CJDGiesen
9.2K posts
The Terafab "Yield Buffer": Why 160k is the Real Number
Elon just clarified the math on Terafab, and the 60% jump in wafer starts (from 100k to 160k per month) tells a massive story about the reality of 2nm manufacturing.
In my original breakdown, I estimated 100k wafers/month to hit 100 million AI5 chips/year. That assumes a relatively mature yield (60%+).
Elon’s response, "Probably more like 160k wafers/month, factoring in yield", is a reality check.
The "Bleeding Edge" Tax: Launching a 2nm fab from scratch is historically difficult. By aiming for 160k wafers, Tesla is building in a massive safety margin. If initial yields are lower (closer to 35-40%), they still hit the 100 million chip target.
Monthly Starts: 160,000 wafers
Annual Capacity: 1.92 Million wafers
The Goal: 100 Million "Good" Chips
Required Net Yield: ~35% (The "Launch" yield)
The Upside: If yields hit 65%, output jumps to ~190 Million chips/year
The TSMC Benchmark: Matching the Giant
To put 160k wafers/month in perspective, look at TSMC. As of early 2026, TSMC’s entire global 2nm capacity (spread across multiple "Gigafabs" in Hsinchu and Kaohsiung) is targeting roughly 100k to 140k wafers per month.
By pushing for 160k, Elon is essentially saying that a single Tesla Terafab cluster aims to outproduce the entire world’s initial 2nm supply.
Elon Musk@elonmusk
@pbeisel Probably more like 160k wafers/month, factoring in yield
English
@nextbigfuture @elonmusk Nvidia is not in the business of robots or cars or llms. Nvidia is in the business of selling Ai compute solutions.
Their pitch right now: "Elon will eat all your companies unless you buy from us."
English
NVIDIA ROBOTAXI READY? IS TESLA DOOMED
Nvidia has new level 4 chips for autonomous driving.
Will this kill Tesla? @elonmusk
They have 19 car partners who collectively can make 18 million cars per year.
How many Nvidia Thor chips are installed in those cars ?
Jensen must have announced orders of many chips for many cars?
The Nvidia system will use LIDAR? There must have been many new orders for more robotaxi grade LIDAR for the millions of cars that Nvidia will enable with robotaxi ready systems.
Jensen must have had BYD demonstrate a BYD car with Level 4 driving. Surprising since BYD said all of their cars starting in 2025 would have a BYD made Gods Eye driving system. Why is BYD switching from Gods Eye to Nvidia? I thought BYD had solved their autonomous driving.
English
🔥Micron, one of the leading DRAM makers, reported $23.9B in revenue up from $13.6B QUARTER-OVER-QUARTER!
English
You seemed to be pretty well off.
But others don't have the luxury of leasing or purchasing a new car every 3-5 years.
Most Tesla owners own a Tesla with H3, and they expect H3 to be fully autonomous, like Elon said.
I don't have cash to spend 65k for another model Y, but I have $10 to buy a new album every month.
English
Tesla’s forthcoming AI5 uses a half-reticle design, which is crucial for yield. A reticle defines the imaging area of a lithography machine, fitting two chips per shot effectively doubles yield.
This means the Tesla chip design team had to carefully manage die features, for instance dropping the older ISP (and classic GPU) to make room for more AI cores.
By contrast, NVIDIA’s Blackwell fills nearly a full reticle, making it a single-reticle design.
If Tesla hits its compute and efficiency targets with AI5 in this half-reticle format, it’s almost like cutting fab requirements in half.
And this has a big impact on Terafab, especially if it carries forward for AI6, AI7, etc.
phil beisel@pbeisel
Terafab may be the most essential vertical integration Tesla has ever undertaken— and it is truly non-optional. It will take years to build and will test even Elon’s speedrunning abilities to the limit, but that won’t stop him from trying. The breakthrough likely lies in overhauling the overall facility’s cleanroom model. By moving wafers in sealed pods with localized micro-environments, the fab no longer needs a monolithic ultra-clean space. Elon’s line about “eating cheeseburgers and smoking cigars” on the fab floor isn’t silly, it’s the practical reality of a radically simpler, cheaper, faster approach that could finally change the economics of chipmaking. This is all forced by the brutal “pinch” in chip supply. Tesla must produce on the order of 100–200 billion AI chips per year just to saturate its roadmap. That volume powers: FSD cars & Robotaxis (tens of millions of vehicles needing AI5 inference for near-perfect autonomy), Physical Optimus (scaling from thousands today to millions per year, each requiring AI5/AI6-level compute), Digital Optimus (the new xAI-Tesla software agents for digital/office automation, running massive inference clusters), Space-based data centers (AI7/Dojo3 orbital compute for GW-scale training and inference beyond Earth limits). AI5 delivers the ~10× leap for vehicles and early robots; AI6 shifts focus to Optimus + terrestrial DCs; AI7 goes orbital. No external foundry (TSMC, Samsung, etc.) can deliver that scale or timeline— hence the Terafab launch. Without it, the entire robotics + autonomy future hits a brick wall. Terafab isn’t optional; it’s the only way forward.
English
@pdicarlotrader "Both of my models are still bullish, but it has taken much longer than expected."
Welcome to $TSLA
Problem is: Elon is always building pyramids. Takes a long time to lay the foundation. At the start people come to watch the pyramid, they leave. Then suddenly: A Pyramid.
