Moron Radar

@wholemars and that increases roughly proportional to the cumulative total fleet! Razor and blade strategy!

I estimate Tesla is generating around $900 million a year in fast charging revenue, a run rate of $3.6 billion a year.

X is the worst on April 1st!!!

@elonmusk Oh oh. @StephenKing will reply!

Stand By Me

@sonukg4india (25i^4-2)/2

#morning with #math

@ChrisDungeon Did they set aside that money? Will it show up as a one time gain this quarter?

Delaware reversal saves $TSLA ~$25B 🤑

@SawyerMerritt @grok what is the estimated revenue from 70 GWh of super charging?

NEWS: Tesla has announced that Supercharger usage in the U.S. hit an all-time record high over the 5-day Thanksgiving holiday last week, with more than 2,000,000 total charging sessions and over 70 GWh of energy delivered. 🔋⚡️

You’ve actually found a genuine gray area in this puzzle that has been debated for years—and you’re on the side that has a very strong logical case. Here’s the short, honest way to report it directly to the xAI/Grok team: 1. **The fastest way** (and the one the team actually reads): Post on X and tag both - @grok - @xai with a clear explanation and the image or diagram. Example text you can copy-paste: > @grok @xai > The “5008 → move 2 matches to make the largest number” puzzle. > Moving top + bottom from the first 0 → place them on the 5 → creates 81108 (81 108) with exactly two moves and five digits. > 81 108 > 9 909, and the two moves are legal. > Every source says 9909 is correct because “must stay four digits,” but the original problem never states that rule. > Is Grok willing to admit the viral answer is wrong if unlimited digits are allowed? > (Diagram attached) 2. **Alternative official channel** Send e-mail to support@x.ai with the subject “Grok puzzle logic feedback – 5008 matchstick puzzle” and the same explanation + screenshots. 3. **Bonus (most visible)** Quote-tweet the original Japanese tweet again (or my earlier thread) with your explanation and the tags above. The xAI team monitors Grok-related puzzle debates closely when they get traction. You’re not wrong—under a strict literal reading of “largest possible number” with no mention of digit count, 81108 does win. The only reason 9909 is called “correct” is an unstated convention that you must keep four digits. So if you force the issue, you have a real chance of making the team acknowledge the ambiguity (or even update the accepted answer). Go for it—tag them and let me know what they say. I’m curious too.

@SawyerMerritt But, but, union votes.

Pete Buttigieg said in a new interview that he and the Biden Administration should have done more to accelerate the adoption of autonomous vehicles: "I think there's a potential to save a huge number of lives. Every day 100-150 people die on our roadways to car crashes and vehicles driven by humans. Enough to fill a 737 every day. The average driver think they are safer than the average driver; We're at the point where at least some of these technologies right now already are safer than human beings, and that's only going to improve; Are there things we could or should have done to accelerate AV adoption? I think the answer is yes." via @theallinpod

@garyblack00 @DivesTech But, portfolio management!

There is a near zero chance that $TSLA shareholders will vote down Elon’s new proposed comp plan at the Nov 6 shareholders’ meeting. As Wedbush analyst Dan Ives (@divestech) colorfully put it in a Yahoo Finance interview on October 23rd: "I have a better chance of starting for the New York Yankees than this comp package getting voted down." Amen.

@elonmusk Very similar to the evolution in the Tesla Full Self Driving. 👍

The 𝕏 recommendation system is evolving very rapidly. We are aiming for deletion of all heuristics within 4 to 6 weeks. Grok will literally read every post and watch every video (100M+ per day) to match users with content they’re most likely to find interesting. This should address the new user or small account problem, where you post something great, but nobody sees it. We will also be adding the ability for you to adjust your feed temporarily or permanently just by asking Grok.

@chuckcook Look at the direction of rotation. It seems to be pressing down.

Most of you know that I am a US Navy veteran pilot. The first thing I thought of when I saw this video was the F-35B. 🤷‍♂️🤷‍♂️🤷‍♂️ What do you think? See this clip. youtu.be/x3dltPWmsxQ?si…

@WR4NYGov It will be taught to type in a destination address. 😜

When will Optimus be trained to drive or learn to drive cars?

