The Shape of Inference
We’re hosting a small omakase dinner for people building, investing in, or simply curious about the next chapter of AI inference.
As workloads become more varied, inference is becoming less one-size-fits-all. Over dinner, we’ll trade notes on questions like:
how different workload shapes change what “good” infrastructure looks like,
where memory constraints start to matter,
how people are thinking about multi-silicon systems,
and what’s working in practice versus what still feels unnecessarily hard.
This won’t be a formal technical session. The idea is to bring together thoughtful people across AI, infrastructure, and startups for good food, honest conversation, and a few unexpected connections.
Come for the omakase, stay for the inference takes.
Hosted by MoE Capital and Emergence.