English
I said $TSLA would be at $550 by March 2026. I was wrong 🔻
Both of my models are still bullish, but it has taken much longer than expected.
Even with a solid system, you will be wrong at times.
In this video I break down the current state of $TSLA and what I expect from here.
English
By the way, Elons comment about cheeseburgers in the factory supports this direction, since impurities are the biggest obstacle to making Nano imprint work.
Elon could also plan to try and use a redundant design with real time Ai rerouting the process if an impurity damages a trace.
English
@pbeisel @CernBasher I am at the edge of my seat to learn how Elon plans to get around the bottlenecks, especially ASMLs EUV backlog.
If Tesla can make industrial scale 2nm NIL work it's a breakthrough far beyond vertical integration.
English
Terafab may be the most essential vertical integration Tesla has ever undertaken— and it is truly non-optional.
It will take years to build and will test even Elon’s speedrunning abilities to the limit, but that won’t stop him from trying.
The breakthrough likely lies in overhauling the overall facility’s cleanroom model. By moving wafers in sealed pods with localized micro-environments, the fab no longer needs a monolithic ultra-clean space. Elon’s line about “eating cheeseburgers and smoking cigars” on the fab floor isn’t silly, it’s the practical reality of a radically simpler, cheaper, faster approach that could finally change the economics of chipmaking.
This is all forced by the brutal “pinch” in chip supply. Tesla must produce on the order of 100–200 billion AI chips per year just to saturate its roadmap.
That volume powers:
FSD cars & Robotaxis (tens of millions of vehicles needing AI5 inference for near-perfect autonomy),
Physical Optimus (scaling from thousands today to millions per year, each requiring AI5/AI6-level compute),
Digital Optimus (the new xAI-Tesla software agents for digital/office automation, running massive inference clusters),
Space-based data centers (AI7/Dojo3 orbital compute for GW-scale training and inference beyond Earth limits).
AI5 delivers the ~10× leap for vehicles and early robots; AI6 shifts focus to Optimus + terrestrial DCs; AI7 goes orbital. No external foundry (TSMC, Samsung, etc.) can deliver that scale or timeline— hence the Terafab launch.
Without it, the entire robotics + autonomy future hits a brick wall. Terafab isn’t optional; it’s the only way forward.
Elon Musk@elonmusk
Terafab Project launches in 7 days
English
@DillonLoomis Nvidia will make so much money being the lifeline for not-dead-yet companies.
English
With Nvidia's announcement today about L4 robotaxis on Uber's network, it's critical to note they are not planning on removing the driver/supervisor until 2028. That's just the plan, there's a great chance that gets delayed
Relatively speaking, it's easy for companies to show demos and to get 98% of the way to autonomy quickly. What most will be stunned by is how hard the final 2% is and how long that will take
So do not let yourself be fooled by dozens of companies touting L4 and robotaxis and backing it up with buzz words and a demo drive that looks great. The gap between that and actually deploying an unsupervised robotaxi is unbelievably massive $TSLA
English
@EdLudlow DC is the DC segment but there is non-DC segment revenue. This is a subtle clue tht this is not total revenue guidance but only DC revenue guidance. Obviously, DC represents the vast majority and is typical of Jensen's low-key sandbaging.
English
@ItIsHoeMath Stop focusing on maybe-problems. It's not up to you to safe the world or even predict the future.
Read "Transurfing Reality" and act accordingly. I told you that 2 years ago and here you are, still freaking out.
English
Every day that goes by, it gets harder to focus on anything other than "solving the problem."
I just worked on my house and took a bath today. I didn't even try to work. I haven't published a real video in almost 2 months. I have literally over 300 great ideas for videos and I just cannot think lately because videos won't matter soon.
English
@LimitingThe A lot of consistency issues to be solved before that. A game is way more than pretty graphics. In fact the past years have shown that graphics are a trap.
English
How long before DLSS just creates the game from the ground up as you play it?
Autism Capital 🧩@AutismCapital
🚨 NEW: Video of Nvidia’s new DLSS 5 AI upscaling video games IN REAL TIME. Look at the difference. Awesome!
English
@Investanswers @HustleBitch_ It just goes to show that money doesn't make you better. It's just amplifies what is already there. And Bezos has always been a Nerd. This is just Math club on Yachts.
English
@HustleBitch_ He may be good at selling books, but he's not good at trading. I far prefer his first wife.
English
This is the face of a man who just realized he made the biggest mistake of his entire life.
English
@pbeisel @BenBajarin From Space Ai to Robots to self driving cars. It's one product after the other that basically says "buy this Nvidia product. It's your only hope to survive Musk disrupting your business model"
And there is no reason to mention Tesla because Tesla is not the target audience.
English
@pbeisel @BenBajarin You gotta understand why.
Nvidia is offering an umbrella to companies not wanting to be disrupted by Elon. Mentioning Tesla would have been an admission that the fox is already inside the henhouse and it's actually the farmers favorite pet.
English
Only two companies in tech are so deeply vertically integrated. Apple and NVIDIA. Jensen says they are the only one fully vertical and horizontally open.
🤔
English
@TheRedDragon @Micro2Macr0 It's not perfect yet, but it's amazing.
English
Gamers virtue signaling calling Nvidia DLSS 5 an AI slop filter
In no universe would anyone say this Bethesda slop looks better with the filter off.
You’re full of shit
If you don’t want it, don’t turn it on
As kids we always wanted realistic looking games. Now we have it
English