That slide on models on Gpus or mac silicon. Reminded me of the first time our Apple hardware dev team had to buy a windows machine for the rtx card to use nvidia omniverse software because Apple dont place nice with nvidia gpus these days. And have split with their silicon.
I wonder if apple will let nvidia external gpu play nice with apple hardware in the future. It may not make sense competitively. However, dev teams can also prefer apple toolchains but also want to co-dev with nvidia software. My guess is that we will see nvidia external gpus for mac silicon to reduce enterprise and creative dev team friction. Especially since nvidia and apple in alliance for openusd. They would want the software dev velocity not to be hindered.
I agree with Coatue's approach about focusing on the application layer. This isn't the same paradigm as focusing on the platforms like the last 15 years of B2B SaaS - so much about the models is open that there are foundational models evolving constantly.
I believe value is going to start accruing to the application of this technology to solve problems, rather than trying to capture a sliver of the value that the underlying LLM APIs provide.
I'm already building my product architecture in a manner that allows me to switch out the foundational model without too many changes. THAT'S A HUGE DEAL when it comes to evaluating where value is going to accrue in this AI phase in tech.
That slide on models on Gpus or mac silicon. Reminded me of the first time our Apple hardware dev team had to buy a windows machine for the rtx card to use nvidia omniverse software because Apple dont place nice with nvidia gpus these days. And have split with their silicon.
I wonder if apple will let nvidia external gpu play nice with apple hardware in the future. It may not make sense competitively. However, dev teams can also prefer apple toolchains but also want to co-dev with nvidia software. My guess is that we will see nvidia external gpus for mac silicon to reduce enterprise and creative dev team friction. Especially since nvidia and apple in alliance for openusd. They would want the software dev velocity not to be hindered.
Thanks as always Casey.
I agree with Coatue's approach about focusing on the application layer. This isn't the same paradigm as focusing on the platforms like the last 15 years of B2B SaaS - so much about the models is open that there are foundational models evolving constantly.
I believe value is going to start accruing to the application of this technology to solve problems, rather than trying to capture a sliver of the value that the underlying LLM APIs provide.
I'm already building my product architecture in a manner that allows me to switch out the foundational model without too many changes. THAT'S A HUGE DEAL when it comes to evaluating where value is going to accrue in this AI phase in tech.