BangleWei

@VittoStack I and @SaahirVazirani would be interested

4 days ago we launched Jailbroken, a PRIVATE Discord community to learn AI red teaming and safety. Since then: - Over 250 security researchers joined - Top resources have been collected - People shared countless techniques and discoveries Today, we've secured over 100B in FREE AI tokens for all the members. If you want to join, drop a comment.

@teaganyuen1 @bondhq_ Bullish

hi x, i'm teagan!! ✨ finally reviving this account, so here’s a quick intro: 2 months ago, i dropped out of boston college to move to sf and work full-time at @bondhq_ > i post startup content 20k+ @ teagan_talks on IG :) > im getting my skydiving certification soon!! > i used to yodel professionally when i was 7 new to the area so if you like trying new restaurants, extreme sports (bungee jumping perhaps), or bouncing GTM ideas off each other, would love to connect!!

@CorinWagen @berkbuilds @jamessinka super cool dude thanks! I respect the hustle!

@BangleWei reactorfield! cc: @berkbuilds @jamessinka reactorfield.com

pretty sure this combination of logos hasn't been seen in the wild before

@teaganyuen1 @bondhq_ Yodelling is new lore😭

The REAL "overnight success" math nobody wants to admit: - 18 months of 20-hour days building Cal AI - built on 10 years of jobs I got fired from - side projects that went nowhere - cold emails, free work, and coffees with strangers Luck didn't find me, and I’m nothing special. Almost every person I know who “made it” had 5-10 years of slow building and too many failures to count that nobody witnessed

When you hit a wall in math, coding, or any hard skill, do not immediately conclude that you lack talent. Most walls are just prerequisite debt finally coming due. Go back, fill the gaps, make the basics automatic, and the wall often turns into a staircase.

Innate talent definitely matters the further you go up, but it's no excuse not to try and give it your all.

@quanttejas @elonmusk bullish

I invented a way to supercharge any piece of content. It's a digital clone of the human brain that morphs to any audience and offers action insights Every post I make for the next 100 days will be optimized for exactly one reader: @elonmusk Day 1 of getting Elon to reply.

@jiahanjimliu Incredibly well done analysis and read.

