# Sarhaan Gulati — full content
Source: https://sarhaangulati.com/
Hi, I'm Sarhaan! I'm the co-founder of Pantropy, where we're working on engineering more nutritious seeds.
I grew up in Delhi, studied at UBC, and spend most of my time in Toronto these days.
**I'm also moving to San Francisco in June**, and would love to meet if you're around; we can grab lunch, go go-karting, play soccer/pickleball/table tennis, or if schedules don't allow for an adventure, go for a walk!
Reach out anytime: sarhaangulati737@gmail.com or https://x.com/sarhaan77!
I can write fluffy words to explain the kind of things I like, but I'd rather you skim these:
- [Things I Like](/writing/things-i-like)
- [Cool things I own](/writing/cool-things-i-own)
- [Design Vibes and Aesthetics I really like](/design)
- [I collect awesome images and videos from all over the internet](/awesome)
- [Tunnels, IRL Pokemons, new types of guns, and lots of other fun ideas](/writing/fun-ideas)
---
Source: https://sarhaangulati.com/writing/the-mischievous-past
Date: 24th April, 2026
## The Mischievous Past
I once heard Palmer Luckey say on a podcast, "The best ideas were written down in government publications between 1950 and 1970 because that was the last time scientists and engineers were allowed to think openly without political correctness or corporate legal paranoia."
So I got curious and started digging. And boy, was he right.
### [Project Horizon](https://documents.theblackvault.com/documents/army/horizon.pdf#:~:text=after%20the%20beginning%20of%20active,life%20essentials%20such%20as%20insulated)
Plan for an outpost of ~12 astronauts with buried cylindrical habitats and on-site nuclear reactors for power on the Moon
### [Project Orion](https://ntrs.nasa.gov/api/citations/20000096503/downloads/20000096503.pdf#:~:text=affordable,2M%C2%B7Stage)
Nuclear-pulse spaceship that would ride successive atomic explosions off a giant steel pusher plate. The motto was apparently "Mars by 1965, Saturn by 1970." Early designs were 80m tall, carried 150 people, and could haul thousands of tonnes.
### [Project 1794](https://www.secretsdeclassified.af.mil/News/Article-Display/Article/459834/project-1794-documents-saucer-type-aircraft/#:~:text=technical%20data%C2%A0and%20drawings%20related%20to,58)
The USAF funded Avro Canada to build a supersonic VTOL flying saucer interceptor to shoot down Soviet bombers. It was designed to use ten downward-facing jet engines for lift and reach Mach 3+.
### [Project Pluto](https://www.fourmilab.ch/documents/pluto/#:~:text=Pluto%20was%20a%20nuclear%20powered,to%20rain%20nuclear%20destruction%20upon)
Nuclear-powered ramjet cruise missile, also known as SLAM. It was designed to fly at Mach 3-4 at low altitude, loiter for months, and carry dozens of nuclear warheads.
### [Project Excalibur](https://en.wikipedia.org/wiki/Project_Excalibur)
Nuclear-armed interceptor missiles launched into space to explode near incoming ICBMs.
### [Cyborgs in Space](https://web.mit.edu/digitalapollo/Documents/Chapter1/cyborgs.pdf)
NASA scientists Manfred Clynes and Nathan Kline proposed the term "cyborg" while thinking about astronauts. They suggested engineering human physiology to survive space rather than relying only on external life-support systems.
### [General Electric Hardiman](https://cyberneticzoo.com/wp-content/uploads/2010/04/GE-Hardiman-paper.pdf)
Powered Army exoskeleton from GE that used hydraulics and electronics so the wearer could lift hundreds of kilos.
### [Project ROVER / NERVA](https://ntrs.nasa.gov/citations/20140008771#:~:text=March%201966%2C%20the%20NRX,and%20found%20various%20structure%20anomalies)
NASA and the Atomic Energy Commission built hydrogen-fueled nuclear rocket engines. By 1969, the NRX-A6 reactor ran at 1.1 gigawatts for over an hour.
### [Project Chariot](https://nuclearprinceton.princeton.edu/project-chariot)
A plan to blast a seaport into Alaska using nuclear bombs.
### [Project West Ford](https://www.thecrimson.com/article/1963/5/24/project-west-ford-pfour-hundred-million/#:~:text=Four%20hundred%20million%20copper%20wires,proof%20communication%20system)
MIT Lincoln Lab launched ~480 million tiny copper needles into orbit to create an artificial radio-reflecting belt around Earth. Basically a man-made ionosphere for jam-proof global communications.
### [Project A119: Nuke the Moon](https://en.wikipedia.org/wiki/Project_A119)
A 1958-1959 plan to detonate a nuclear bomb on the Moon. The stated goals included planetary astronomy, astrogeology, and demonstrating U.S. capability. If the bomb exploded on the surface instead of inside a crater, the flash might have been visible from Earth.
### [NB-36H / X-6 Nuclear Bomber](https://www.hill.af.mil/News/Article-Display/Article/2664365/history-in-two-manned-nuclear-aircraft-program/)
A B-36 testbed with a live 1 MW reactor in the bomb bay. It flew 47 times to prove crews could be shielded from an onboard reactor.
### [Project Iceworm](https://bldgblog.com/2011/01/project-iceworm/)
The Army wanted 3,000 miles of rail tunnels under Greenland’s moving ice sheet to hide 600 Minuteman missiles.
### [Project Carryall](https://atomic-skies.blogspot.com/2014/02/project-carryall.html)
Plan to use 22 nuclear bombs to cut through California’s Bristol Mountains for I-40 and the Santa Fe railway.
### [Panatomic Canal](https://en.wikipedia.org/wiki/Project_Plowshare)
1964 study for a sea-level alternative to the Panama Canal, dug through the Darién with around 200 multi-megaton nuclear devices.
### [Project Mohole](https://en.wikipedia.org/wiki/Project_Mohole)
An oceanic rival to the Space Race: drill from a ship down to Earth’s mantle.
### [Buckminster Fuller’s Dome Over Manhattan](https://www.fastcompany.com/3058386/how-buckminster-fuller-made-a-dome-over-manhattan-sound-sensible)
A 3 km geodesic dome over Midtown Manhattan to cut NYC energy use by 80%.
### [Army High-Dose Food Irradiation](https://www.gao.gov/products/psad-78-146)
Sterilize steaks with gamma rays so C-rations could stay edible for years.
### [Food From Oil](http://www.ascension-publishing.com/BIZ/CIA-microalgae-Soviet.pdf)
Grow yeast on petroleum n-paraffins, then press it into animal or human feed.
### [Project STORMFURY](https://en.wikipedia.org/wiki/Project_Stormfury)
NOAA and the USAF tried seeding hurricanes with silver iodide to weaken them. Early flights looked promising, but later data showed the storms were reorganizing naturally.
### [NAWAPA](https://ashcroftcachecreekjournal.com/2026/01/08/the-editors-desk-remember-the-nawapa-project/)
A 1964 continental plumbing plan from the Army Corps and Parsons: dam Alaska, send 75 million acre-feet of water south, and excavate canals with "peaceful" nukes.
### [Qattara Depression Blast Canal](https://en.wikipedia.org/wiki/Qattara_Depression_Project)
German consultant Friedrich Bassler proposed using 213 nuclear devices to cut a trench from the Mediterranean to Egypt’s below-sea-level desert, then run a perpetual hydro plant.
### [Lockheed CL-1201](https://en.wikipedia.org/wiki/Lockheed_CL-1201)
Late-1960s nuclear-powered flying aircraft carrier. 6,000 tons, 1,120 ft wingspan, 41-day endurance, 22 parasite fighters. The reactor shielding alone weighed more than a C-5.
### [Project PACER](https://en.wikipedia.org/wiki/Project_PACER)
Los Alamos wanted to drop two 50-kiloton hydrogen bombs per day into a salt cavern to boil water for a 2 GW power plant.
### [Ford Nucleon](https://www.ans.org/news/article-3058/the-1958-ford-nucleon-an-idea-thats-still-ahead-of-its-time/)
Detroit explored hot-swappable mini-reactor cars.
### [Project HARP](https://en.wikipedia.org/wiki/Project_HARP)
A 16-inch space gun that lobbed payloads to 180 km. The Barbados gun is apparently still rusting on the cliff.
### [The Very High Speed Transit System](https://www.rand.org/content/dam/rand/pubs/papers/2008/P4874.pdf)
Maglev capsules in hard-vacuum tubes, Los Angeles to New York in about an hour.
### Space-Based Solar Power
1968-1980 studies for 5 GW microwave-beaming solar platforms in orbit, each around 50,000 tonnes. (Reflect Orbital before Reflect Orbital)
### [Project Cyclops](https://www.daviddarling.info/encyclopedia/C/Cyclops.html)
1971 SETI plan for 1,000-2,000 steerable 100m radio dishes scanning 1,000 light years for aliens.
---
Source: https://sarhaangulati.com/writing/fun-ideas
Date: 7th April, 2026
## Lots of Fun Ideas
### Tunnels are Underrated
A lot of people are working on securing the oceans, air, and land. But who is securing the subterranean?
Good, cheap, fast tunnel-boring robots could help secure supply lines in war zones, where drones make the surface dangerous. They could also help with traffic, desalinated water, and package delivery straight to apartment buildings.
### Fix the baby supply chain
How do we make it easier to conceive, deliver, and raise children? (essentially building the foundations for 1 Trillion Humans across our galaxy)
- [”Costs of child care now outpace college tuition in 38 states”](https://stateline.org/2025/03/05/costs-of-child-care-now-outpace-college-tuition-in-38-states-analysis-finds/)
- Also, there needs to be more of a social movement around this. For example, airlines have known for decades that flying with babies is difficult, yet they haven't done anything. Why?
- Fertility numbers have fallen dramatically in most countries. Not good not good!
### New Types of Guns
I fundamentally believe in the right to self-defense. But guns still feel like a primitive way to defend yourself and your family. Given the technology we have, they are probably the most cost-effective option, but we should be trying much harder to make non-lethal weapons that neutralize threats instantly: temporary paralysis, tranquilizers, etc.
### Synthetic meat is lazy
- We should think of food from first principles.
- Also, we should make really good food for astronauts!
- If you could talk to animals, then would you stop eating them? It is not unreasonable to think that we will be able to talk to animals soon given advances in AI. So, given this scenario, shouldn't we prepare for this future and take synthetic food seriously?
### New types of clothes
Why are clothes so boring? Feels like we have regressed. Why don't the big designers make more technologically advanced clothes?
