$1,000 follows their calls into $1,064
All in the world market (VT). No called asset is counting right now.
himanshu@himanshustwtsWhat a week for NVIDIA with Poolside and Hugging face! They are fucking everywhere across the AI stack (from chips to models to products), but use licenses + investments + talent deals for expensive model/chip startups. The AI Superpower!!
himanshu@himanshustwtsLlama 3 moment for Meta. They are back in the game!
Mark Zuckerberg@finkdToday we're also opening the weights for Muse Glimmer, a great 30B parameter dense model that can run locally. Soon we'll also release the weights for Muse Spark 1.2, our latest foundation model. Meta is a strong supporter of open source and I'm proud of these releases. Congrats
himanshu@himanshustwtsactually meta has the trio in place - (data, talent, compute). i think the most underrated fact here is how meta has started preserving the good talent now (having good talent to stay in org for long time > frequently hiring and firing good talent). also, they have the
kipply@kipperriimeta is so clearly the third best lab that i'd be surprised if they lost their place in the next year. i didn't see it coming because i lost faith in "money can be exchanged for goods and services" but buying the compute and talent worked. i must never lose faith again
himanshu@himanshustwtsNVIDIA is on a roll. After Poolside, they are going for Perplexity in very same way. > Pay billions for a exclusive / non-exclusive tech license > Hire 100s of staff and leave the company independent. > Now Nvidia has specialized expertise for its efforts in AI without going in
Valida Pau@validapauNEWS: Nvidia is investing in Perplexity. BUT it considered paying Perplexity billions of dollars to license some of the startup’s technology and hire some of its staff. Also Perplexity's revenue 📈📈 @_pheebini @amir h
himanshu@himanshustwtsMETA IS SO BACK! This costs like $3 per 1000 voice minutes. Zuck is making AI an extreme commodity if you just think about it.
Mark Zuckerberg@finkdMuse Voice Transcribe is MSL's first real-time audio perception model -- rolling out today. SOTA in streaming speech-to-text, it handles speaker diarization, and endpointing natively in a single model.
3:03
