The tweet archive.

“@alex_avoigt Almost all of Tesla’s value long-term will be from AI & robots, both vehicle & humanoid”

“@WholeMarsBlog @SawyerMerritt @Tesla He doesn’t get enough negative feedback. This platform is great for that :).”

“@herbertong Gates also said the Tesla Semi was impossible, even though it was literally being driven all over the country. When I asked what battery Wh/kg and truck Wh/mile he was using to reach the conclusion that it’s impossible, Gates had no idea, but still stuck with his conclusion. 🤷‍♂️”

“@teslaownersSV No, but Robotaxi will look like the future”

“@teslaownersSV Likely to be 10% to 20% longer in later versions”

“@teslaownersSV No”

“@teslaownersSV Yup”

“@Tesla_Asia So many cool places to visit in China!”

“@SawyerMerritt @Tesla PW3 is optimized for ease of installation & high power, which means that a single Powerwall can serve as an uninterruptible power supply for most homes. This is a big deal for ensuring that the lights stay on and you can power all your devices in the event of a power outage.”

“@SawyerMerritt @Tesla Cool”

“@farzyness @WalterIsaacson Managing that data is extremely difficult. We ignore (don’t store at all) or delete almost all of it. The training videos that matter are rare ones, like weird & busy intersections, so less than 1% of all videos. Tesla is limited by AI training compute. Nvidia + Dojo fix it.”

“@Tesla_Asia Congratulations!”

“@teslaownersSV Karma is real”

“@TeslaSynopsis @Kristennetten At Masada”

“@teslaownersSV Hope in, we’re goin to Mars!”

“@teslaownersSV You’re welcome”

“@teslaownersSV @DrKnowItAll16 @alx Advertisers avoid controversy, so all that is needed for ADL to crush our US & European ad revenue is to make unfounded accusations. They have much less power in Asia, so our ad revenue there is still strong. This “controversy” causes advertisers to “pause”, but that pause is…”

“@teslaownersSV @imPenny2x @ADL If this continues, we will have no choice but to file a defamation suit against, ironically, the “Anti-Defamation” League. If they lose the defamation suit, we will insist that they drop the the “anti” part of their name, since obviously … 😂”

“@teslaownersSV Yes”

“@SawyerMerritt @Tesla Yup”

“@TeslaSynopsis @WalterIsaacson Very little”

“@ChrisJBakke @XHiring @Tesla_AI @coinbase @ycombinator @axios Cool”

“@SawyerMerritt @Tesla Accurate”

“@mezaoptimizer Tesla AI can see incredibly well”

“@teslaownersSV Not at first, but we will add the ability to turn encryption on or off dynamically. There is necessarily a slight lag for encryption. Most of the time, encryption isn’t important and quality of call is better.”

“@Teslaconomics True”

“@jamesdouma @farzyness Training compute should soon not be so much of a limiting factor. Very difficult bringing the 10k H100 cluster online btw. Similar experience to bringing our now 16k A100 cluster online. Uptime & performance are low at first, then improve with lots of work by Tesla & Nvidia.”

“Total game changer. If you haven’t tried Tesla Autopilot, you don’t know how awesome it is.”

“@ray4tesla V12 is still training”

“@thejefflutz @Tesla @Tesla_AI Yup”

“@ray4tesla Did my best to warn people that LiDAR isn’t optimal for cars. Roads are designed for biological neural nets & eyes, so digital neural nets & cameras will work best. I say this even though Dragon uses LiDAR for docking with Station, so I don’t intrinsically hate that technology.”

“@ray4tesla Some of them are indeed doing so – we’re well aware. Part of the value of having our own hardware is that our software will only run on that specific hardware, warts & all. As with Apple, you can’t just copy the code or unpack the binaries. You need their hardware too.”

“@DirtyTesLa It’s a real 6 months. Maybe less.”

“@farzyness Having a vast fleet of cars for FSD Autopilot training is a massive Tesla advantage. Essential for the hundreds of millions of unusual situations cars encounter on roads every year.”

“@WholeMarsBlog HW4 software will lag HW3 by at least another six months, as our focus needs to be on getting FSD on HW3 working super well and provided internationally.”

“@WholeMarsBlog Designing in the cameras, liquid-cooled Tesla computer (always go liquid-cooled!) and high security internet gateway will probably take major car companies about three years. Might as well start now! We want to be helpful, as we were with making our patents available for free,…”

“@aelluswamy True, it was super hard to do this! Even when FSD is 2X as safe as the average human driver fully unsupervised in all scenarios, then we will want to make it 3X, 4X … 10X. About a million people die every year in automotive accidents and ~10 million are seriously injured.”

“@thejefflutz @CNBC @business @YahooFinance @Tesla They don’t get it”

“@Teslaconomics He’s such a jolly little fellow”

“@Teslaconomics It is very easy! 11.4.7, which is mostly, but not entirely, AI is with Tesla employees and a few external users. Rolls out to anyone who bought FSD in the North America next week.”

“@Scobleizer An accurate assessment. What is also mindblowing is that the inference compute power needed for 8 cameras running at 36FPS is only about 100W on the Tesla-designed AI computer. This puny amount of power is enough to achieve superhuman driving! It makes a big difference that we…”

“@teslaownersSV To inform & entertain is our goal!”

“@DirtyTesLa Maybe I can squeeze it in tomorrow night”

“@teslaownersSV Yes!”

“@Teslaconomics It is in the set of possible outcome. I see a path.”

“@teslaownersSV It’s true”

“@astro_greek I would just move my desk around the factory to wherever the biggest problem was. These days, I don’t have a desk, but the room I use for meetings at HQ overlooks the Model Y end of line.”

“@Teslamojo @PicturesFoIder 😂”

“Just drove the production candidate Cybertruck at Tesla Giga Texas! https://t.co/S0kCyGUBFD”

“@teslaownersSV Tesla Model S Plaid will do 0 to 60mph (~100km/h) in 2 seconds”