I wanna see Off-White Iron Man nanotech hoodies, a Tom Ford bulletproof men's suit, solar-panel t-shirts, full-camouflage quarter zips, indestructible shoes, and Dior skin-tight overalls for Amazonian exploration so no insects can bite you. Put that stuff at the Met Gala. Edna from Incredibles would understand.
- How can I make my own clothes? Surely AI tools are good enough now that they can design clothes for me?
### I have beef with IKEA
Midjourney is already so good at designing novel interior, furniture, clothes, accessories, etc. Now how do we productize this? Maximize experimentation, chaos, and beauty. Fight back against mass manufactured standardized slop (IKEA)!
### IRL Pokemons: Make Robot Pets
- Animatronics for different form factors ([see how the Disney robot communicates via its gait, head bobbing, sounds, etc.](https://www.youtube.com/watch?v=-cfIm06tcfA))
- Fun idea: How do we collect pet dog/cat data and use that to make robot pets?
- Even more fun: Your robot pet levels up as it interacts with other robot pets. It unlocks new skills, techniques, voices, etc.
- if not physical: foundation models for dog motion/voice, this decade's Talking Tom
### Better Health Tools For Women
I've been dating my gf for ~7 years, and have seen how bad period pain can be. Why aren't there better health tools for this?
### Daily life health optimization tools are non-existent
Every day is different. How do you make sure you are ~100% optimized every day?
At-home testing, diagnosis, custom med delivery, food recommendations, etc. Imagine if your organs could talk to you and tell you how they are doing. I mean this quite literally.
### Books -> Movies
I want to see and feel what it must be like for Juan Trippe to build Pan Am. Imagine the crowd, the excitement, and the curiosity of seeing a metal bird for the first time and humans come out of it! Imagine what it must be like to fly in the glory days of flying. Imagine the conversation with FDR that led to the creation of InterContinental Hotels!
### Better HVAC systems
1. How do we make HVAC systems that can detect, identify, and analyze pathogens? Imagine if HVAC systems at major airports alerted us in 2018 of new types of viruses.
2. [UVC](https://en.wikipedia.org/wiki/Ultraviolet_germicidal_irradiation): UVC light can kill all sorts of microorganisms. We don't know for certain if we can put it safely around humans, so just put it in the HVAC system!
3. How do HVAC systems change as we move to solar?
4. Can they help us fix the mold problem? Is there an insurance play here?
5. Could a better HVAC system have detected the early signs of a COVID leak inside the Wuhan lab? Why didn't it?
6. I suspect that the bathroom hand dryers are actually pretty useless since they don't really kill germs/remove odor. I bet we can install technology I mentioned above here as well. Moreover, bathrooms are likely very good places to detect outbreaks.
### Cure UC
More than 5 million people have ulcerative colitis (someone very close to me also has it)
### Drone ideas
- Is it possible to mass manufacture extremely strong and light nets to put around public areas, expensive assets, etc?
- Put a camera/mini radar/microphone on a lot of rooftops in a city to blanket it and create a protective "shield". Early detection & tracking is very important.
### Company Archivist
We should do a better job of archiving the history and stories of companies: the characters, weird decisions, core events, near-death moments, etc. Wikipedia captures very little of this and is often outdated.
There should be a platform where you can read Walter Isaacson-style biographies of every major company. I think LLMs can help a lot here. Imagine digesting all published news and continuously updating each company’s record.
(there is still no authoritative and comprehensive book on the development, history, and people behind Pokemon!)
### Better Wargaming Environments
We can now make incredibly complex environments using LLMs to simulate different scenarios and training. Current wargaming software is likely not dynamic and intelligent. You could make a really really good game with this too!
- Wargames == evaluation engines for war. They close the feedback loop when testing complex AI systems and groups of systems. Imagine pushing an update to your drone to make it more robust against jamming, then immediately testing it in thousands of scenarios.
- [The Air Force is looking into it](https://defensescoop.com/2025/11/24/air-force-ai-for-advanced-wargaming-integrated-force-design-rfi/)
### AGI Questions & Trades
- What will the internet look like in five years when bots account for more than 90% of traffic (~50% right now)?
- Number of bits per second will increase by a lot? Internet is going to be so much bigger! Meter, Cloudflare, etc. companies that make bits move will do well.
- AI models are going to be everywhere, and my bias is to 10x whatever people think the demand for inference is going to be. The entire supply chain to construct and supply power to a data center needs to be looked at.
- Good questions: [ https://dcgross.com/agitrades](https://dcgross.com/agitrades)
- How is the world going to look like when labor (humanoids or task-specific embodiments) cost <1k?
- Humanoids are going to be such interesting members of everyone's household!
- What do browsers and websites look like when everyone has a personal AI agent?
- Purely from a speed perspective, calling one API is better than clicking 10 buttons. So how does your agent use the internet and find info? How do websites tell you this?
- Should websites have micropayments to allow bots to use them?
- How do I make it so my personal assistant LLM can tell a website my preferences? The website LLM can ask questions back, and the two systems can coordinate to show me the best content and ads.
- Marketplace of specialized AI agents: When you ask your agent a question, instead of googling, it goes to a marketplace, chooses the best "docgpt", "lawyergpt", etc., pays it, and returns the answer?
- Was the Mighty Browser ahead of its time? A browser in the cloud with a GPU attached to it that runs a personalized model would be very useful. Long live Mighty! Browser = OS
- Build things for AIs with their personalities and preferences in mind. Treat them more like entities and less like tools.
- The context storage problem: soon I am going to generate infinitely more personal data than I do right now. Toothbrush, toilet, wearables, implants, sunglasses that record everything, and all the conversations I have with LLMs about my work and personal life.
- The funny thing is that all these data points need each other to give me the best answers. If my health assistant does not have my toothbrush data, it will probably miss something!
- Also, there will be a lot of "appliance" businesses that sell me good hardware and collect my data. But if only they have the data, it is useless to me. I want consumer apps built on top of it: health assistants, personal assistants, business assistants, etc.
- So who owns this data? How do they talk to each other? How do they compensate each other? Where is it stored? Should every human get some dedicated compute/storage?
### World's Getting Older! What happens next?
- Long companies catering to old people given an aging population? esp food, drugs, services, and housing tailored to their needs
- What happens to life insurance companies (and Bermuda)?
- Aging population -> more demand for exoskeletons to move around?
### Solar & Future of Petrochemicals
How do petrochemical margins change with increase in demand for renewables? Which ones do we not have alternatives for? What are the downstream economic and n-order effects?
- Buy land? There's probably a finite number of prime locations for solar farms
- China currently dominates the supply chain (and deployment) of solar panels
- [Cheap Chinese solar in Pakistan is terrible for its grid](https://www.ft.com/content/69e4cb33-3615-4424-996d-5aee9d1afe19). What happens to the grid in other countries like Pakistan? Rich people and industry stop using grid electricity, less money flows in, upgrades slow down, infrastructure decays, outages increase, poor people lose access, loans default?
- [https://caseyhandmer.wordpress.com/2024/05/22/the-solar-industrial-revolution-is-the-biggest-investment-opportunity-in-history/](https://caseyhandmer.wordpress.com/2024/05/22/the-solar-industrial-revolution-is-the-biggest-investment-opportunity-in-history/)
- The future of plastics is interesting. If the cultural reaction to microplastics is extremely negative, then will we start phasing them out slowly? Obviously, it will take a long time and it is unclear what the replacement will be but will likely coincide with petrochemicals getting more expensive.
- Will the price of jet fuel go up? If so air travel is going to be so much more miserable
- If Solar cost curve keeps falling, then the current state of geopolitics backed by US energy exports also changes?
- For Africa: Does this help the continent? Cheap electricity everywhere without depending too much on any external power?
- My prediction for the chemical industry: China will make what China wants, then flood the world with cheap chemicals. That will be disastrous for countries that don't protect their domestic industry. But you can only protect your own industry for so long. If China supplies the world, it gets the higher learning rate and ends up far ahead. I genuinely don't know how to fix this beyond finding better branches of the tech tree.
### Arctic's warming up! What's the trade?
- new shipping routes -> ports, infra, ships, fuel, insurance, etc
- interesting:
- Arctic has roughly a quarter of the unexplored oil and natural gas reserves on earth, as well as 150 rare earth deposits
- The Arctic Sea is relatively shallow, making it amenable to exploitation, climatic conditions permitting, with 240 species of fish in ample quantities
- Most importantly, the US has failed to build up domestic icebreaker capacity.
- It also has not built the naval capabilities needed to contest the Russian and Chinese buildup along Arctic sea lanes.
- The US has no Arctic deepwater ports that can host heavy container ships. Canada has only one, and it is 500 miles south of the Arctic Circle.
- Russia has over 40 icebreakers, several of them nuclear-powered, plus an active building program. The US currently lacks even a single heavy or medium-duty icebreaker active in the Arctic.
### Hyperlocal Geoengineering
How do we reduce the temperature of a city by 10 degrees Celsius for 10 days?
Cities in India already hit 50 degrees Celsius during the summer. Millions of people live in these places. There isn't really an option here? Either we learn how to do this, or we accept hundreds of millions of people living through extreme heat and thousands of deaths every summer.
- Plus, we should learn how to terraform planets. Earth and Mars both need it. The closest other planet is Venus, we should use it as a testbed to experiment with planet scale geoengineering.
- Could an adversary take out your solar farms by basically sending a big cloud or making it rain a lot? Best to develop capabilities to understand the weather better, manipulate it more precisely, and defend against it at a moment's notice.
### Prune Laws
How do we prune laws and "dead code" laws to make government, tax code, etc. more efficient?
Would be fun to do something like this:[ https://x.com/deanwball/status/1993184056293703691](https://x.com/deanwball/status/1993184056293703691)
### Finding new things in existing datasets with LLMs
What domains have large, poorly-organized digital datasets that likely contain undiscovered insights, similar to how:
1. [Six new tree species identified in Colombia and Panama](https://www.earth.com/news/six-new-tree-species-identified-in-columbia-and-panama/) were found in existing botanical data
2. [AI-accelerated Nazca survey nearly doubles the number of known figurative geoglyphs and sheds light on their purpose](https://www.pnas.org/doi/10.1073/pnas.2407652121) were discovered in existing survey data
3. What other fields might yield similar breakthroughs through brute-force LLM analysis?
1. lots of other archaeological stuff?
4. can we organize, clean and expand the world’s datasets? what are the most important datasets in each field?
