 has completely changed by 2026, and it now sits in a major turnaround phase with strong industrial exposure normalguy.co.in/post/idealist-…

@growthcc_ I am not hating :) Indian IT is changing fast, those who not update themselves will get the heat :)

Sir, if anyone has guts, please build AI, but stop hating on Consulting services companies. They are doing their job exceptionally well, and also AI implementation part of business they are growing 10-30% per quarter. China mostly distilled Anthropic and openAI, which is blatant violation of user code. There are million of opensource models now, AI is already getting commoditised. AI is US's own innovation. Just like operating system, internet, AC motor, Medical patents ( that Indian pharma distills and rebrands) and thousands of innovation that are unique to USA. We will make it eventually but hating on our consulting companies deprives them of much needed capital and drive to innovate.

OpenAI gave Indian users ChatGPT Go free for a full year starting November 2025, Plan normally costs Rs 399 per month. India already has over 100 million weekly ChatGPT users. Training frontier model in 2026 costs $200 million minimum and heading toward a billion. Any Indian startup building rival product starts day one competing against something that is free, used by hundred million people, and backed by tens of billions in funding. No rational investor takes that bet without guaranteed demand on other side China did not build frontier AI because sanctions somehow made it stronger That reading is wrong. Chinese AI labs grew because OpenAI and Anthropic were blocked in China. 1.4 billion people needed AI tools and only domestic companies could serve them That unmet demand funded DeepSeek, Qwen, and dozens more. India faces opposite conditions. Every global model from GPT to Claude to DeepSeek is freely available here. Indian startups have to compete against all of them simultaneously with no captive market anywhere. India spent 0.64% of GDP on R&D in 2025 That number has not moved in 30 years. Private sector contributes only 41% of even that tiny amount. For comparison, China private sector funds 77% of its R&D and South Korea 79%. Top 10 Indian non financial firms made $43 billion in profits in 2021 and spent under $1 billion on R&D combined. Building frontier AI requires pouring billions into something with no guaranteed return. Indian corporate culture has never shown that appetite, protected market or not. But they will do Buyback for sure :)

Marquee investors like Norway pension fund, Invesco, ICICI Pru and BlackRock backed the issue. Just think about it :) When these big fund invested in this Auto player, Yu have to Connect the dot & Find Why? :) From Market Outlook :)

With Handpicked Filings Dashboard

This Week of Market Outlook is now live :) With fresh updates in battery chemicals, distribution expansion across industrial, consumer-care and agrochemical businesses, and companies moving towards higher-value segments and turnaround breakouts. normalguy.co.in/post/market-ou…

This battery shift is premiumization showing up in the most boring product category you can imagine. Indian consumers are trading up across everything. Premium FMCG brands are growing nearly 2x faster than mass market and now make up 27% of all FMCG sales. By 2030, 80 million Indian households will be affluent. $2 trillion of India projected consumption growth is expected to come from consumers upgrading to better versions of things they already buy. Alkaline replacing carbon zinc is just one small example of that pattern. 85% of India battery market is still carbon zinc, Cheapest option. Works fine in a wall clock or a basic flashlight. But Indian homes now have wireless mice, gaming controllers, smart TV remotes, blood pressure monitors, cordless doorbells. Carbon zinc cells die in these devices within days. Alkaline cells cost 3x more but last 3 to 10x longer. The battery shift is not happening because consumers suddenly care about battery chemistry. It is happening because the devices they own now demand it. Eveready alkaline battery sales grew 82% year on year in Q4 FY26. Alkaline is now 19% of their total battery volume, up from single digits a few years ago. Company just opened India only alkaline battery manufacturing plant in Jammu with Rs 200 crore investment and 456 million units of annual capacity. They are targeting 53% share of India alkaline market within four years. Source - @drprashantmish6

Trade Deficit Soon

@Naresha54993928 Really?

