Transcript
0:02 The AI networking model it's a networking industry model that tracks and covers networking custom bomb as well as of networking custom shipment units. We think about say, how would meta deploy its networking clusters? How would that be different at Aldo? How would that be different at Nubia? How would that be different at Crystal? And once we do a bottoms up build of how these networking clusters are being designed, we get a pretty decent view on like what networking on equipment 10 is, what the bomb for every cluster is.
0:34 say for example, if you look at like Amazon, they would do like three to four layers of switches. But say if you go to like a new cloud, they could do two layers of Nvidia switches on Spectrum-X. Our team is really global. We are pretty small in terms of the networking model core team. Me, I am based in SF. We have an analyst based in Europe. We have an analyst based in Singapore. We have like all the AI software engineers and then AI hardware engineers who help to triangulate our insight. One is pure supply chain information. We travel to China. We travel to Taiwan. We talk to engineers.
1:08 The second source of information is like the demand side. We basically hear from like new clouds. How they thinking about deploying and designing their clusters. And then I would say the third source of information is just talking to a lot of these equipment vendors. I would say our biggest edge is like triangulating what the data points we get from supply chain with our technical knowledge we have on like how network clusters are being designed. Investor clients are very interested in what these technologies mean for the stocks that they cover. If they're thinking about modeling the transition towards CPO, what would that mean for say the transceiver market? What would it that mean for say the laser market?
1:47 Industry clients care more about their competitors, what their competitors roadmap would look like, and then they're also very interested in the very long term view of how the industry could look like say in the 2028 to 2030 timeline. I would say that's where the most amount of uncertainty, but that's also where they see the most amount of opportunity.