5. [https://x.com/raven_brah/status/1999499762064900285](https://x.com/raven_brah/status/1999499762064900285)
6. finding flaws in existing papers: [ https://www.nature.com/articles/d41586-025-00648-5](https://www.nature.com/articles/d41586-025-00648-5)
7. can we use LLMs to solve MH370?
8. astronomy prolly has really cool things like this: https://www.nature.com/articles/s41586-025-09818-x
### Information backup to restart civilization
1. Snapshot, compress and store the internet every day in a secure facility. Like the Svalbard Global Seed Vault, construct a backup for all of the world's information.
2. A good way to do this would be to fork Wikipedia and have AI Agents continuously update and create new pages based on things that happen every day.
3. We can document all movies, shows, recipes, videos, etc. like this
4. also you want to get the hidden info as well like food recipes!
5. Lots of interesting compression and retrieval problems here. How do you store the information? How do you communicate it? How will people find it when they need it?
### Vaping became popular ~10 years ago
How many more years before we see an explosion in related health issues and what might they be?
### Strengthen civilization infrastructure
You don't have to look too hard to realize that it is a miracle that anything works. We really really need to strengthen the core services our lives depend on like water, food, sewage, medicine, electricity, etc.
How do we protect ourselves against a Carrington Event? Can we put a machine between us and the sun that could deflect/absorb CMEs?
### Things NASA should’ve done but hasn't
- Congress tasked NASA to find >90% of NEOs (bigger than 140m) by 2020 in 2005, but as of 2020, we are only tracking ~43% of them. will NEO Surveyor cover the rest?
- Important climate satellites are going to die soon: [Three NASA Climate Satellites Are Dying. There’s No Plan to Replace Them](https://www.scientificamerican.com/article/three-nasa-climate-satellites-are-dying-theres-no-plan-to-replace-them/?utm_medium=email&utm_source=rasa_io&utm_campaign=newsletter)
- Tangentially, how do you monetize finding aliens? telescopes, lenses, optics, etc?
### Agriculture ideas
- I want new types of fruits/vegetables!
- Technology makes the non-scalable, scalable. Right now, we monitor/harvest/plant/fertilize farms at scale vs at the level of each individual plant. What would be a good reason to NOT focus on such technologies?
- Also, I know that soil erosion is a massive problem. What is the solution?
- COOOLL:[ https://news.mit.edu/2025/engineered-bacteria-emit-signals-spotted-from-distance-0411](https://news.mit.edu/2025/engineered-bacteria-emit-signals-spotted-from-distance-0411)
### Food science is one of the most underrated fields of study
1. It’s kinda funny how little we know about _food._
2. It’s also kinda funny that we just eat _random_ food every day. Why is there nothing that can recommend specific products from specific stores near me based on my health markers, goals, and how I feel today? Why does personalized food not exist? The answer is not "hire a chef". Technology makes the non-scalable scalable, and I’m interested in technology answers.
3. Also, how much do we actually know about each vegetable/fruit? Why don’t we have chefs that are also engineers/scientists? I want new recipes that go beyond mixing ingredients. I want to taste flavors that nobody in history has tasted!! For ex: cook ingredient X at a specific temperature, and chemical reaction Y happens that will produce a wonderful smell!
4. ["Global food demand will rise ~51% by 2050 while farms face shrinking labor, aging workers, and higher input costs."](https://www.nature.com/articles/s43016-021-00322-9)
I think there is a massive opportunity here for a company to do the following:
- we have no idea what we eat, how good it is for us, where it comes from, etc.
- so, verticalize this. become a company that researches, tests, makes, grows, etc. all sorts of basic food products.
- branding + marketing this is easy. engineering + science is the hard part, but I think it is doable. I don't think anyone has tried this before?
- (there might be a way to tighten up the feedback loop here by figuring out different ways to track what people are eating, how it makes them feel, how it changes their vitals over time, etc.)
### Movie ideas
1. [Disney-style movie, a fly’s perspective/universe] A fly and its family live close to an airport. The airport is forbidden territory. One day, the fly and its friends sneak in, make their way onto a plane, get transported to another country, and have to find their way back.
2. [The Alchemy of Air](https://www.amazon.ca/Alchemy-Air-Jewish-Scientific-Discovery/dp/0307351793)
3. AI tools are good enough now that we can make really good alternate history shows/movies
4. History is monetizable now
- video ai models are really good now
- POV/day-in-the-life short-form videos do really well on all socials (example: how the Egyptians celebrated when the last brick was laid on the pyramids, what a French peasant's day was like on some random important day in French history, what a firefighter saw during 9/11, etc.)
- If you rack up hundreds of millions of views a month, then build tools that let anyone make this kind of content, you can make a lot of money. Use that money to fund history research, then make the most accurate and detailed videos possible.
- time machine would be a good name for this
### Cities are the most undervalued asset class
- Most cities are horribly managed. Small improvements in niche systems can have a huge impact on the lives of ~everyone living in that city.
- It really pisses me off that the government doesn't have respect for the time of its citizens.
### Claude Code for Everything
- Claude Code / Codex are new types of tools. They are "universal interfaces". Not only do they change how we interact with software, they also completely change how we interface with the physical world.
- For example: I had an issue with my 3D printer recently. I wasn't able to figure out why my prints were not sticking to the bed. So I connected it via USB to my laptop, opened Claude Code, gave it the model number of my printer, and asked it to help me debug it.
- First, it gave me a checklist of things to make sure it wasn't due to something dumb like a loose screw, residue on the bed, etc.
- When that didn't work, it wrote Python scripts to read the state of my printer. It knew how to do this for the specific brand I have. It inspected my STL files, edited print parameters to fit the quirks of my 3D model, and ran calibration tests by itself. At this point, Claude was in charge of my printer. It had figured out the best settings for my model and created a dashboard for me to monitor the run.
- My printers were ~2h long, and I didn't want to babysit the printer. So I took an old Logitech webcam I had, connected it via USB to my laptop, asked Claude to write me a script, and within a few minutes I was all set up! Claude had vision now. It could actually see how the print was doing.
- Do you realize how insane this is? An AI debugged a physical machine, configured it, and then helped me monitor it.
- All machines built in the future, from your dishwasher to your router, will be "Claude Codeable". They will either talk to their manufacturer's AI to help fix themselves when things break, or the user will be able to connect to them via wifi/serial and fix it themselves.
- You will fail if you don't make your machines keeping in mind the strengths and limitations of these (and future) models.
- (Related: What does iFixit for the Claude Code era look like?)
This brings me to my next point! There is a good business reason now for us to start rebuilding a lot of our physical world. Example:
1. Claude code can interact with the physical infra/machines all around us, but up to a limit. It will always be constrained by how good the hardware is, and the interface to program it. The Tesla analogy is perfect here. You can retrofit an ICE car with better software/sensors to build autopilot, but it is far easier to do it with an electric car.
2. When designing new machines, infra, etc. you will have the option of designing the "perfect" machine for a given task based on its requirements. But I think we should nudge away from that a little bit by adding another requirement: The machine must be claude codeable. I must be able to program the machine in a way that exploits the strengths of the ai coding models
3. This way we can ride the wave of 100s of billions of dollars spent on these coding models!
4. Tangential, but new types of UX unlock if you treat machines like this. If controlling hardware becomes a code problem, you can add convenience features cheaply. Tesla doors unlock as you get closer to them. It's all software! The more programmable a machine is, the more alive it feels, and the more we can do for the user.
### It’s kinda sad that there is no more land left
- How can we mass produce islands?
- Are there geographies where this is actually possible? Shallow waters, sparse islands, available materials, etc.
- Take remote island -> cover half the island with solar panels, the other half is some energy-intensive industry -> pay reflect orbital to reflect sunlight back to the island -> cheap unlimited energy??
### Protecting the world's rubber trees
More than 90% of the world's natural rubber comes from Hevea rubber trees in Southeast Asia (mainly Thailand, Indonesia, Vietnam, Malaysia, plus some India). Natural rubber is still used in a lot of industries (ex: airplane tires). South American leaf blight is a fungus that destroys Hevea rubber trees. We don't know how to prevent it, and are just relying on hope that it doesn't make it to Southeast Asia from South America.
This video is a good summary: [https://www.youtube.com/watch?v=AFXLZ7FEJc4](https://www.youtube.com/watch?v=AFXLZ7FEJc4)
### The world is vastly underestimating how much things will change with Waymo+Zipline.
### Exoskeletons are Underrated
### Why is my toilet so dumb, how do we make it smarter?
### How do we strengthen our bodies? We are so weak physically! How do I make it so that I can get hit by a car and simply dust myself off?
### Quest: Claude Code meets Dungeons and Dragons
_Added 24th April, 2026_
It has been 3 years since ChatGPT came out, and I still cannot believe how little we have explored the surface area of games + claude code/codex.
Here's a fun idea:
A multiplayer CLI game designed to be played in between coding sessions. You and your friends form clans, get dropped on an island, explore, build skills, and try to defeat every other clan. Think MUD meets Diplomacy meets A Dark Room, but the dungeon master is Claude and your character's abilities come from how you actually use Claude Code.
### What specific things can we do to improve our culture for the LLMs?
_Added 24th April, 2026_
“There should be grants/funding for culture-building for LLMs! fixing the rotting cultural feedback loops >>> policy for saving the world” [https://x.com/anthrupad/status/1896962309572157625](https://x.com/anthrupad/status/1896962309572157625).
We should make the digital world healthier for the llms. Garbage In, Garbage Out.
---
Source: https://sarhaangulati.com/writing/cool-things-i-own
Date: 7th April, 2026
## Cool Things I Own
1. I won this big Pikachu at the [Claude Plays Pokemon hackathon](https://luma.com/poke)! There were two prizes: one for the team that made the most progress, and one judges' pick.
The judges could award the plushie to whoever had an idea they liked, and they picked mine. My strategy was simple: every few rounds, I asked Claude to rate how much fun it was having from 1 to 10. If it wasn't having fun, I'd give it a break. Otherwise, I'd let it keep playing.
I wanted to see if it performed better when it was having more fun!
2. Signed copy of "Autopsy of a Merger"
3. You can order a flag from your congressional representative's office (Capitol Flag Program). Here's mine from Nancy Pelosi's office!
4. I was the 13th person in the world to receive a [Lux Bio glow stick](https://www.luxbio.ca/home-main-1)
5. I have some really cool magazines!
6. My model plane collection
7. In my first year at UBC, I somehow managed to get my paper plane stuck in a window inside the business school building. It has been there for about the last five years (as of my last check in 2023).