@Normal_2610 I appeared for #ISRO Group A scientist interview last year in august and even selected for the post. Do you know what they offered in breakfast before the interview? 🍌 Unriped Bananas 🍌 Glad i didn't join( thanks to the bananas,haha joking)

India opened space to private companies in 2020, Government promoted startups, gave them access to ISRO infrastructure, transferred SSLV rocket technology to HAL Private space companies went from 11 in 2019 to over 400 by 2024, Space startup funding crossed 147 million dollars in FY2025-26 alone. Now same government issues memo saying scientists cannot resign from Gaganyaan till mission is done. You created competing market, funded it, celebrated it Then try to trap people from joining it, Fix is not restriction, Fix is better pay. Meanwhile around 120 scientists working on key missions walked out in past year. ISRO has 1,050 vacant scientific positions it cannot fill fast enough. When people who held deep mission specific knowledge leave and replacements do not come in at same rate, missions start failing ISRO pays entry level scientist about 72,000 rupees per month in hand. Senior scientist with 20 years maxes around 2 lakh with all allowances. Aditya Rallapalli ran simulation team that generated 25 terabytes of data validating Chandrayaan 3 moon landing, He quit. Skyroot founders who left ISRO in 2018 hold combined 2,440 crore in billion dollar company. Startup employees with stock options can earn more in one IPO or buyout than ISRO pays over full career. Cadre review added 460 posts, Changes nothing when private pay runs multiples higher.

@Sambbhav765 They are making themselves read the recent documents

@Normal_2610 Polycab are not making any solar panels. Have they disclosed which brand of solar panels they are selling?

Polycab FMEG division went from 2.1% EBIT margin to 8% in one year. Solar inverters and Topcon modules are now the single largest category in the FMEG basket, bigger than fans, bigger than lights, bigger than switches. Solar products grew over 2x YoY This margin expansion is not operational efficiency. It is product mix. Solar carries higher margins than traditional FMEG products like fans and LED panels. A wires and cables company found its highest-margin consumer product by selling rooftop solar gear to the same electricians who buy its cables. India installed 2.7 GW of rooftop solar in Q1 2026, up 125% from a year ago. Residential systems made up 82% of new installs. PM Surya Ghar has already connected 2.4M homes. Polycab FMEG growth of 71% YoY is what happens when a company with India deepest electrical distribution network plugs into that wave. The same distributor who sells wires and MCBs to an electrician now sells him solar inverters. Pure solar companies cannot match that kind of reach. Polycab did not enter solar, Solar entered Polycab existing pipeline. When a consumer electrical company fastest growing product is not fans or LED lights but solar inverters, that tells you where Indian household spending is going. India residential solar inverter market is projected to grow from $1.8B in 2026 to $8.4B by 2035. Cumulative rooftop capacity reached 30 GW by May 2026. Polycab hit 8% EBIT margins in FMEG for the first time, reaching its Project Spring target of 8 to 10% four years early. The energy transition is showing up in quarterly earnings now, not just policy papers. This is How yu can build yur own thesis :) i did write up in the past and going really well :)

In 2018 Geely (Li Shufu) bought 9.7% stake in Mercedes, becoming its largest single shareholder In 2024 Geely and Renault formed Horse Powertrain, a joint venture with 17 plants and 19,000 employees making engines. In 2026 the new CLA hybrid ships with a 1.5L turbo engine designed by Mercedes but built entirely in China by Horse. A decade ago Chinese automakers sought German engineering to catch up, Now Germany most famous car brand sources its engine from a Chinese-led factory, flow reversed Mercedes sourcing Chinese-built engine for the CLA is one data point in a larger shift. European automakers warned Brussels that chip inventories from Chinese suppliers could run out in weeks if restrictions hit. BYD is building a factory in Hungary. Chinese auto component exports to Europe rose 62% in early 2026. EU now requires 70% local content for vehicle subsidies because supply chains already depend on Chinese inputs. Europe is not losing its car industry in some future scenario. It is losing it in this quarter purchase orders Horse Powertrain runs 17 factories and 5 R&D centers, supplying engines to Renault, Geely, Volvo, Nissan, Mitsubishi, and Proton. Planned capacity is 5M units a year. Mercedes chose Horse for the CLA because it builds at cost and quality levels German plants cannot match for a car starting around 50,000 euros. CLA is the entry point of the Mercedes lineup. Entry-level is where cost pressure arrives first. Same pattern played out in electronics two decades ago. Design stayed in Europe. Manufacturing moved to Asia. Engines were the last holdout. Playbook always same :)

