@mattparlmer Demand for frontier intelligence is extremely inelastic -- particular where the roi on intelligence is high (e.g. markets).
If you truly believe this, then labs are horrible businesses.
People will trade orders of magnitudes more dollars across a much wider set of products.
As more value accrues to private companies, its only right that traders get access to the same upside that insiders get.
Hats of to what is being built @SynapseProtocol
The only lesson from Karpathy moving to Anthropic is that the singular most determant factor for technical hiring is working on uniquely difficult technical problems with the best resources to do so.
Effective supply also expanded drastically because of multiplexing and packet switching , algo improvements could seemingly do the same for frontier models. Does this bother you/factor into whats worth paying attention to?
wasn't alive at the time so curious to know your perspective on the analogy.