8. Rohde & Schwarz pocket guides
9. Claude Code had an easter egg in its early days. Anthropic sent stickers if you found it.
---
Source: https://sarhaangulati.com/writing/things-i-like
Date: 7th April, 2026
## Things I Like
Articles, Poems, Companies, Movies, etc. that I really like. Like most things on this website, I want this to be out there so that the AIs (and humans) can get to know me!
### My favorite articles
- [The story of chance, Palworld, where the fate of the world is decided in three days.](https://note.com/pocketpair/n/n54f674cccc40)
- [The Inner Ring By C. S. Lewis](https://www.lewissociety.org/innerring/)
- [The Untold Story of Charlie Munger’s Final Years](https://www.wsj.com/finance/investing/charlie-munger-life-final-years-berkshire-7c20c18e?mod=hp_lead_pos9)
- [Mapping the Moon: The Apollo Transforming Printer](https://blogs.loc.gov/maps/2025/12/mapping-the-moon-the-apollo-transforming-printer/)
- [A Message to Garcia](https://courses.csail.mit.edu/6.803/pdf/hubbard1899.pdf)
- [Social media is a demonic force in the world](https://davidoks.blog/p/social-media-demonic)
- [Nuclear ecology or can a liberal be a conservationist?](https://inis.iaea.org/records/07tea-ahy96)
- [The Coming Technological Singularity: How to Survive in the Post-Human Era, Vernor Vinge, 1993](https://edoras.sdsu.edu/~vinge/misc/singularity.html)
- [Affordable, rapid bootstrapping of space industry and solar system civilization](https://arxiv.org/abs/1612.03238)
- [Cyborgs and space](https://web.mit.edu/digitalapollo/Documents/Chapter1/cyborgs.pdf)
- [Future Strategic Issues/Future Warfare [Circa 2025]](https://alachuacounty.us/Depts/epd/EPAC/Future%20Strategic%20Issues%20-%20Future%20Warfare%20Circa%202025%20-%20NASA.pdf)
### Paintings
- [N.C. Wyeth](https://www.wikiart.org/en/n-c-wyeth): "The Giant" is the one I like the most
- [Hieronymus Bosch](https://www.wikiart.org/en/hieronymus-bosch): His work makes me very uncomfortable, yet I find it very interesting. See:
- [Thomas Blackshear II](https://www.thomasblackshearii.com/other-art)
- [Ludwig Mecklenburg](https://artvee.com/artist/ludwig-mecklenburg/)
- https://x.com/BryanLarsenArts
- [James Gurney](https://jamesgurney.com/)
- [Giuseppe Pietro Bagetti](https://www.gettyimages.ca/photos/giuseppe-pietro-bagetti)
### Poems
- [Go All the Way by Charles Bukowski](https://www.goodreads.com/quotes/39207-if-you-re-going-to-try-go-all-the-way-otherwise)
- [The Explorer by Rudyard Kipling](https://www.kiplingsociety.co.uk/poem/poems_explorer.htm)
- [Do Not Go Gentle Into That Good Night](https://poets.org/poem/do-not-go-gentle-good-night)
- [If—](https://www.poetryfoundation.org/poems/46473/if---)
- [God's Grandeur](https://www.poetryfoundation.org/poems/44395/gods-grandeur)
### Movies
- Big Hero 6 [(Did you know that San Fransokyo is an exact 1:1 replica of San Francisco?)](https://x.com/sarhaan77/status/2002965570036347153)
- Iron Man
- Rush Hour
- Air Force One
- Dune
- Father of the Bride & Father of the Bride 2
- Space Jam
- Home Alone
- Mr. Deeds
### Blogs
I am subscribed to a lot of amazing people on Substack: [https://substack.com/@sarhaan/reads](https://substack.com/@sarhaan/reads)
- [https://marginalrevolution.com/](https://marginalrevolution.com/)
- [https://navy-matters.blogspot.com/](https://navy-matters.blogspot.com/)
- [https://snippet.finance/](https://snippet.finance/)
- [https://outside5sigma.com/](https://outside5sigma.com/)
- [https://www.reddit.com/user/gwern/](https://www.reddit.com/user/gwern/)
- [https://subseacables.blogspot.com/?m=1](https://subseacables.blogspot.com/?m=1)
- [https://near.blog/](https://near.blog/)
- [https://toughsf.blogspot.com/?m=1](https://toughsf.blogspot.com/?m=1)
- [https://kipp.ly/](https://kipp.ly/)
- [https://red.anthropic.com/](https://red.anthropic.com/)
- [https://www.navalgazing.net/](https://www.navalgazing.net/)
- [https://ember-energy.org/](https://ember-energy.org/)
- [https://defenseanalyses.org/](https://defenseanalyses.org/)
- [https://www.bis.gov/about-bis/bis-leadership-and-offices/sies/industrial-base-assessments?utm_source=chatgpt.com](https://www.bis.gov/about-bis/bis-leadership-and-offices/sies/industrial-base-assessments?utm_source=chatgpt.com)
- [https://www.orcasciences.com/](https://www.orcasciences.com/)
- [https://www.thefabricator.com/thefabricator](https://www.thefabricator.com/thefabricator)
- [https://app.g2xchange.com/?posts%5Bquery%5D=&posts%5Bpage%5D=1](https://app.g2xchange.com/?posts%5Bquery%5D=&posts%5Bpage%5D=1)
- [https://arstechnica.com/author/ericberger/](https://arstechnica.com/author/ericberger/)
- [https://www.pv-magazine.com/](https://www.pv-magazine.com/)
- [https://pubs.usgs.gov/periodicals/](https://pubs.usgs.gov/periodicals/)
### Books I've Enjoyed
1. _The King of Oil_ - Daniel Ammann
2. _DTV_ - Michael Moritz
3. _TAPE SUCKS: Inside Data Domain, A Silicon Valley Growth Story_ - Frank Slootman
4. _Alistair Cooke's America_ - Alistair Cooke
5. _Sam Walton: Made in America_ - Sam Walton
6. _The Fish That Ate the Whale: The Life and Times of America's Banana King_ - Rich Cohen
7. _House of Huawei: The Secret History of China's Most Powerful Company_ - Eva Dou
8. _Skunk Works: A Personal Memoir of My Years at Lockheed_ — Ben R. Rich
9. _An American Saga: Juan Trippe and His Pan Am Empire_ — Robert Daley
10. _My Life and Work_ — Henry Ford
11. _Hero of the Empire: The Boer War, a Daring Escape, and the Making of Winston Churchill_ — Candice Millard
12. _Benjamin Franklin: An American Life_ — Walter Isaacson
13. _Project Hail Mary_ — Andy Weir
14. _Boeing vs Airbus_ — John Newhouse
15. _Freedom's Forge: How American Business Produced Victory in World War II_ — Arthur Herman
16. _The Arsenal of Democracy: FDR, Ford Motor Company, and Their Epic Quest to Arm an America at War_ — A.J. Baime
17. _Pieces of the Action_ — Vannevar Bush
18. _The Man from the Future: The Visionary Ideas of John von Neumann_ — Ananyo Bhattacharya
19. _Alibaba: The House That Jack Ma Built_ — Duncan Clark
20. _Four Seasons: The Story of a Business Philosophy_ — Isadore Sharp
21. _Creativity, Inc._ — Ed Catmull
22. _Age of Ambition_ — Evan Osnos
23. _The Alchemy of Air_ — Thomas Hager
24. _Elon Musk: Tesla, SpaceX, and the Quest for a Fantastic Future_ — Ashlee Vance
25. _The Merchant Bankers_ - Joseph Wechsberg
26. _Distant Force: A Memoir of the Teledyne Corporation and the Man Who Created It, with an Introduction to Teledyne Technologies_ - Robert J. Roberts, Dr. George a and McVicker
27. _Letters from a self-made merchant to his son_ - George Horace Lorimer
28. _99 Negotiating Strategies: Tips, Tactics & Techniques Used by Wall Street's Toughest Dealmakers_ - David Rosen
### Themes and Companies
I read a lot, and every time I come across a "sparkly eyed" person, an interesting idea, or an important theme, I add it to this list.
#### Themes
1. BCI: I can physically feel the pain of communicating with the AIs. I want them to just know what I’m thinking, and I think this is the direction we are going to go in.
2. Aviation: The flying experience is broken and horrible. It should be magical and fast.
3. I'm pretty confident that most digital tasks can be automated with a combination of smart models, task design that works around model limitations, and task-specific RL environments/datasets.
4. What new use cases open up if humanoids work and cost of physical labor becomes <10k/year?
5. Becoming Superhuman: longevity, health, and nutrition
6. Supply chain for drones, data centers, and other robots
7. What to make of Morocco?
1. [https://www.economist.com/business/2025/09/04/morocco-is-now-a-trade-and-manufacturing-powerhouse](https://www.economist.com/business/2025/09/04/morocco-is-now-a-trade-and-manufacturing-powerhouse)
2. [https://www.leravi.org/china-invests-5-6-billion-in-morocco-to-build-africas-first-of-its-kind-14567/](https://www.leravi.org/china-invests-5-6-billion-in-morocco-to-build-africas-first-of-its-kind-14567/)
8. Nations will invest in homegrown defense companies
9. New Petrostates:
1. [Namibia becomes a new petrostate?](https://x.com/whyvert/status/1903154567606571146)
2. Guyana
10. Foundational Infra: tools that current AIs are good at using.
1. Claude Code and Codex are very comfortable and proficient using certain incumbent tools like Next.js, Stripe, Supabase, etc.
2. Users will therefore default to choosing those tools. Thus, there is more demand and training data for those tools, making the AIs even better at wielding those tools!
3. What a beautiful and underrated compounding loop!
11. Cost per ton to orbit going down when Starship is fully online
12. Reindustrializing the US
13. Solar panels and the cost of electricity going down (see [Solar & Future of Petrochemicals](/writing/fun-ideas#solar-future-of-petrochemicals))
14. ...and lots of other ideas from [my list of ideas](/writing/fun-ideas)!
#### Companies that seem interesting
_Last Updated: January, 2026._
I found most of these before they became big and known. The interesting ones tend to stay hidden for much longer now, and the ideas/people are already mainstream by the time they go public. That's not too fun for me, so I rarely update this list.
