The cost to tokenize seems inevitably to go in the opposite direction, even though deep seek has come out, saying in your article that they might increase prices, and I mean that collectively overall, for the long-term of how LLMs are utilized inside the complete framework of AI. Since we’re still in the buildout phase, things like harnesses and plug-ins and connectors seem to really ground and make LLMs actually usable versus them alone now these big companies like Anthropic and OpenAI are obviously racing to fill in the gaps. The argument for more specialized, smaller labs providing very specific LLMs for use cases I think becomes a bigger story and whether the Hardware or software becomes more intuitive towards AI and thus cheaper to run as far as compute goes.
Good framing, and I think the FCF point you land on at the end is the real thing to watch, not token demand. Token growth stays spectacular regardless, so it does not actually tell you where the loop gets tested.
The theory seems sensible, but I believe that you are not considering the costs of data centers which are running on heavy subsidy and once that is removed, they will incur a much higher cost in running those mega centers. They also might be benefitting from taxation perspective which can change if the government starts taxing them differently.
Can you reference these subsidies specifically? I understand any large corporation that’s investing millions upon millions and billions will get consideration from local state and federal. However there is a clear need infrastructure wise for buildout and historically any large Industry has seen proportionate subsidies agree it’s definitely something to keep in scope.
THANKS CRACK
The cost to tokenize seems inevitably to go in the opposite direction, even though deep seek has come out, saying in your article that they might increase prices, and I mean that collectively overall, for the long-term of how LLMs are utilized inside the complete framework of AI. Since we’re still in the buildout phase, things like harnesses and plug-ins and connectors seem to really ground and make LLMs actually usable versus them alone now these big companies like Anthropic and OpenAI are obviously racing to fill in the gaps. The argument for more specialized, smaller labs providing very specific LLMs for use cases I think becomes a bigger story and whether the Hardware or software becomes more intuitive towards AI and thus cheaper to run as far as compute goes.
Good framing, and I think the FCF point you land on at the end is the real thing to watch, not token demand. Token growth stays spectacular regardless, so it does not actually tell you where the loop gets tested.
The theory seems sensible, but I believe that you are not considering the costs of data centers which are running on heavy subsidy and once that is removed, they will incur a much higher cost in running those mega centers. They also might be benefitting from taxation perspective which can change if the government starts taxing them differently.
Can you reference these subsidies specifically? I understand any large corporation that’s investing millions upon millions and billions will get consideration from local state and federal. However there is a clear need infrastructure wise for buildout and historically any large Industry has seen proportionate subsidies agree it’s definitely something to keep in scope.
Wow, great article! And great mental model, a new one for me