Japan is upset India chose ETCS over its proprietary DS-ATC signaling for the bullet train But ETCS is the global rail signaling standard, used by China, Saudi Arabia, South Korea, Australia, Taiwan, and across Europe. DS-ATC works on Shinkansen lines in Japan and nowhere else. India choosing ETCS means any future high speed corridor can use multiple vendors and interoperate with any rolling stock. Japan is not complaining about safety, Japan is complaining about losing a permanent customer India is doing this across sectors now. ETCS over DS-ATC for rail. DAP 2026 demanding indigenous design ownership for defense. UPI instead of adopting Visa and Mastercard protocols for payments. NCRTC already runs ETCS Level 2 on the Delhi-Meerut RRTS, first deployment of its kind in India. Countries that adopt open global standards get supplier competition and lower prices. Countries that accept proprietary systems get a single vendor who controls every upgrade for 30 years. India is learning to pick the standard, not the vendor The funniest part is that Japan uses ETCS on its own network for interoperability. Shinkansen DS-ATC is a legacy format no other country adopted in 60 years. A Japanese railway engineer published a complaint saying India broke its promise by switching standards. But what he calls a broken promise is India insisting on internationally certified signaling with third-party verification, which DS-ATC did not meet. India did not break a promise. India asked for a global standard and Japan did not have one ready

Let’s wait and see what else happens :)

China compressed US factory margins over 30 years that hurt but factories have a cost floor because every unit costs money to make and move Software does not. Kimi K3 beat Fable 5 on Arena coding leaderboard with 76% pairwise win rate and Moonshot releases full model weights for free on July 27 Anybody downloads it and runs it on their own servers. Chinese open-weight models went from under 1% of global API traffic to over 30% in one year US hyperscalers plan to spend $725B on AI infrastructure in 2026, Token costs dropped from $60 to roughly $0.50 per million for GPT-4 level work in under two years. Chinese open-source models are 60 to 90% cheaper than Anthropic and OpenAI on equivalent tasks. In manufacturing you needed factories, ports, shipping routes. Building those took decades. In AI the weapon is a downloadable file. Moonshot went from $4.3B valuation to $30B in six months on the back of models US developers are already using Washington spent 3 years trying to slow China with chip export controls. Chinese AI labs responded by building sparse architectures that cut compute needs by 90%. Kimi K3 runs 2.8 trillion parameters on infrastructure the US tried to deny them, Cursor uses Kimi to build its coding agent. DoorDash routes lower-level work to K2.6. Export controls were designed to starve Chinese labs of computing power, Instead they pushed those labs to optimize so hard that American companies now voluntarily send them revenue Even Indian are using it even me personally :)

Cybersecurity

Broking = acquisition. Mutual funds/wealth/AMC = retention. Credit (MTF, loans against securities) and insurance = cross-sell. AI cuts the cost to serve. But look closely: every single one of these three is growing its lending book fastest. Groww's MTF +264%. Angel One's funding book +31% in a quarter. Anand Rathi's +55%.

There are too many moving parts in the US market right now. A wave of IPOs is coming, and nearly $1 trillion of liquidity could get absorbed. We have already seen the kind of capital appetite in SpaceX IPO China is the new wildcard. Its rapid progress in AI could pressure margins across the entire sector. Korean markets have already gone sideways, debt remains another major risk, and models like Kimi K3 show how quickly the competitive gap is narrowing. For now, the US market may remain sideways in some short & now more selective. The better opportunity is to connect the dots and focus on China AI supply chain and Europe defence cycle.

@signulll OpenAI feedback same thing when yu chat :) in past it gives 2 response :) other copy it

whoever built this long thread navigation ux at openai absolutely cooked. the hover states give you a quick glance of the message context incl artifacts. & the non hover states give you a sense of your context vertically within the thread too with subtle opacity changes. lovely.