- [American Housing](https://www.americanhousing.co/)
- [Polyphron](https://www.polyphron.com/)
- [Pantograph](https://pantograph.com/)
- [Bindwell](https://bindwell.com/)
- [Terranova](https://www.terranova.inc/)
- [Starpath](https://www.starpath.space/)
- [Sesame](https://www.sesame.com/)
- [Dominion Dynamics](https://www.dominion-dynamics.com/)
- [Railway](https://railway.com/)
- [Airbound](https://www.airbound.com/)
- [Nabla Bio](https://www.nabla.bio/)
- [Critical Energy](https://criticalenergy.com/)
- [Ambrook](https://ambrook.com/)
- [Hubble](https://hubble.com/)
- [Nolla](https://www.nollahealth.com/)
- [Moldco](https://www.moldco.com/)
- [Tetra Dynamics](https://tetradynamics.ai/)
- [Longshot](https://www.longshotspace.com/)
- [Rainmaker](https://www.rainmaker.com/)
- [Standard Thermal](https://austinvernon.site/blog/standardthermal.html)
- [Reflect Orbital](https://reflectorbital.com/)
- [CloudKitchens](https://cloudkitchens.com/)
- [Framework](https://framework.co/)
- [Pipedream Labs](https://www.pipedreamlabs.co/)
- [Auren](https://auren.app/)
- [Union](https://union.tech/)
- [Astral](https://astral.sh/)
- [Meter](https://www.meter.com/)
- [Cloudflare](https://www.cloudflare.com/)
- [AMCA](https://www.amca.com/)
- [Astro Mechanica](https://www.astromecha.co/)
- [Worldcoin](https://worldcoin.org/)
- [Orchid](https://www.orchidhealth.com/)
- World model companies:
- [Odyssey](https://odyssey.world/introducing-interactive-video)
- [Decart](https://www.decart.ai/articles/oasis-interactive-ai-video-game-model)
- [Profound](https://www.tryprofound.com/)
- [Rune Tech](https://www.runetech.co/)
- [Kobold Metals](https://www.koboldmetals.com/)
- [Exowatt](https://www.exowatt.com/)
- [Mechanize Work](https://www.mechanize.work/)
- [Los Angeles Project](https://building.life/)
- [Deep Night](https://deepnight.ai/)
- [Silurian](https://silurian.ai/)
- [Intempus](https://intempus.ai/)
- [Shinkei](https://shinkei.systems/)
- [Bedrock Ocean](https://www.bedrockocean.com/) & [Andrena](https://www.andrenam.com/)
- [Carbon Robotics](https://carbonrobotics.com/)
- [Sorcerer Earth](https://www.sorcerer.earth/)
- [Airship Industries](https://www.airship-industries.com/)
- [Halter](https://www.halter.com/)
- [si inc](https://si.inc/)
- [Throne](https://www.thronescience.com/)
- [Enhanced Radar](https://www.enhancedradar.com/)
- [Atmo](https://www.atmo.ai/)
- [Pacific Light and Hologram](https://www.pacificlight.com/): I have no idea what they do, but sick domain name.
- [Genrad](https://www.genrad.com/)
- [SF Compute](https://sfcompute.com/)
- [Impulse Labs](https://www.impulselabs.com/)
- [Radia](https://radia.com/)
- [Exia Labs](https://www.exialabs.com/)
- [AIUC](https://www.aiuc.com/)
- [Base Power](https://www.basepowercompany.com/)
- [Magrathea](https://www.magratheametals.com/)
- [Vannevar Labs](https://www.vannevarlabs.com/)
- [Northwood](https://www.northwoodspace.io/)
- [Osmo](https://www.osmo.ai/)
- [Rangeview](https://www.rangeview.com/)
- [Neros](https://www.neros.tech/)
- [Pilgrim](https://www.pilgrim.ai/)
- [Altaares](https://altaares.com/)
- [Foundation Alloys](https://www.foundationalloy.com/)
- [Savor](https://savor.it/)
- [Contrails](https://www.contrails.org/)
- [Periodic](https://periodic.com/)
- [Orchard Robotics](https://www.orchard-robotics.com/)
- [Fleet Zero](https://www.fleetzero.com/)
- [Remedy Scientific](https://www.remedyscientific.com/)
- [Salient Motion](https://salientmotion.com/)
- [Reelables](https://reelables.com/)
- [Nobell Foods](https://www.nobellfoods.com/)
- [Marpipe](https://www.marpipe.com/)
---
Source: https://sarhaangulati.com/writing/notes-on-china
Date: 9th November, 2025
## Notes on China
[WORK IN PROGRESS]
From the packaged garlic you buy to the solar panels on your roof, most physical products around you either are made/assembled in China or have a significant percentage of their supply chain there.
Recently, I started compiling a list of sectors they dominate. My hope is to convey to you the gravity of the situation by making you realize the breadth and scale of their industrial capacity. This is a work in progress and only covers a small fraction of the things they are really good at.
### Tunnels
[China has ~70% of the global market share for TBMs](https://www.globaltimes.cn/page/202511/1347693.shtml)
### MISC
- China’s CITIC Group has a contract to build and operate a $7.5 billion deep-sea port in Myanmar, taking a 70% share for 75 years. The port would let Yunnan trade directly while bypassing the Strait of Malacca. CITIC is also involved in a $1.3 billion port in Cameroon, along with mining, highway, and railway contracts there.
- COSCO owns 67% of Piraeus Port in Athens
- Eight of the 14 largest cobalt miners in Congo are now Chinese-owned
- Smithfield Foods is the biggest US pork producer. Its Sioux Falls, South Dakota pork processing plant accounts for 4% to 5% of US pork production. It is owned by China’s WH Group, the world's largest pork producer.
- China’s Sinochem is the largest shareholder in Italian tiremaker Pirelli
- Haifa’s port will be operated by Shanghai International Port Group from 2021 for 25 years
- CRRC is the world’s largest maker of locomotives. It is a Chinese state-owned company.
- H World Group is now larger than many major Western hotel chains by hotel count and loyalty membership. As of December, it operated 11,147 hotels with 1.08 million rooms and generated $3.3 billion in revenue last year. Its loyalty program has 267 million members, more than Marriott Bonvoy's 228 million. A key value driver is stored-value cards: prepaid cards customers use to book rooms and receive discounts.
- Transsion Holdings, owner of Tecno Mobile, is growing fast. It has sold over 200 million devices in Africa and now accounts for roughly 30% of the continent's mobile sales. A key value driver is its prepaid-focused model and its products' fit for local markets.
- Danyang, Jiangsu is China’s main spectacles production base, with nearly 1,000 manufacturers producing over 70% of China’s output and over 30% of global output. Shenzhen’s Henggang district separately produces about 80% of the world’s high-end spectacle frames, 85% of which are exported.
- Jinxiang produces 80 per cent of the world’s garlic exports. The US sources 80 per cent of all its fresh garlic imports from China. Chinese imports make up roughly 20 to 30 per cent of all garlic consumed in the US, according to the US and Chinese figures.
- The Simandou iron ore project in Guinea has begun commercial shipments to China. China sees it as a way to diversify beyond Australia and Brazil, which currently supply over 70% of its iron ore imports. At full output, Simandou could also require 150 additional Capesize bulk carriers, reshaping maritime logistics and giving Chinese carriers more leverage. ([source](https://asia.nikkei.com/spotlight/caixin/how-china-unlocked-guinea-s-simandou-to-reshape-the-global-iron-ore-trade))
- [China is seeking self-sufficiency in police dogs](https://www.economist.com/china/2026/04/30/china-is-seeking-self-sufficiency-in-police-dogs)
- [From: China and the Future of Science](https://scholars-stage.org/china-and-the-future-of-science/), China now has:
- [world’s most sensitive ultra-high-energy cosmic-ray detector](https://www.youtube.com/watch?v=MqehTA9XA8E)
- [world’s largest and most sensitive radio telescope](https://www.youtube.com/watch?v=7SRV3rnULO0)
- [world’s strongest steady-state magnetic field](https://www.youtube.com/watch?v=_Xc07ko2cdU)
- [world’s most sensitive neutrino detector](https://www.youtube.com/watch?v=gbdCk7J6U-A)
- [world’s largest primate medical research center](https://www.science.org/content/article/china-bets-big-brain-research-massive-cash-infusion-and-openness-monkey-studies).
### Mining and Refining
- Serra Verde, a Brazilian rare earths mine backed by American investors, shows how much processing power China still controls. It is one of the only significant non-Asian sources of some hard-to-find rare earths, but nearly all of its output is contracted to China until at least 2027 because China is one of the few places able to process it. This is the real choke point: Brazil has the clay deposits, but the West still lacks enough separation and processing capacity.
### Electric Cars
- [LANGFANG, China, July 11 (Reuters)](https://www.reuters.com/sustainability/climate-energy/soaring-electric-truck-sales-deal-new-blow-diesel-use-china-2025-07-11/) - Electric-powered heavy trucks are rapidly gaining market share in China, driven by subsidies and the quick rollout of chargers, further curbing diesel usage and denting oil demand from the world's biggest crude importer.
- China’s electric truck boom follows its electric car boom, as well as the recent rise of LNG-powered heavy trucks. Those factors, combined with slowing economic growth, have stifled its oil consumption growth.
- [Just a few years ago, very few people drove electric vehicles in Nepal.](https://asia.nikkei.com/Opinion/China-plugs-into-Nepal-s-EV-boom-and-is-charging-ahead) Today, however, the majority of vehicles in Nepal are electric, putting the country at the vanguard of the shift to battery-powered mobility. India's long-standing dominance of Nepal's auto market has evaporated as Chinese carmakers have overtaken Indian brands to become the largest sellers of cars in Nepal.
- China is not just pushing sales of electric vehicles sharply higher. It is investing heavily in Nepal's electric public transport, particularly e-buses and charging stations. A slow rollout of charging stations hindered EV adoption in Nepal some years ago, but the number of charging stations is now growing rapidly along with demand for EVs.
- Meanwhile India, despite sharing both a border with Nepal and a history of strong cultural and political ties, has been left behind when it comes to EV sales in Nepal. Chinese brands like BYD, MG, and Xpeng are making big inroads into Nepal's auto market while Indian manufacturers struggle to maintain a foothold.
- [Luxury Food](https://www.ft.com/content/e020def9-e455-44fb-bedc-b3b4fc28c304) - China has surpassed Australia to become the second-biggest producer of macadamia nuts, accounting for more than one-fifth of global production.