$OUST: The Truth About Elon and Lidar This is my current long term view on Ouster: $OUST. However, with the stock up 67% 1-month, 157% 3-months, the short term is possibly overheated so I'm looking to sell CSP to increase my position. Elon doesn't hate lidar (1). Elon hates lidar for self driving but personally led the design to use lidar for SpaceX Dragon docking (1). What I see in Ouster is a technically strong engineering team building lidar for robotics, industrial, drones, and smart infrastructure. History Ouster started with mainly selling lidar for self driving cars and it's stock price's peak during the self driving hype of 2021 at 126.90. However, most of the self driving companies outside Tesla and Waymo either went bankrupt or were not competitive. Tesla famously does not use lidar and Waymo has been developing analog lidar in house starting in 2011 so $OUST bottomed at 3.72 in 2023. We have seen the resurgence of companies previously left for dead companies but with sufficiently complex to replicate engineering moats like SNDK, MU, CRDO, LITE, NOK go on a tear. It's hard to imagine a world when MU was left for dead but in 2023, fame value investor Mohnish Pabrai sold his entire 1.59 million share position of $MU for $64/share after holding it for 5 years because the stock did not move. For those looking to do the math, that's a 16.6x gain missed and Mohnish Pabrai was known for his patient value investing philosophy. The main take way here is that sufficiently complex engineering met with the right trends can deliver great returns. Ouster has a very component engineering team, which I will detail later, and refocused on lidar for robotics, industrial, drones, and smart infrastructure. This aligns with an possibly upcoming physical AI trend. I am not saying that OUST will be the next MU as HBM is more complex and has faster scaling demand compared to lidar. On the other hand, SK and Samsung have more dominant HBM positions than MU, and MU started at 72B market cap while OUST is at 3B market cap today. Engineering Tradeoffs: Lidar vs Camera Elon chose camera because he forecasted that with advancement in AI, FSD would be able to use camera to calculate depth, remember occlusion from glare and become effective in darkness. Today, we see that Waymo cannot go without lidar but Tesla can go without camera. However there is advantage that lidar innately has over camera: millimeter precision. For self driving cars that move at miles per hour, no one cares about millimeter precision. But for robots picking up things, pouring industrial chemicals, doing surgery, and Elon's use case of docking Dragon in space, millimeter precision matters. Yes, camera can work around this with tactile feedback, calibration, and visual servoing which one reason is why Elon focuses so much on Optimus' hands. Differences between Robotics, FSD, and LLMs LLMs utilizes mostly the entire internet for data with a mix of proprietary data and reinforcement learning. The resulting dynamics is LLMs is a winner takes most category for whoever has the best researchers and most compute. Robotics segments on the other hand are more distributed. Tesla and Waymo are not contenders for the best surgery robots, that's Intuitive Surgical $ISRG. Meanwhile Intuitive Surgical don't have a chance in full self driving. Granted within certain segments like full self driving, it's currently dominated by two top players: Tesla and Waymo but that's because these two have the most training data to imbue onto their models. Full self driving training data and surgery data cannot be sourced from the web, that's why OpenAI and Anthropic are not in the conversation for FSD or surgery robots. Thus it's likely that many segments of robotics will have their owner leaders. Open/Ant are not in the conversation for military drones: that's Anduril. Anduril is top for military drones but $ONDS has carved itself a great niche in defense drones in the civil government / private security domain. Even within military drones, Anduril dominates high tech drones while $RCAT has a niche in low cost disposable drones. Component Choice in Robotics Segmentation is likely in robotics, but there will probably be either a variety of industry standard platforms or at least components. For industry standard platform dynamic look no further than the Apple iPhone. The iPhone has both camera and lidar. Application developers on top of that can choose to use either, both or none. Whether robotics will have a variety of standard platforms or have industry standard components, lidar is a strong choice of sensor for various applications. Camera can serve as a workaround in some application but is not a universal sensor. Can you imagine using camera instead of $KRKNF sonar? You think reducing complexity would have prevented Blue Origin's landing pad explosions? You think Elon was reducing complexity when he chose to use lidar for DragonX docking? I am not a proponent of lidar in self driving cars but in many other applications, lidar reduces complexity. Reduction in complexity allows both human and Human-AI hybrid teams to iterate faster and focus on innovation instead of fixing regressions. Ouster: Pioneers of Digital Lidar Ouster pioneered the digital lidar. Analog sensors and compute rely on continuous signals converted from physics based feedback while digital relies on 1s and 0s. Analog is found in the natural world and in many domains starts out as the winner. Digital disrupts because it has Moore's Law working for it. Digital cameras disrupted film cameras. Digital thermometers disrupted liquid thermometer. MP3 disrupted Vinyl. Digital TV disrupted Analog TVs. VoIP is used for Zoom nowadays, not landlines. Most importantly, many analog digital signal processing circuits moved to digital signal processing. Ouster invented the digital lidar (4) and in 2026 Hesai, the lidar champion of China introduced their Picasso SPAD-SoC, their first digital lidar. Aeva is halfway digitalizing components in their FMCW lidar. Quanergy is has some digital components in their OPA lidar but is still focused on optical waves. The moat is not the concept of digital lidar itself but rather that Ouster's engineers moves faster than it's competitors. Digital lidar has best cost/performance improvements, lower manufacturing complexity, higher reliability, and has software defined behavior, and is more power efficient which is big in drones. The Team at Ouster Evaluating the team at Ouster: 1. Angus Pacala (5), CEO, BS, MS, Stanford Mechatronics. Was a top performing engineer at Quanergy, an incumbent lidar company, and saw improvements that could make a better product. This has vibes of Eric Yuan of Zoom who was previously a top performer at Cisco. 