日本财务大臣片山皋月警告Mythos等先进AI构成重大威胁，TSLA涨3.89%科技板块分化

I own both, albeit older van and SUV (model y). I do like van a lot, and kids were probably more excited than when getting a Tesla. Agility is not comparable. For cargo, Tesla is surprisingly good. As we speak I'm planning to pick up a 2meter tall hot water cylinder (heat pump baby). I think telling people what they should do is a bad idea. On paper I should choose a wagon (I l…

📊 Group 3: Post-Market 1-Hour Charts Updated $TSLA is advancing from the April low in a strong impulsive structure, looking for more upside into the $453–480 zone before the next pullback shows up. For now, momentum stays constructive while that sequence remains intact. #Elliottwave

While I agree in principle, at least Tesla has delivered something pretty impressive. FSD is not really something I am personally interested in, but I cannot deny that it has gotten quite good. Better than any other system being offered, certainly. Waymo is better, but you cannot buy your own Waymo.

There were plenty of free email services before gmail. Google isn't at fault here because they provided a better experience.

@SK326 @Tradermikes40 $tsla take 5 seconds to use $GOOGL like cubie teaches

I know we're talking about Google here, but the privacy violations and concerns from this sort of search are massive. We need local AI ASAP.

I wish Debian had an Arch style bleeding edge fork. Till then I've been happy using Arch, I had my last straw when a program needed a more up to date GLIBC on Debian. That's such a can of worms to resolve, I just went ahead and gave Endeavour (Arch based) a try and havent gone back or changed distros ever since. If someone ever makes a Debian distro that is bleeding edge and su…

@johh22 semis are tough to chase up here. Doesn't mean you can't trade them on a smaller time frame but probably prudent to wait for pullbacks. I posted a setup on $NVDA at 200. The "edge AI" is interesting place to look for potential ideas imo.

> directly to PTX Weird. There's a recent NVIDIA MLIR that is quite good and fast. Or they could target the even easier and more recent/fashionable tile IR [1] used by CuTile [2] (a little bit higher level but significantly easier to target, only loses on epilogue fusion and similar). [1] https://docs.nvidia.com/cuda/tile-ir/ [2] https://developer.nvidia.com/cuda/tile

I took the time to write a meta thesis about this earlier using Golang as an example of a highly consistent langue which represents extremely well in the training data https://jry.io/writing/use-boring-languages-with-llms/

$META all the money is in ai bottleneck. Once it flips this rockets right to 800. I promise you all!!!!! They are so dialed in and built their own AI backbone. The capex looks bad now but just wait 2 years. They will get it all back and then some. I really need a pullback one more time. I already own 2k shares and looking to add another 2k.

I wrote about the meta thesis of programming languages in the training data here https://jry.io/writing/use-boring-languages-with-llms/