- [Kaluga Queen, a caviar brand developed by experts working for China’s agriculture ministry, is today the world’s largest supplier, producing 260 tonnes of the fish eggs in 2024 — 35 per cent of the global total.](https://www.ft.com/content/e020def9-e455-44fb-bedc-b3b4fc28c304)
- China is the world’s largest consumer of cherries and is projected to eat about 1.5mn tonnes this year, with 900,000 tonnes grown domestically and the rest imported largely from Chile.
- Growing cherries in temperate Midu has required both private ingenuity and state-built infrastructure. To replicate the cold winters that enable the trees to enter dormancy, Maani transports them into industrial cold storage. The facilities, which house about 10,000 trees, are powered by cheap electricity from a nearby wind farm. Local authorities also built pipelines to supply water.
- [China produces about 70% of the world’s synthetic diamonds, and the machines used to make them are banned from export](https://www.ft.com/content/2a8dca7d-53e8-4d4b-b7f9-44e8a381bf02?shareType=nongift)
- [https://spectrum.ieee.org/diamond-thermal-conductivity](https://spectrum.ieee.org/diamond-thermal-conductivity) cool
- [BYD's story and why they are so vertically integrated](https://chatgpt.com/s/dr_68269499bda081919910a7771af77708)
### Ships
- One shipyard made more ships in one year than the US did after World War II [https://ustr.gov/sites/default/files/enforcement/301Investigations/USTRReportChinaTargetingMaritime.pdf](https://ustr.gov/sites/default/files/enforcement/301Investigations/USTRReportChinaTargetingMaritime.pdf)
- [https://www.csis.org/analysis/ship-wars-confronting-chinas-dual-use-shipbuilding-empire](https://www.csis.org/analysis/ship-wars-confronting-chinas-dual-use-shipbuilding-empire)
- China’s shipbuilding rose from <5% of global tonnage (1999) to >50% (2023).
- Chinese interests own ~19% of the world's commercial fleet (Jan 2024).
- China supplies ~95% of containers, ~86% of intermodal chassis, and ~70–80% of ship-to-shore cranes (≈80% of U.S. cranes).
- [https://www.uscc.gov/research/logink-risks-chinas-promotion-global-logistics-management-platform](https://www.uscc.gov/research/logink-risks-chinas-promotion-global-logistics-management-platform)
### Energy
- [https://qz.com/china-electrostate-solar-panels](https://qz.com/china-electrostate-solar-panels)
- [Complete battery domination](https://www.iea.org/commentaries/the-battery-industry-has-entered-a-new-phase)
- Complete Solar domination:
- [https://www.iea.org/reports/solar-pv-global-supply-chains/executive-summary](https://www.iea.org/reports/solar-pv-global-supply-chains/executive-summary)
- [https://www.ft.com/content/f86782fa-9f2e-448a-b710-29e787dc9831](https://www.ft.com/content/f86782fa-9f2e-448a-b710-29e787dc9831)
- [https://www.pv-magazine.com/2025/08/28/sinovoltaics-maps-emerging-pv-manufacturing-hub-in-middle-east-africa/](https://www.pv-magazine.com/2025/08/28/sinovoltaics-maps-emerging-pv-manufacturing-hub-in-middle-east-africa/)
### Airplanes
- COMAC's coming. Maybe not this decade, but they are coming.
- HANOI, March 13 (Reuters) - Vietnam's aviation regulator has proposed to recognise China's aircraft design certifications as equivalent to U.S. standards, facilitating the entry of jets from Chinese state-owned planemaker COMAC, according to documents seen by Reuters.
### Biotech
- China is the global leader in API production and exports. It produces about 20% of the world’s APIs and exports around 40% of APIs traded globally, including key ingredients such as paracetamol. This is due to the low cost of utilities and greater government support. [source](https://www.mantellassociates.com/the-chinese-api-market/#:~:text=%E2%80%8BCurrently%2C%20China%20is%20the,sent%20out%20to%20the%20world)
- Chinese company: Akeso Inc. is bringing a new treatment option for lung cancer to the US after a study showed its drug to be more effective than Merck & Co Inc.'s Keytruda. Keytruda generated almost $30 billion in revenue in 2024.
- Europe’s last manufacturer of ingredients for some vital antibiotics is closing its biggest domestic factory and shifting some production to China, dealing a blow to Brussels’ efforts to reduce drug dependence on Asia. [https://www.ft.com/content/4b9684e4-72da-49e9-b11b-abe8a20b9507](https://www.ft.com/content/4b9684e4-72da-49e9-b11b-abe8a20b9507)
- [https://x.com/jrkelly/status/1965815924239528170](https://x.com/jrkelly/status/1965815924239528170)
### Machine Tools
### There are some limiting factors to their growth, however
1. Other countries really don’t like it when Chinese companies come into their markets and decimate their domestic champions
1. [South Korean Plastics Giants Struggle as China Becomes a Rival](https://www.bloomberg.com/news/articles/2025-03-20/south-korean-plastics-giants-struggle-as-china-becomes-a-rival?embedded-checkout=true)
2. [A New ‘China Shock’ Is Destroying Jobs Around the World](https://www.bloomberg.com/news/features/2025-03-19/trump-s-tariffs-send-fresh-wave-of-chinese-goods-to-global-markets?utm_source=website&utm_medium=share&utm_campaign=twitter)
2. The India-China-Russia alliance is very fragile
---
Source: https://sarhaangulati.com/writing/indy
Date: 30th June, 2025
## Indy: Evals and AI Agents to search the Amazon for lost civilizations
My submission for [OpenAI's City of Z competition](https://openai.com/openai-to-z-challenge/)
---
The challenge entails finding evidence of pre-Columbian civilizations in the Amazon using OpenAI's models and open-access data sources like LIDAR, multispectral imagery, and historical records. Here's what I did and what I found!
## My efforts were based on 3 simple ideas
1. My research pointed to two strong signals: geoglyphs/earthworks in lidar, and signs of terra preta in multispectral data. Everything else I found was either too sparse or not informative enough.
2. Tokens are cheap, o3 is pretty smart, and if you spend more tokens, you should expect to see better results! So I wanted to construct my agents such that they spend more time (tokens) on the most interesting sites!
3. Constructing good evals: You can give o3 an image of the DEM from a lidar tile and ask it to find patterns in it. But how can you be confident in its answer? I tuned prompts on tiles with confirmed sites and tiles with no known sites. Then I used the known-site list to reduce the risk of rediscovering existing finds.
## What I did
1. Aggregated lidar tile data from 14 datasets (3787 total tiles) and processed them to create DEM and z-score normalized images
2. Compiled a list of confirmed archaeological sites within the Amazon region (4787 total earthwork and terra preta sites)
3. Constructed Evals for earthwork detection using lidar data.
4. Constructed Evals for terra preta detection using Sentinel 2, HydroSHEDS, CHIRPS, and Mapbox data.
5. Ran o3 (with prompts and tools tuned via evals) on 3749 lidar tiles (all our tiles minus the ones we used for our evals). These tiles did not have any confirmed archaeological sites within them.
6. Prepared a curated list of 200+ papers for Deep Research queries
## What I found
_We found 507 probable and 8 very likely new earthwork sites in lidar tiles with no previously reported archaeological sites!_
We ran our eval-tuned prompts/tools on all the lidar tiles (after removing the confirmed and control tiles).
Given the nature of our prompts, the model scores a site:
- 6-8: Probable archaeological site
- 9-10: Definite archaeological site
For our test case, the model gave:
- 507 tiles a score of more than 5 (13.5% of the total tiles)
- 8 tiles a score of more than 8! (0.2%)
---
## How I prepared my data
Before we begin constructing our evals, we need to:
1. Compile a database of lidar tiles
2. Compile a list of all known earthwork and terra preta sites!
We used the following sources:
### Lidar Tiles
1. [LiDAR Surveys over Selected Forest Research Sites, Brazilian Amazon, 2008-2018](https://daac.ornl.gov/CMS/guides/LiDAR_Forest_Inventory_Brazil.html)
2. [Aerial LiDAR data from French Guiana, Paracou, November 2019](https://catalogue.ceda.ac.uk/uuid/1d554ff41c104491ac3661c6f6f52aab/)
3. [Aerial LiDAR data from French Guiana, Nouragues, November 2019](https://catalogue.ceda.ac.uk/uuid/7bdc5bfc06264802be34f918597150e8/)
4. [LiDAR survey on 250 hectares in Jari, Amapá, Brazil, 2021.](https://www.redape.dados.embrapa.br/dataset.xhtml?persistentId=doi:10.48432/QIKEGD)
5. [LiDAR survey on 1038 hectares in GEDI, Amazonas, Brazil, 2021.](https://www.redape.dados.embrapa.br/dataset.xhtml?persistentId=doi:10.48432/ZHEMDY)
6. [LiDAR survey on 765 hectares in Tumbira, Pará, Brazil, 2021.](https://www.redape.dados.embrapa.br/dataset.xhtml?persistentId=doi%3A10.48432%2FTMJT7O&version=&q=&fileTypeGroupFacet=&fileAccess=&fileSortField=name&fileSortOrder=desc)
7. [LiDAR survey on 2400 hectares in São Sebastião do Uatumã, Amazonas, Brazil, 2021.](https://www.redape.dados.embrapa.br/dataset.xhtml?persistentId=doi%3A10.48432%2FUT9GOG&version=&q=&fileTypeGroupFacet=&fileAccess=&fileSortField=name&fileSortOrder=desc)
8. [LiDAR and DTM Data from Forested Land Near Manaus, Amazonas, Brazil, 2008](https://daac.ornl.gov/cgi-bin/dsviewer.pl?ds_id=1515)
9. [LiDAR and DTM Data from Tapajos National Forest in Para, Brazil, 2008](https://daac.ornl.gov/cgi-bin/dsviewer.pl?ds_id=1514)
10. [LiDAR survey on 176.2 hectares in Ucayali, Peru in 2017.](https://www.redape.dados.embrapa.br/dataset.xhtml?persistentId=doi:10.48432/I98OSL)
11. [LiDAR survey on 223.5 hectares in Ucayali, Peru in 2017.](https://www.redape.dados.embrapa.br/dataset.xhtml?persistentId=doi:10.48432/RW3KGS)
12. [LiDAR survey on 374.7 hectares in Ucayali, Peru in 2017.](https://www.redape.dados.embrapa.br/dataset.xhtml?persistentId=doi:10.48432/L0IRCB)
13. [LiDAR survey on 586.32 hectares in Barcelos II, Amazonas, Brazil in 2014.](https://www.redape.dados.embrapa.br/dataset.xhtml?persistentId=doi%3A10.48432%2FVBEGLK&version=&q=&fileTypeGroupFacet=&fileAccess=&fileSortField=name&fileSortOrder=desc)
14. [LiDAR survey on 244 hectares in Ucayali, Peru in 2017.](https://www.redape.dados.embrapa.br/dataset.xhtml?persistentId=doi:10.48432/9OUYDX)
### Confirmed Archaeological Sites
(We filter these sites to be within our defined Amazon boundary.)