2. Mark Frichtl (6), CTO, BS, MS Physics. Interned at Palantir in 2012 and Apple Special Projects Groups (15). Mark and Angus were friends, startups co-led by friends is a huge green flag for long term success. Furthermore, interning Palantir in 2012 and Apple SPG in 2015 were top students technically. 3. Onur Ulusel, FPGA Lead (7) and Anil Semizoglu, Lead Calibration and Test (8), Feichen Shen, FPGA (9). Onur, Anil, Feichen all worked on ArgoAI's lidar. ArgoAI had the 2nd best lidar in self driving cars to Waymo. I didn't work with Onur, Anil, Feichen directly but they had excellent work and were well respected when I worked at ArgoAI. 4. Aman Bindra, Sr Staff Embedded Engineer (10). Worked 3 years 8 months at SK Hynix Memory Team. Memory firmware team at SK Hynix is among their best. Not every team at every company is bad or good. You have to recognize which team were the elite teams at each company. Ads Infra team at Meta is elite but metaverse team at Meta doesn't mean much. 5. Yan Zhao, Lead ASIC Architect (11). Worked 9 years 11 months at Ouster and 4 years at Oracle before that. Working a tenure like 9 years 11 months at Ouster is a great sign. This is continuation of knowledge and serious expertise building. This is a great indication of belief in the mission and strong outcomes at work. Yan graduated bachelors from Shanghai Jiao Tong which is equivalent to the Stanford/MIT of China. 6. Joshua Sakwa, Sr Firmware Engineer (12). Worked 4 years as Sr Staff IC Design at Broadcom. Although through acquisition, Broadcom makes targetted strategic acquisition and the fact Broadcom kept him for 4 years and promoted him to Sr. Staff means that he was key to the acquisition technology. Always known in Silicon Valley but now known by FinX is that Broadcom $AVGO is a top tier hardware company. 7. Karthik K. Principal Hardware Engineer (13). Worked 7 years 4 months at Ouster graduating from Berkeley. I'm not as impressed by staying at one company many years for a 40/50 years old but like Yan, early career, and Karthik fresh grad from UC Berkeley, being able to develop at one company signals a few things to me: - Ouster has the ability to attract and retain talent from top universities. - Ouster has institutional knowledge built over many years by long standing employees who worked here in the prime of their years. - These people had all the job opportunities during the hottest times in the bay area but saw something at Ouster that made them stay. Adoption 1. Lightning, developed by Chinese company Honor, has the world record for half marathon by a robot and uses both Ouster and Hesai lidars (15, 16). The event where Lighting set the record was in China. Can you imagine China importing EOSE batteries or $FSLR solar panels in China? China is dominant in lidar, yet their winning robot use Ouster in some parts of their robots over Hesai, Huawei, RoboSense, their national champions. If you move faster than Chinese companies in some electronic sensor component, that's a moat. This is not to mention that many applications are security sensitive that require US made key components for US robots. I mean, if Ford manufacturing plants need to be ready to be retrofitted to produce tanks again ala WW2, don't you think all the industrial lidars should be made by Ouster/Luminar or an American company instead of Hesai/Huawei? Seyond, the last key lidar company that should be mentioned in this post, is "headquartered" in Silicon Valley but is obviously a Chinese Company 2. Komatsu - Autonomous Mining. Multi-million supplier agreement with Ouster. Mining can be dangerous for humans, robotics are key here. 3. Utah Department of Transportation - Smart Infrastructure. Ever wonder why red lights could be smarter and turn red and green based on traffic instead of pre-set timers? Every wonder if road repair and expansion can be process by data instead of sending humans to survey? Utah Department of Transportation tested solutions from 6 lidar companies and found Ouster to be the best and put Ouster in 100+ of their intersections (17). 4. Chattanooga, Tennessee - same applicaiton to Utah Department of Transportation and for 120+ intersection (18). 5. Balyo - autonomous forklifts. 6. Trombia - electric street sweepers. 7. Microavia - Avalanche safety drones at ski resorts to monitor snow levels and avalache conditions. 8. Deep Forestry - surveying forests. 9. Mine Vision System - mining mapping. 10. SafeHaus - Crowd analytics at live events. 11. Avanti West Coast Rail - crowd analytics and station management at rail stations in UK. 12. Digital Mortar + Global Partners - shopper journey and layout analytics. 13. Argus - Germany Counter-UAS drones company will uses Ouster (20). Big field where precision matters for trajectory prediction for drone interception. Vertical Integration Ouster has true vertical integration because it designs the Integrated Circuits, ASICs, FPGAs, Firmware, and Software, and lidar domain specific AI models (19). Ouster provides software for object detection, classification, tracking and navigation for it's lidars. Ouster has shown the ability to expand to other product categories besides lidar. Their ZED cameras uses camera to produce depth in addition to all the camera functions. Remember what FSD does with Cameras? Ouster is adapting here too as there are some applications that prefer Cameras. Their vertical integration is notable because their sales to Utah DOT is Blue City - a turn key real time traffic management system which includes all of the software on top of their lidar. Financials I mostly analyze tech potential. Many other people analyze financials better than me so I'll just do the basics here. 30-50% guided revenue growth (19) and 49m revenue with 49% revenue growth - Q1 2026 (21). Q4 2025 62m revenue was an abnormally because of IP license contract. 43% gross margins (21). 175m cash in hand with Adjusted EBITDA of 7m loss. Ouster is pretty exciting from a tech perspective but at 3B market cap and 196m revenue that's a forward PS of 15.3 and 30-50% guided growth that's very expensive. I bought a tiny amount at $34 but am working to sell CSP and in parallel slowly DCA to increase my position. Forward PS of 15.3 is very expensive. For reference, IREN has a forward PS of 6.7, NBIS 8.3, MU 7.3, Nvidia 12.4. MU and Nvidia have 80% margins.

