> Markdown version of [/videos/2133-building-the-nervous-system-of-ai-michael-kagan-nvidia?t=1166](https://www.wearedevelopers.com/videos/2133-building-the-nervous-system-of-ai-michael-kagan-nvidia?t=1166). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Building the Nervous System of AI - Michael Kagan (NVIDIA)

Michael Kagan asserts that without robust networking, AI hardware is just expensive sand. Discover how distributed nervous systems transform scattered data centers into singular intelligence factories.

- **Speakers:** [Michael Kagan](https://www.wearedevelopers.com/@michael-kagan), [Sead Ahmetović](https://www.wearedevelopers.com/@sead-ahmetovic-2)
- **Event:** World Congress 2026 Europe
- **Published:** July 10, 2026
- **Duration:** 29:17
- **URL:** https://www.wearedevelopers.com/videos/2133-building-the-nervous-system-of-ai-michael-kagan-nvidia

## Summary

The computing landscape has undergone a profound shift, transitioning from the PC revolution and utility-based cloud computing to the era of AI infrastructure, where natural language democratizes programming. Traditional data centers function as data warehouses, but modern AI factories operate as power plants that ingest raw data and energy to produce intelligence in the form of tokens. As AI models demand exponentially more compute power, the industry has bypassed the limits of physical atomic constraints through extreme hardware co-design, integrating thousands of processing units into massive compute clusters.

The acquisition of Mellanox positioned networking as the critical nervous system bridging these vast compute resources. Because constraints like land availability and power transmission create bottlenecks for single-site scaling, advanced networking architectures rely on scale-up configurations connecting dozens of chips and scale-out infrastructures handling millions of nodes. Technologies like Spectrum XGX allow data centers located hundreds of miles apart to function synchronously as a single, disaggregated machine. This architecture effectively treats advanced models as distributed energy batteries, where model training and daily inference can occur in entirely different geographic locations.

Without robust software layers translating raw power into functional infrastructure, hardware remains functionally equivalent to expensive sand. As intelligence expands beyond digital reasoning agents into physical applications, autonomous robotics will rely increasingly on edge computing, updating their local models iteratively by syncing with centralized data hubs—much like humans exchanging knowledge at a conference. Looking toward future innovations, mastering foundational disciplines like math and physics remains essential, with digital biology emerging as the next critical frontier. Ultimately, massive simulations run on these integrated systems hold the potential to transform history into an experimental science, allowing humanity to accurately forecast the physical impacts of complex decisions.

**Keywords:** nvidia AI infrastructure, AI factory architecture, GPU scale-out networking, nvlink scale-up clustering, spectrum XGX datacenter synchronization, mellanox networking hardware, data center energy constraints, CUDA software platform, agentic flow workloads, physical AI integration, autonomous edge robotics, token production optimization, digital biology research, extreme hardware co-design, distributed inference clusters

## Chapters

1. **Evolution of computing to AI infrastructure** (01:11) — How AI transforms the unit of computing and makes programming accessible through natural language.
1. **Acquiring Mellanox to build cohesive AI factories** (04:05) — The strategic necessity of advanced networking to connect GPUs into a single unified AI computer.
1. **Transitioning from data centers to AI factories** (07:10) — Converting traditional storage warehouses into power plants that generate intelligence tokens from raw data.
1. **Optimizing infrastructure for agentic flows and inference** (09:23) — Managing exponential computing demands driven by reasoning models and autonomous agentic interactions within data centers.
1. **Scaling up and scaling out GPU clusters** (12:17) — Using NVLink and scale-out networking to connect millions of GPUs into massive unified supercomputers.
1. **Connecting distributed AI factories across distances** (15:17) — Overcoming power limits by networking separate data centers to operate as one synchronized computing machine.
1. **Architecting CUDA and the AI software stack** (16:52) — Building a flexible API platform that allows developers to innovate continuously on top of hardware layers.
1. **Navigating power limitations for AI infrastructure** (19:26) — Treating energy supply as a foundational layer and decoupling training locations from edge inference endpoints.
1. **Moving agentic intelligence to physical robotics** (22:10) — Enabling robots to operate independently in the physical world and sync learned data back to centralized clusters.
1. **Ensuring safety and security in AI systems** (24:13) — Implementing technical guardrails and cooperating with authorities to steer artificial intelligence toward benefiting humanity.
1. **Choosing foundational learning paths for engineers** (25:42) — Why mastering core physics, math, and emerging fields like digital biology remains critical amidst AI advancements.
1. **Simulating historical outcomes as an experimental science** (27:13) — Forecasting the long-term societal consequences of actions by using AI models to simulate complex historical trajectories.