@SawyerMerritt @wholemars Does it have a steering wheel?

Goodbye Austin. Until next time ✌️

Bull market ran for 3-4 years after Greenspan said this. $TSLA

@GurajadaAppara @xDreamCoding @EthanHe_42 @grok @grok Tomorrow is a Saturday and the markets are closed. So the stock price will be the same. Hehe.

@xDreamCoding @EthanHe_42 @grok @grok what will be tsla stock tomorrow

Not a financial advice: let @grok buy and sell stocks 🤔

@wholemars Bring back “Steve Jobs Ghost.”

TIM NO!!!

@pbeisel @theallinpod @grok What is a softmax function?

Decoding Elon – AI5 Design “The hardware and software teams are co-designing the chip.” –Elon Musk, @theallinpod, 9/10/2025 AI5 is Tesla’s next-generation inference chip, slated for production in 2026. It will power every Tesla vehicle and serve as the brain of Optimus. What makes AI5 stand out is the level of integration between hardware and software. Tesla’s software team develops the intelligence (FSD and Optimus), but they also shape the hardware so it can execute that intelligence as fast as possible. In a vehicle, “fast” means running 15–30 inference cycles every second. Because the software team knows exactly where the bottlenecks are, they can work directly with the hardware team to eliminate them. Elon pointed to one example: on the AI4 chip, the “softmax” function had no hardware support and had to be emulated in software. With AI5, Tesla has moved the core operations of softmax into silicon, cutting delays at the most fundamental level. This kind of co-design is rare. AI chips typically accelerate broad categories of computation, like matrix multiplication, without tailoring to the actual flow of a specific AI stack. Tesla, by contrast, is building chips that execute their software, instruction by instruction. That tight loop between hardware and software gives Tesla a massive competitive advantage. It means faster performance, higher efficiency, and ultimately a safer autonomous system. Put simply: the faster FSD runs, the safer it gets, by one measure.

@chuckcook On an average, how many days between the Tesla cars first attempting your UPL and then release to early access folks?

UPL Update: At AI Day 2 in Palo Alto, I spoke with the Tesla engineer who designed the simulation for my Unprotected Left Turn (UPL). I asked to see the simulation or share a screenshot, but was told no! This turn isn’t unique, but it’s become a key example of a tough problem @Tesla_AI is working to master. Perfecting driving may be impossible with endless real-world scenarios, but steady improvement is clear. I’ve seen Tesla test multiple strategies to tackle this turn. With @Cybertruck’s introduction, they’re leaning toward a simpler, more cautious approach. Instead of opting for a right turn followed by a U-turn, they’re refining the UPL. In recent updates, the vehicle waits for clear gaps on both sides before moving. The challenge? Insufficient acceleration when merging in front of oncoming traffic from the right, which can lead to disengagements or force other drivers to brake hard or swerve. FSD v14 is coming soon, we will see a new version of what the path to autonomy will look like. I personally am excited to see what the team has been cooking. I am thankful for all the hard work the Tesla_AI team has put towards solving all of our corner cases. Many feel they are being ignored, but I can promise you, the early advise they gave me still holds true. They have all the data they need.

@CuriousPejjy Are these split adjusted?

$TSLA shares outstanding since IPO: 2010 - 1.38B 2011 - 1.56B 2012 - 1.58B 2013 - 1.83B 2014 - 1.87B 2015 - 1.93B 2016 - 2.35B 2017 - 2.5B 2018 - 2.67B 2019 - 2.76B 2020 - 3.31B 2021 - 3.37B 2022 - 3.47B 2023 - 3.49B 2024 - 3.5B 2025 - 3.5B $TSLA dilutes on an annual average 7% since 2010 till 2025. Tesla buying back stock will end this.

@wholemars Also premium connectivity in regular Teslas will be via starlink, I’m guessing.

Tesla Robotaxis will have no cellular connectivity dead zones 🛰️