Danila Shtan

What Danila Shtan said

1. Unsupervised Learning · Nov 2, 2026

   The Window Before AWS Wakes Up: Nebius CTO on Racing the Hyperscalers

   8 moments · 28 min

   4:30
   1. 39:13

      Chip scarcity gave AI clouds room to compete ProPro pick

      “this is a massive investment on their part that needs to be justified by the size of a potential business. And it's not that there yet.”

   2. 25:05

      Open-model inference still has major room to improve ProPro pick

      “95% of our customers do not run vanilla open source models. These are modified, fine-tuned”

   3. 53:19

      Nebius lets engineers use AI but keeps accountability human

      “whatever you've done with the agent, it's still your result. It's your responsibility. It's your pull request. It's your code.”

   4. 15:36

      AI customers need shared storage built for extreme throughput

      “five, six, seven hundred gigabytes per second of throughput.”

   5. 10:12

      New Nvidia chips require constant iteration before launch

      “they are constantly iterating while these chips are being produced.”

   6. 21:47

      Prefill/decode disaggregation remains a case-by-case tool

      “we are in the stage where everyone's talking about that, and no one's doing that properly.”

   7. 34:25

      Execution capacity is the bottleneck across AI infrastructure

      “an experienced project manager is probably in highest demand I've ever seen throughout my career.”

   8. 57:53

      Agent reliability may matter more than model size

      “I think a lot of future AI adoption lies in how reliable your agentic scenarios are and how reliably you reach those outcomes.”