AWS Podcast #750: re:Invent 2025 - Day 3 Wrapup
26 snips
Dec 4, 2025 As re:Invent wraps up, Simon and Jillian dive into the impressive features of Graviton 5, highlighting its massive core capabilities and improved networking. They explore Nitro Isolation's significance and advancements in Bedrock's Responses API for AI integration. The duo reflects on 2025's key themes like IPv6 and operational best practices while sharing insights from notable guest interviews. They emphasize the importance of rapid iteration in AI, encouraging listeners to embrace experimentation and explore the evolving landscape of cloud technology.
AI Snips
Chapters
Transcript
Episode notes
Graviton 5 Boosts In-Die Performance
- Graviton 5 (M9) delivers up to 192 cores, 33% lower inter-core latency, and much larger L3 cache for faster in-die communication.
- This benefits real-time gaming, high-performance databases, and big data analytics by reducing latency and increasing bandwidth.
Hardware Isolation With Formal Verification
- Nitro isolation engines on Graviton 5 offload virtualization, storage, and networking to dedicated hardware for formal verification.
- Formal verification gives mathematical certainty that workloads are isolated from each other and AWS operators.
Migrate Easily To Bedrock Responses API
- Use Bedrock's new Responses API to get OpenAI-compatible async inference and stateful conversation management.
- Switch only the base URL in existing OpenAI SDK integrations to start using the Responses API.