The cluster section is the one that stands out for us.
Compute capacity only becomes useful when the rest of the system can keep up. That includes the data layer feeding the workload.
As compute becomes more distributed, keeping that data accessible across changing compute
We made an interactive explainer of the AI compute stack: from grid power and data centers to chips, racks, clusters, and workloads, plus the public companies at each layer.
Explore the infrastructure behind what @JensenHuang calls the five layer cake:
intelligence.mts.now/compute
Compute is increasingly dynamic - where it runs, what’s available, what it costs and what fits the workload.
But changing compute often means another cycle of moving, copying, staging or synchronizing data.
Shelby is built to change that - keeping data accessible as compute
We’ve been heads down hardening Shelby for Private Beta, and we’re beginning to onboard a select group of customers in stages.
Over the next few weeks, we’ll share more about what testnet clarified, how those learnings shaped the product, and the problem Shelby is built to solve