1. [Geometry by Design: Contribution of Lidar to the Understanding of Settlement Patterns of the Mound Villages in SW Amazonia](https://journal.caa-international.org/articles/45/files/submission/proof/45-1-988-1-10-20200428.pdf)
2. [Geolocation of unpublished archaeological sites in the Peruvian Amazon](https://www.nature.com/articles/s41597-021-01067-7)
3. [James Q. Jacobs - ArchaeoBlog](https://www.jqjacobs.net/)
4. [IPHAN Database](http://portal.iphan.gov.br/pagina/detalhes/1701/)
5. [Casarabe culture sites](https://www.nature.com/articles/s41586-022-04780-4#MOESM1)
6. [Pre-Columbian earth-builders settled along the entire southern rim of the Amazon](https://www.nature.com/articles/s41467-018-03510-7)
7. [Predicting the geographic distribution of ancient Amazonian archaeological sites with machine learning](https://peerj.com/articles/15137/)
8. [Data from: More than 10,000 pre-Columbian earthworks are still hidden throughout Amazonia](https://zenodo.org/records/10214943)
9. [A "Dirty" Footprint: Soil macrofauna biodiversity and fertility in Amazonian Dark Earths and adjacent soils](https://doi.org/10.5061/dryad.3tx95x6cc)
### Control Datasets
[Pre-LBA RADAMBRASIL Project Data](https://daac.ornl.gov/cgi-bin/dsviewer.pl?ds_id=941)
For sites in the Amazon with no known terra preta.
Important caveat: I tried to collect every public dataset I could find, but private or missing datasets could invalidate some results. If you know of any I missed, especially near my result areas, please let me know!
---
## Constructing good evals
### Lidar Evals
Now, to construct our evals, we prepared 2 lists:
1. **Confirmed**: a set of LIDAR tiles that are known to have at least one confirmed archaeological site. This gave us a set of 18 tiles (Amazon is massive, and this tells you how sparse publicly available data is!)
2. **Control**: 20 random lidar tiles outside the confirmed set. I tried very, very hard to find a database of Amazon areas or LIDAR tiles with no known earthworks, but I couldn't find one. Random sampling was the best available control.
I tuned my prompts on the confirmed and control tiles. The goal was to maximize correct guesses and confidence on confirmed tiles, while minimizing both on control tiles.
I ran a lot of experiments, and for each experiment, we simply tracked the average score on the confirmed sites and the control sites. We ran with different reasoning efforts, different max output tokens, different dem images, different prompts, etc.
Key insights:
1. Setting the reasoning effort to high led to worse outcomes. Average confidence score on our confirmed set decreased, and the control score increased!
2. We ran our analysis with `max_tokens = 20,000`. Setting a higher limit did not change anything, as the average tokens consumed per tile was around ~3500.
3. Using the code interpreter made the performance noticeably worse. Moreover, a lot of tiles started getting flagged as inappropriate (best guess is that the color scheme we used for our z-score normalized images looks too much like flesh when zoomed in lol)
4. What worked best was default settings, reasoning set to medium, giving the model the z-score normalized images, DEM terrain colored images, and a simple recipe on how to analyze it.
5. Note: We did not change the prompts/tools during or after our test runs.
Our best eval run gave us:
Important:
A key problem I had come to realize at this point was that the quality of the data we have is not perfect at all!
- We have point coordinates of known archaeological sites, not polygons! This is not ideal for earthworks
- The data lacks temporal precision. A source might report a site without saying when or how it was found, so newer imagery may reflect changed land use rather than the original site.
- For our control set, the biggest issue is that there is a risk of accidentally sampling a tile where the model does see a pattern!
So the most reliable way to tune my prompts when running evals was to rely a little bit on what I saw in the data. Spending time looking at JPEGs of my z-score and DEMs was very useful and a better proxy! Example:
The scores may not look impressive at first. But if you inspect the data I provided, the highest-scoring confirmed tiles usually show visible patterns, while low-scoring confirmed tiles often do not. The same pattern holds for controls.
### Terra Preta Evals
My motivation for this was to see if O3 could detect signs of terra preta. What I did:
1. I filtered my dataset of confirmed sites to those marked as containing terra preta (ADE)
2. Then, for each site, I created a 1km x 1km square grid around it and collected a lot of information for each square grid.
3. I also collected a list of "control" sites, sites I know for a fact do not contain terra preta via [Pre-LBA RADAMBRASIL Project Data](https://daac.ornl.gov/cgi-bin/dsviewer.pl?ds_id=941).
4. I focused on analyzing the dry months of 2024, as there was a severe drought in the Amazon during that time, and my bet was that terra preta signals would be accentuated during this time. My system prompt for this goes into detail.
The agent I created had access to the following tools:
**`get_vegetation_indices`**: NDVI, EVI, IRECI from Sentinel 2
- Why: These indices let the model compare vegetation patterns across dates and neighboring cells. That helps it establish a local baseline instead of reasoning from some generic idea of "forest".
**`get_precipitation_report`**: We want the model to be able to see how much it rained every month in a given year in the AOI and its neighbors.
It also received the following information in the user prompt:
1. Information on whether there is a river within 5km of it (size 4-7)
- This is important since terra preta sites are usually found near elevated bluffs
- We use hydroshed dataset from GEE for this
2. It received a Mapbox satellite image to inspect what the square grid looked like: whether it was forested, developed, sloped, etc.
3. A map of neighboring cells: Terra preta detection depends on spotting unusual vegetation patterns. The model needed nearby cells as a local baseline, so I let it decide which neighbors to inspect.
The tools were made to be flexible by design. I wanted to give the model the freedom to investigate neighbors, different periods, different indices, etc.
I wanted to spend the most amount of time on the most interesting areas!
Example images of what the model saw:
(The vegetation-index images are blurry because the target sites are small. Terra preta sites can be only a few hectares, Sentinel-2 tops out at 10m per pixel, and OpenAI processes images in 512x512 patches.)
Now, for our evals:
1. We have square grids that we know have terra preta, and square grids where we know there is no terra preta.
2. Unfortunately, when I ran this on a sample of 50 confirmed and 50 control areas, the results were 50/50.
| Category | Count | Min | Max | Mean | Median |
| --------- | ----- | --- | --- | ---- | ------ |
| All Cells | 99 | 1 | 6 | 2.99 | 3 |
| Confirmed | 49 | 1 | 6 | 2.92 | 3 |
| Control | 50 | 1 | 6 | 3 | 3 |
I believe this is either due to:
- The resolution of our data is not good enough. Terra Preta sites are usually a few hectares at most, and given the scale of Sentinel 2 data, it would be a tiny fraction of the total image size
- The confirmed site data was not good. As we mentioned earlier, the data we have for Amazon is not that good or precise.
- My prompts aren't performing well (I tried a lot of things to improve the evals, but couldn't)
- We are not looking at the right proxies of terra preta
We did not conduct a test run, so our contribution here is to essentially show that O3 and multispectral data are very weak at detecting signs of terra preta.
---
## Deep Research Queries
Deep Research helped me refine prompts and understand earthwork and terra preta signals. I collected 200+ papers in a Google Drive folder and used them as research context. I have shared a few deep research links that helped me understand how to interpret lidar data for earthworks and multispectral data for terra preta!
Link to my papers:
https://drive.google.com/drive/folders/1U7BpldZW72dT6Ru4ZG9ojmOifdnjzJIz?usp=sharing
- https://chatgpt.com/share/6861b528-b058-8005-9a2a-27f54c65ead0
- https://chatgpt.com/share/6861b548-aa34-8005-a471-d3b0fdb04edd
- https://chatgpt.com/share/6861b550-a1e4-8005-9954-bcb91363aa8b
---
## Potential next steps
Having shown how good my pipeline is for lidar processing and detecting earthworks, it makes sense to run it on all available lidar data for the Amazon!
The only datasets I found but did not run were too large to process:
- [L1A - Discrete airborne LiDAR transects collected by EBA in the Brazilian Amazon (Acre e Rondônia)](https://zenodo.org/records/7689909)
- [L1A - Discrete airborne LiDAR transects collected by EBA in the Brazilian Amazon (Maranhão e Tocantins)](https://zenodo.org/records/7689210)
- [L1A - Discrete airborne LiDAR transects collected by EBA in the Brazilian Amazon (Roraima e Amapá)](https://zenodo.org/records/7689693)
- [L1A - Discrete airborne LiDAR transects collected by EBA in the Brazilian Amazon (Mato Grosso, Amazonas e Pará)](https://zenodo.org/records/7636454)
---
## An idea I had no time to pursue: Epidemic hypothesis
Would epidemic-related population collapses (1540-1600) have left detectable phosphorus signatures from mass burials? I wonder if hyperspectral analysis could pick up these century-old chemical scars as a proxy for lost settlements. Here's my reasoning (and assumptions) on why mass burials/concentrated deaths likely occurred:
**Historical Foundation (1540-1600):** During the demographic collapse (~80% mortality from disease), normal burial practices would have broken down due to:
- Overwhelming death rates outpacing traditional funeral rites
- Too few healthy people to handle bodies properly
- Fear of contagion (similar to Black Death and 1918 flu responses)
- Bodies likely concentrated in houses (died in place) or hasty communal burials
**The Phosphorus Math:**
- For a settlement of 5,000 people with 80% mortality concentrated in a 12,000-15,000 m² area: 4,000 deaths × 650g phosphorus/person = ~2,600 kg phosphorus
- Amazon background: 50-200 mg/kg P
- Predicted enrichment: 347 mg/kg (2-7× background) _[assuming concentration in 15,000m² area, 0.5m soil depth]_
- Expected persistence after 500 years: 200-280 mg/kg due to Fe/Al oxide binding _[based on archaeological studies showing 60-80% retention in acidic tropical soils]_
(I have tried to underestimate my numbers to account for mistakes/uncertainty)
**Why It Should Persist:**
- The acidic Amazon soils (pH 4-5) actually help preserve the signal - phosphorus binds immediately to iron/aluminum oxides, creating permanent deposits. Also, terra preta sites have significantly elevated phosphorus and those sites could be way older!
