$1,000 follows their calls into $976
Piotr Mazurek@tugot17Nvidia moving beyond the cuda-moat to things that are hard to replicate through “Claude make no mistakes“
SemiAnalysis@SemiAnalysis_NVIDIA LPU supports 3 types of disaggregated inferencing: 1. Rubin Prefill + LPU Decode for the fastest interactivity 2. Rubin Prefill + Rubin Decode Attention + LPU Decode FFN for the middle of the curve 3. Rubin Prefill + Rubin Decode Verification + LPU Drafter for the
Piotr Mazurek@tugot17the bull case for AMD is that it builds this insanely good hardware; just the software is not 100% there; once this is bridged through "make no mistakes", there are trillions of shareholder value to be unlocked here
Piotr Mazurek@tugot17Excited for the future of Apple, hopefully less stock buybacks and more folding screens and 2nm semiconductors
Om@Om_Codes_- Meet John Ternus - CEO of Apple - competitive swimmer - No MBA - Studied mechanical engineering - had a minor in Psychology, - Joined Apple as a engineer - Kept climbing, one product line at a time, for over a decade - Became VP of Hardware Engineering - Promoted to Senior VP -
Piotr Mazurek@tugot17been talking about this with a German friend last night; imagine being @Siemens_Energy, arguably one of dozen or so European companies 🇪🇺 that will thrive post AGI, and instead of offering "stock that will keep you out of permanent underclass" they offer "subsidies for the h
