AI Infrastructure's Next Bottleneck: Networking and Power Constraints
Analysts Meltem Demirors and Kelly Greer identify networking as the next major trade in AI infrastructure after memory, drawing parallels to NVIDIA's 2019 Mellanox acquisition. They also highlight power as the ultimate constraint, with gigawatts of backlog to connect to the grid.
RT @MTSlive: What are the physics constraints holding back AI infrastructure? @Melt_Dem and @kellyjgreer: "AI is basically just taking el…
What are the physics constraints holding back AI infrastructure? @Melt_Dem and @kellyjgreer: "AI is basically just taking electrons from the ground and running them through GPUs, you get tokens out of your computer." "Power is the ultimate constraint, gigawatts of backlog to connect to the grid.
What's the next big trade in AI infrastructure after memory? We asked @Melt_Dem and @kellyjgreer: "Memory is ripping now. We think the next narrative is gonna be around networking." "Jensen is doing the exact same thing now that he did with Mellanox in 2019 that turned NVIDIA from a GPU company i