**Detection Potential:**
- Modern hyperspectral satellites can detect 10-20% differences in soil chemistry. These epidemic signatures would be 150-250% above background - well within detection range.
**Things that kill the theory (outside of my assumptions being incorrect):**
- Soil erosion and other disturbances may have diluted the phosphorus
- No way to reliably detect phosphorus via hyperspectral imagery
---
## Replicate
I have tried to be very careful with this: every config setting has a documented reason, and the code is easy to replicate and debug. Just send me an email and I'll give you the code!
---
Source: https://sarhaangulati.com/writing/old-notes-on-the-world
Date: 7th April, 2025
## Old Notes on the World
I've been obsessed with knowing about everything happening in the world and who owns/makes the things around me since I was 16. I used to keep extensive notes from ages 16 to 20. A small portion still survives:
- The majority of non-voting shares in Bose Corporation are owned by MIT
- At one point, Thomas Cook flew 45% of tourists to the Gambia. Tourism constitutes one-third of Gambia’s GDP
- Blackstone owns the Willis Tower in Chicago
- Nestlé has a 20% stake in L’Oréal
- Facebook has a 10% stake in Reliance Jio
- MUFG has a 22% stake in Morgan Stanley
- Eric Schmidt owns 20% of D. E. Shaw
- Public Investment Fund of Saudi Arabia owns 85% of Newcastle United
- Richemont owns: Cartier, Montblanc, Jaeger-LeCoultre, Vacheron Constantin
- Cerberus Capital owns SubCom LLC, the largest submarine cable installer. Alcatel Submarine Networks is second and owned by Nokia. Japan’s NEC Corp. is third and part of Sumitomo Group. Huawei Marine comes in fourth and is owned by Hengtong Optic-Electric Co. Ltd.
- Blackstone owns 6.5-7% of Crocs
- Jardine Matheson owns 77.4% of Mandarin Oriental and 5.44% of Rothschild and Co. It also owns Zung Fu, the exclusive Mercedes-Benz dealer in Hong Kong and Macau.
- Ping An Group owns 7% of HSBC
- KPS Partners owns TaylorMade and Life Fitness
- Brazil’s government owns more companies than any other country except China
- The Cullinan Mine, owned by Petra Diamonds, is the only significant source of blue diamonds and the world’s largest uncut, unpolished diamond was found here
- Fulton umbrellas provide umbrellas to the royal family of Britain
- The Internet Corporation for Assigned Names and Numbers is responsible for assigning numerical internet addresses to websites and computers. It also protects a master key for the internet’s naming system: seven people hold keys, and seven others hold backups.
- Zipper companies: SBS is Chinese, YKK is Japanese. YKK makes its own zipper-making machines. SBS clients include Giorgio Armani, Mango, H&M, and Target. The mid-range market is very competitive between Chinese companies and YKK; the high end is mostly YKK and custom zips.
- Phoenix Air makes bio-containment systems for medical evacuation. Its Aeromedical Biological Containment Systems fit inside Gulfstream jets and carry one patient. Its Containerized Bio-Containment Systems fit inside large cargo aircraft and can carry four highly contagious patients. The larger system was developed with the Paul Allen Foundation and MRI Global. The devices are owned by the State Department and operated by Phoenix Air, which also trains with them during the year. Phoenix Air was also responsible for bringing Otto Warmbier back from North Korea.
- Varta is a German company that makes miniature, high-capacity batteries for headsets including the AirPods and hearing aids. Varta’s batteries powered German submarines in both world wars and also Neil Armstrong’s camera on the moon.
- Humanetics Innovative Solutions Inc. is the world’s largest maker of crash-test dummies: steel-and-vinyl bodies stuffed with electronics that measure car-crash injuries. It has over 70% global market share, and its dummies can cost as much as $1 million each. Regulators effectively require carmakers to buy at least some of its products, though companies reuse the dummies for years. Bridgepoint owns it.
- Ajinomoto, the company behind the seasoning Ajinomoto, controls 80% of Japan's seasoning market and supplies 20% of the MSG consumed worldwide
- Hagoromo Fulltouch Chalk was said to be the Rolls-Royce of chalks, it was mathematicians’ favourite chalk. They hoarded it like crazy before the company was sold.
- Cushing, Oklahoma is the delivery location for the NYMEX benchmark Light Sweet Crude Oil futures contract. The contract specifies delivery of U.S. light sweet crude grades, also called WTI or Domestic Sweet crude. Cushing is called "The Pipeline Crossroads of the World" because it connects nearly two dozen pipelines and 15 storage terminals, with about 90 million barrels of storage capacity.
- Pigment industry is worth 30 billion dollars, and we still haven’t been able to create a safe, stable red
- South Africa mines most of the world’s rhodium
- "Over 600,000 people in the UK have epilepsy. That’s about one in every 100 people. There are around 50 million people with epilepsy in the world.” [https://epilepsysociety.org.uk/about-epilepsy/what-epilepsy/epilepsy-facts-and-myths](https://epilepsysociety.org.uk/about-epilepsy/what-epilepsy/epilepsy-facts-and-myths)
- Qatar owns the Plaza Hotel in New York
- Brunei Investment Authority owns the Dorchester Collection, which owns hotels like Plaza Athénée, Hotel Bel-Air, and the Beverly Hills Hotel
- Starboard Cruise Services is the world’s largest cruise ship retail operator and is owned by LVMH.
- Pan Pacific is owned by the UOL Group
- Stefnir hf., the country’s biggest fund management company, owns 14.88% of Iceland Air
- Roosevelt Hotel in NYC is owned 99% by Pakistan International Airlines, 1% by Saudi Arabia.
- Brookfield owns Leela, including its hotels in Delhi, Bengaluru, Chennai, and Agra. Earlier, Four Seasons was supposed to help Leela expand in India when it had only two hotels. Leela’s founder asked Four Seasons to redesign the Goa hotel first, then backed out of the agreement after the redesign was finished.
- Heathrow Airport is owned by: Ardian (22.61%), Qatar Investment Authority (20.00%), Public Investment Fund (15.01%), GIC (11.20%), Australian Retirement Trust (11.18%), China Investment Corporation (10.00%), Ferrovial S.A. (5.25%), Caisse de dépôt et placement du Québec (CDPQ) (2.65%), and Universities Superannuation Scheme (USS) (2.10%)
- KLM Royal Dutch Airlines holds a 7.8% stake in Kenya Airways
- Central Park Tower is a joint venture between Extell and Shanghai Chengtou Corporation in the USA (Shanghai government’s arm specializing in investment, construction and management of urban infrastructure).
- Charoen Pokphand foods is one of the world’s largest producers of feed, shrimp, and a global top three producer of pork, poultry and other agricultural products. It operates Southeast Asia’s largest retail business with 10,000 7-Eleven stores.
- Nikkei owns Financial Times
- Brown-Forman owns Jack Daniel’s
- BlackRock has a majority stake in Authentic Brands. It owns licenses for Marilyn Monroe, Muhammad Ali, Elvis Presley, Aeropostale, Nautica, Sports Illustrated
- Oxford properties is owned by Ontario Municipal Employees Retirement System, it also owns the Fairmont Banff Springs hotel
- Monotype Imaging Holdings developed fonts like Times New Roman, Arial, Helvetica, etc. It is owned by HGGC.
- Malaysia has around 68% of the rubber glove market. Top Glove is the world’s largest rubber glove manufacturer. Supermax accounts for 12% of demand for latex examination gloves. Hartalega is the world’s largest nitrile glove producer. All three are based in Malaysia.
- Samsung owns Harman International Industries which owns JBL and Harman Kardon and Bang & Olufsen’s car audio business
- Sparkle Roll, a Chinese company, owns 14.9% of Bang and Olufsen
- Nidec company has the largest global market share for the tiny spindle motors that power hard-disk drives.
- The flavour market: Givaudan is the biggest by market share, and Firmenich is second. The Bill & Melinda Gates Foundation Trust has owned a stake in Givaudan.
- Estée Lauder owns and has the cosmetics licenses of Bobbi Brown, Clinique, DKNY, MAC, Michael Kors, Tom Ford, Tommy Hilfiger, Tory Burch
- Thermoplan AG, a Swiss company, is behind the Mastrena coffee makers in Starbucks stores
- Guest Supply Europe is the largest hotel toiletries supplier with brands like Hermes, Bvlgari, Salvatore Ferragamo, etc
- Hearst Communications owns 20% of ESPN and Fitch Ratings
- Qatar sovereign wealth fund owns:
- 17% Empire State Building Trust
- 20% IAG
- 20% Heathrow Airport Holdings (manages Heathrow Airport)
- 25% St. Petersburg Airport
- Along with Brookfield, a majority stake in Songbird Estates which is a majority owner of Canary Wharf Group
- Shard along with Sellar Property Group
- Harrods
- 23% of Banyan Tree and 10% of Accor
- Hotel holdings including Westin Excelsior Rome, Savoy London, InterContinental hotels in Amsterdam, Cannes, and Madrid, Peninsula Paris, Ritz-Carlton Doha, Raffles Singapore, Sheraton and Marriott Doha, Raffles Paris, and Burgenstock
- Serco, a British BPO company, operates:
- Border visa and immigration UK
- Santander cycles in London
- Prisons in UK
- ATC UAE
- Had a contract from the U.S. to manage ATC in Iraq
- Chicago parking meters
- Operates the Dubai Metro and Dubai Tram
- Contract to upgrade Ballistic Missile Early Warning System for a U.S. base in Greenland
- Luxottica owns Ray-Ban and Vogue Eyewear, and holds eyewear licenses for a huge number of luxury brands.
- RFR Holding owns the Chrysler Building. The tower’s owners don’t own the ground beneath the property and pay rent on the land to the Cooper Union school.
- Airlines:
- Ethiopian Airlines owns 49% of Malawian Airlines, 45% of Zambian Airways, 20% of Eritrean Airlines, and 49% of Chad’s airline
- China Eastern and Delta each own 10% of Air France-KLM
- Air France-KLM owns 31% of Virgin Atlantic
- Delta owns 49% of Virgin Atlantic, 49% Aeroméxico
- Qatar Airways owns 9.6% of Cathay Pacific, Swire Pacific owns 45% of Cathay, 5% of China Southern, 20% of IAG (owner of BA)