Now that we've covered Vera Rubin and its modular racks, let's focus on NVIDIA's strategic moves to spur downstream demand, secure upstream supply, and expand its ecosystem. This includes their push to accelerate enterprise agentic AI adoption, their ecosystem investments across the supply chain and neoclouds, and the latest wave of major AI buildouts.

  • Agentic AI is now driving a major inflection in inference, and it is still just getting started.
  • NVIDIA is providing open-source LLM models and new agentic software layers to help spur enterprise adoption of agentic AI.
  • Agents will also thrive on data access. NVIDIA is leveraging existing CUDA-X libraries to help accelerate agentic data workloads, spurring even more GPU demand while freeing up the CPU for other agentic tasks.
  • Deeper agentic AI adoption will magnify compute needs from here across long-running orchestration, spawned off sub-agents, calling or running separate AI models specific to task, and accelerated data processing.
  • NVIDIA’s ecosystem investments and partnerships help secure upstream capacity in the supply chain, while seeding downstream capacity for rising neoclouds, frontier AI labs, and sovereign AI.
  • The recent wave of AI buildouts shows demand coming from all directions at once, across hyperscalers, neoclouds, frontier AI labs, sovereign AI, and private capital funds.
  • The largest of individual AI buildouts are projected to grow from an average 400-600MW (housing 350-600K GPUs) this year to >2GW (housing 3.5-4.5M GPUs) in 2028-2029. Mgmt expects buildout costs to rise from $50-60B per 1GW to $80-100B, given the coming increase in density.
  • Frontier AI labs like Anthropic and OpenAI are seeking compute wherever they can find it. Anthropic is finally branching into GPUs, and has committed to capacity from Azure, CoreWeave, and SpaceX.
  • OpenAI is snuggling up with AWS and will adopt Trainium for some coming agentic services – but seem to remain primarily focused on GPUs.
  • Meanwhile, Google is building its own competing TPU ecosystem, and just announced it is heavily ramping up its planned capex spend. And Anthropic is helping fuel that new TPU ecosystem with a brand new deal this week (with hints that OpenAI might be involved).

Part 3:

  • Tidbits from GTC Taipei
  • Agentic AI demand
  • Ecosystem investments
  • AI buildouts continue
  • OpenAI shift
  • Google's TPU push into neoclouds

We aren't quite done with NVIDIA Week. A final part will cover their farther out waves to expand their ecosystem across physical AI, edge compute, and NVLink Fusion, as well as their roadmap after Vera Rubin.