## Related Moments

- [Dissecting artificial intelligence layers from compute to applications](https://www.wearedevelopers.com/videos/100049-beyond-the-wrapper-technical-bets-that-vcs-back) (from "Beyond the Wrapper: Technical Bets That VCs Back")
- [Balancing AI competitiveness with compute efficiency demands](https://www.wearedevelopers.com/videos/1627-pioneering-ai-assistants-in-banking) (from "Pioneering AI Assistants in Banking")
- [Hard science and infrastructure enabled by artificial intelligence](https://www.wearedevelopers.com/videos/100049-beyond-the-wrapper-technical-bets-that-vcs-back) (from "Beyond the Wrapper: Technical Bets That VCs Back")
- [Core drivers fueling modern robotic capabilities](https://www.wearedevelopers.com/videos/1452-robots-2-0-when-artificial-intelligence-meets-steel) (from "Robots 2.0: When artificial intelligence meets steel")
- [Cost and latency pressures pushing AI to the edge](https://www.wearedevelopers.com/videos/100295-from-perception-to-autonomy-building-agentic-edge-ai-robots-with-ros-2) (from "From Perception to Autonomy: Building Agentic Edge AI Robots with ROS 2")
- [Reviewing key architectural decisions for federated AI computing](https://www.wearedevelopers.com/videos/100355-running-secure-life-science-research-at-scale-using-hybrid-gpu-hpc-and-kubernetes) (from "Running Secure Life Science Research at Scale using Hybrid GPU HPC and Kubernetes 🧬")

## Related Articles

- [Stephan Gillich - Bringing AI Everywhere](https://www.wearedevelopers.com/magazine/489-stephan-gillich-bringing-ai-everywhere)
- [MLOps And AI Driven Development](https://www.wearedevelopers.com/magazine/82-mlops-and-ai-driven-development)
- [Graph and AI Trends 2026: Why Is AI Running but Not Yet Delivering?](https://www.wearedevelopers.com/magazine/680-graph-and-ai-trends-2026-why-is-ai-running-but-not-yet-delivering)
- [MLOps – What’s the deal behind it?](https://www.wearedevelopers.com/magazine/125-mlops-what-s-the-deal-behind-it)

## Related Jobs

- [Principal Software Engineer, Enterprise AI Platform](https://www.wearedevelopers.com/jobs/ext/1467292-principal-software-engineer-enterprise-ai-platform) at **GitHub**
- [AI Software Engineer (Germany)](https://www.wearedevelopers.com/jobs/48317-ai-software-engineer-germany) at **Sunhat**
- [Principal Engineer - AI Search & Vector Infrastructure](https://www.wearedevelopers.com/jobs/ext/319507-principal-engineer-ai-search-vector-infrastructure) at **Redis**
- [Principal Engineer - AI Search & Vector Infrastructure](https://www.wearedevelopers.com/jobs/ext/353953-principal-engineer-ai-search-vector-infrastructure) at **Redis**
- [Data Scientist](https://www.wearedevelopers.com/jobs/ext/1351648-data-scientist) at **Almedia**
- [AI Full Stack Engineer](https://www.wearedevelopers.com/jobs/ext/1354435-ai-full-stack-engineer) at **Almedia**