@AntonioSitongLi Dude this awesome, I'd love to talk to you guys about how you got cost down!

A robot for the price of an iphone. Meet Nori L2 Orders open next week

Why you should never bet on yourself. Tomo, launching june 25th

Great read

@seanrobins_ too popular dude

Can’t even go to a yacht party without being solicited

The number of posts from people who missed out on Cursor shows you need to keep putting shots on target. Don't quit, the glass is always half full, and there's light at the end of the tunnel:)

@itunpredictable @AmplifyPartners Applied!

We at @AmplifyPartners are launching a writing fellowship. It's for technical people (devs, researchers, etc.) with a focus area who want 2-3 months, some money, and my editing to research + write. Help us fight back against the onslaught of slop and dumb attention-maxxing.

@cory Still shows you the power of the Z-fellows network

congrats to the Cursor founders and team on the $60B acquisition to SpaceX this is one of those moments where I’m both genuinely excited for the team and kicking myself lol we were around the edges of Cursor in a bunch of ways, but I didn’t do enough to earn the chance to work with the founders directly it's still been cool to see the Z Fellows network show up around the company: the President of Cursor did Z Fellows I invested in the the COO of Cursor's pervious company Cursor acquired a Z Fellows alum company in an all stock (great) acquisition we were introduced to the founder of Cursor by the founder of Etched (Z Fellows alum) big L on my part that we didn't get a chance to work with the Cursor founders but excited for them, the team, and what's next :)

Anthropic CTO Rahul Patil. PESIT graduate. No IIT tag. No IIM tag. Yet he's helping shape the future of AI at one of the world's most influential AI companies. Your college can give you a start. Your work ethic, skills, and consistency decide how far you go. It's never over until you give up.

@DavidZagaynov @poseidonaero Watched your relentless video a few months ago, excited to see the next steps!

We built the world's largest cargo drone @poseidonaero