Jacob Armstrong

Claude is totally useless at this point

everyone get so excited about neuralink but guess who had a buy it now button for years: intantech.com/pricing.html

@OverlyTrev @itskyleconner love how everyone forgets 30 min to 80% 2mw megachargers promised in 2017

@itskyleconner Nope. 1.2MW charge CAPABLE

Ummm Megacharger would indicate at least 1MW ….750kW 🤔 ☹️

@__tinygrad__ we have some work do to bring polysilicon refinement, wafer manufacturing, making dopants, photoresists, cell fabrication and all associated tooling, and final assembly on us soil then we have something.

If America doesn't get serious about changing this quickly, there's going to be no way for any US based AI providers to compete with China.

The price and difficulty of getting electricity in America is infuriating. Why aren't we building this? I nominate Arizona to be turned into a huge solar panel.

trying to one shot a xtensa lx7 emulator in ts 12 hours later....

claude vibecoded kicad schmatic 2/10.

@skibriirb7 @CyberMikeOG I am sure some things like the seat covers, foam, core structure, etc are purchased from suppliers but give tesla some credit, many autos outsource the full seat assembly to tier 1's

@CyberMikeOG Interesting (and a lie) because you can just look at used Tesla parts on eBay as see the mfgr. labels. They have outsourced from Preh, Adient, Faurecia, IGB (Lear) and others, just like everyone else. Many parts are made in Mexico, just like everyone else.

Did you know that Tesla designs and builds its own seats? Most automakers outsource the manufacturing of their seats. Buy American, Buy Tesla!

@yacineMTB this is what I thought of, disappointed it was a vac tube, good interaction bait

if you know what this is, dm me i will hire you

/compact has achieved long running agent capability

The only reason I have a @Starlink mini is for in flight internet. The roam plan being limited to 100mph now is total garbage. Now they want av use to pay $250 for 20 gigs. The enshitification of starlink has begun. It’s squeezing a extraordinarily small minority of customers.

@XueJia24682 This is very doable we have had 6.5c charge rates in some Chinese drone batteries. for years with field proven 1000+ cycle lifespan. dl.djicdn.com/downloads/t50_…

✨🇨🇳This is BYD's megawatt flash charging. It adds 2 km of range per second and can cover 400 km in just 5 minutes.

@nuro what happened to the pill car?

Another day, another autonomous drive on a random route around Mountain View, California. From quiet residential areas to busy main streets, our universal autonomy platform can handle all roads and all rides. #DrivenByNuro #SelfDriving #AutonomousVehicles #AI

@DeepValueBagger or a old mac studio

@DeepValueBagger m1 max 14/16 pro 64gb is the best deal

BIG 2026 PREDICTION: EDGE COMPUTING I went deep into the rabbit hole with OpenClaw and setting it up for my personal uses like finance, and health. It costed me nearly $200 for over 2 weeks use. The $80 claude max ran out in a few days, and tokens from openai ran out fast. I figured that xai has the lowest cost by 10x to 15x factor. That's when I digged into a bit more with local model. With my 12GB VRAM RTX, I could run a decent model suprisingly with newer backend like vllm serving openai compatible apis. I ordered by first Nvidia blackwell gb10 128MB VRAM to do more computation. At the pace I'm using with more frequent updates, the only economical way is local inference. Anyways my whole experience convinced me that in 2026 there will be a big shift into edge computing. Openclaw community have been buying up M3 Ultra 128GB $3500 all the way up to 512GB $10K unified RAM (eg it talks directly to the CPU). Apple is selling them like hotcakes. On the nonmac, Nvidia has been licensing to HP, Asus, and Dell. AMD also has a solution with 395+ CPU, and Intel will soon have one. Mark my words, every company will want their own rack server, and every developer will have their own local inference computer. The price reminds me of the the same cost of my first computer 33Mhz 4GB RAM computer. As cost decreases, it will be more accessible for each person. This will be a new growth segment for $APPL $NVDA and $DELL. All have shared very bullish guidances so far.

@niccruzpatane looks like Waymo and Hyundai are doing some real colab here, the rear cams are in the roof garnish. stuff like this means they are built on a special line just for waymo making the unit econ much better then the ipace

@pr0t0_01 hell no

Should I give my electronics engineer a raise?

@chatgpt21 It’s highly subsidized though.

Waymo & Uber are cooked 💀 Tesla robotaxi users are reporting autonomous rides in Austin for as low as 1.49$ for a half a mile ride and 5$ for a 30 minute ride. Humans can’t compete with this.

@AnthropicAI So if the dod asked lockheed to build a f35c that can backflip in a hover do we all get mad if they say no.

A statement from Anthropic CEO, Dario Amodei, on our discussions with the Department of War. anthropic.com/news/statement…

@gregjoz finally took yall like 2 years

Xcode 26.3 with Claude Agent & Codex hits the Mac App Store today! With advanced reasoning capabilities in Xcode, you can streamline workflows and build faster. And MCP support lets you easily connect other compatible agents.

Idea- airtag for vibe&therm&mag sensing on anything with a motor. This would be for predictive maintenance. Stuff like this kinda exists but it’s proprietary as hell. This should be $20 and be on every commercial air compressor cooling fans etc.