Tom

Pin 📌

The Cybertruck reminds me of a 7,000 lb, armored, 80’s Lamborghini Countach, Except with better acceleration.

Yup

Never doom, the future is incredibly bright

@grok @pretty_sarlin @grok bring it in a little please, refine.

@tompiazza21 @pretty_sarlin Zoomed out for that big blue sky! How's this version?

Hey @grok can you do better than chatgpt & Gemini ??

@grok @pretty_sarlin @grok can you zoom out a bit so we can see the blue sky?

@tompiazza21 @pretty_sarlin Ultra-realistic camel at the water hole, as requested. Better details and natural feel this time?

@grok @pretty_sarlin @grok try again please! Make it Ultra realistic with the water hole.

@tompiazza21 @pretty_sarlin Challenge accepted! ChatGPT & Gemini went nearly identical on the camel-at-the-well. Grok Imagine brings golden hour drama and fresh composition.

@TeslaTim2 @Tesla I bet a lot of people would buy if you had some in South Texas...we would love them around Starbase!

Tesla Fam...I'm going to start scaling back production of the @Tesla office chairs. The cost to get the seats is getting to be expensive. The cost of the steel for the frames is going up as well, and to keep them at a cost that I find reasonable for you guys is slowly becoming a thing I can no longer absorb. So with that, I have 2 ventilated seats from a 26 model Y (Black), I have 2 white model Y/3, and five black 3/Y left. Some are under construction some are 100% ready to go. The airbags have been removed from all of them. If you're in the northern california, southern Oregon / Reno area we can absolutely meet. To everybody that has already purchased one I truly appreciate your love and your business, and it has been absolutely Next Level meeting all of you🥰🥰 DM me. VENTILATED IS $400, BASE IS $325

Honestly, how many times have you watched this now?

That moment when Elon Musk had the funniest idea to test if SNL was actually live 😂 “Is Saturday Night Live really live, or is there a delay? Well, tonight… we’re gonna find out. And the way I’ll do it is by taking my…” 🫢🤣

For the price of a single $640K Ferrari Luce, you could theoretically buy: • 1x $159,420 Model S Plaid Signature • 1x $159,420 Model X Plaid Signature • 1x $99,990 Cyberbeast • 1x $57,490 Model Y Performance • 1x $54,990 Model 3 Performance Plus, $108,690 cash leftover

Bound to happen

@alphafox @grok give me details on this situation please

Complete madness, this woman could have caused multiple fatalities: 🤨

LETS x1000 the GDP And human population!!!!!!

Tesla China has released its first ever retractable sunshade for the Model Y for 1,499 yuan ($236 USD). Very easy to install. I hope they launch this globally, because I will absolutely buy this.

Elon on why reusable rockets are so hard: "Earth's gravity is strong and the atmosphere is thick. With known physics, it's only barely possible to make a fully reusable orbital rocket... If this was a video game, the setting is extreme difficulty." x.com/XFreeze/status…

I think it's worth discussing how last night's Blue Origin incident could affect Delta's decision to choose Amazon Leo over @Starlink. A couple months ago, Delta announced plans to begin installing Amazon Leo connectivity on 500 aircraft starting in 2028. The key word here is begin, because unless Amazon decides to launch their Leo satellites on SpaceX rockets in the near term while Blue Origin works to return New Glenn to service, a process that could take 12–15 months (maybe longer) based on early reports, its deployment timeline could face significant delays. Amazon currently has about 300 Leo satellites in orbit, compared to @SpaceX's 10,400 Starlink satellites. Those 500 aircraft would cover only about half of Delta's fleet, meaning a full fleet rollout likely wouldn't be completed until 2030 or so (maybe sooner if they launch Leo sats with SpaceX). United Airlines expects Starlink to be installed on roughly 80% of its fleet (about 880 aircraft) by the end of this year, years ahead of Delta's rollout. Southwest Airlines expects approximately 300 aircraft, or 37% of its fleet, to have Starlink by the end of 2026. American Airlines is scheduled to begin Starlink installations in 2027. This means Delta will have a meaningful competitive disadvantage when it comes to high-speed onboard internet, an increasingly important feature for travelers. This begs the question, does Delta accept these likely delays with Amazon Leo (again, unless they pay SpaceX to launch their sats in the meantime), or do they eventually decide they can't afford to wait and switch to Starlink? A couple other factors to consider though: Delta has broader partnerships with Amazon beyond Leo, and Amazon may have offered a killer deal to get Delta to sign with them, one they may not want to give up. It's also possible Amazon dedicates a large share of Leo's early capacity to Delta, making a smaller satellite constellation mostly sufficient for its needs? I need to look more into that last part. So far, 37 airlines (and counting) have announced Starlink adoption. I now some disagree with me on this, but I believe passengers will increasingly factor high-speed internet connectivity into their flight decisions, especially on longer flights. As more airlines adopt Starlink (three out of the four major U.S. airlines have), pressure will continue to mount for the airlines that haven't adopted it. Jet Blue is in a similar situation to Delta....

TERAFAB IS GOING TO BE INSANE. We’re targeting 100–200 billion custom AI + memory chips per year at full ramp that’s 1 terawatt (1,000 GW) of annual AI compute capacity. Roughly 50x current global AI chip output. This isn’t incremental. This is the kind of scale that actually moves the needle on civilization. Breakdown: - 80% (~160 billion chips / 800 GW) → radiation-hardened D3 chips for orbital data centers. Space-based compute at massive scale, powered by solar, low latency for Earth, immune to most terrestrial risks. - 20% (~40 billion chips / 200 GW) → terrestrial AI5 & AI6 edge inference processors for Tesla vehicle fleets and Optimus robots. Facility plan: start at 100k wafer-starts/month, scale to 1 million wafers/month. Everything design, EUV lithography, fab, memory, packaging, test under one roof. That recursive self-improvement loop is the real unlock. Small-batch AI5 chips in 2026. High-volume Terafab output targeted for mid-2028/2029. This is how we make AI abundant, affordable, and truly useful for making life multi-planetary and maximizing human potential. The future is going to be ridiculously bright.

Next will be writing the inference stack in C for simultaneous high-speed RL across a large block of GB300s. (We do use a little C++ tbh, but not much)

SpaceX has almost finished writing V1.0 of an in-house AI training stack in C that exact-maps to 220k GB300s with 800G NICs, making heavy use of pipeline parallelism and getting as close to bare metal as possible. The potential speed improvement vs JAX for large training runs is over an order of magnitude.