Premium: Wave after wave of demand
NVIDIA is aggressively driving demand for AI compute through agentic AI software investments, ecosystem partnerships, and supply chain positioning. Major AI buildouts are scaling dramatically, with individual facilities projected to grow from 400-600MW this year to over 2GW by 2028-2029. Frontier AI labs like OpenAI and Anthropic are diversifying compute sources, while Google is pushing its TPU ecosystem into neoclouds.
Summary
The transcript covers NVIDIA's multi-pronged strategy to accelerate AI adoption and GPU demand, focusing on three main areas: agentic AI, ecosystem investments, and ongoing AI infrastructure buildouts.
On the agentic AI front, NVIDIA is providing open-source LLMs and new agentic software layers to drive enterprise adoption. The company is leveraging its existing CUDA-X libraries to accelerate data workloads for agents, which simultaneously increases GPU demand and offloads work from CPUs. The transcript argues that deeper agentic AI adoption will compound compute needs through long-running orchestration, spawned sub-agents, task-specific model calls, and accelerated data processing.
Regarding ecosystem investments, NVIDIA is using partnerships to secure upstream supply chain capacity while simultaneously seeding capacity for neoclouds, frontier AI labs, and sovereign AI initiatives. This dual-direction strategy helps NVIDIA maintain leverage across the value chain.
The AI buildout wave is described as coming from all directions simultaneously — hyperscalers, neoclouds, frontier AI labs, sovereign AI, and private capital funds. Individual AI campus scale is projected to grow dramatically, from 400-600MW housing 350-600K GPUs today, to over 2GW housing 3.5-4.5 million GPUs by 2028-2029. Management expects buildout costs to rise from $50-60B per gigawatt to $80-100B due to increasing density requirements.
On the competitive front, Anthropic is expanding beyond TPUs into GPUs, committing capacity from Azure, CoreWeave, and SpaceX. OpenAI is deepening its AWS relationship and will adopt Trainium for some agentic services but remains primarily GPU-focused. Google is aggressively expanding its TPU ecosystem and increasing capex, with Anthropic signing a new TPU deal and hints that OpenAI may also be involved.
About this episode
Agentic AI demand, ecosystem investments, ongoing AI buildouts
Key Insights
- NVIDIA is using open-source LLMs and CUDA-X libraries not just as developer tools but as demand-generation mechanisms to increase GPU consumption across agentic workloads.
- The largest individual AI buildouts are projected to grow roughly 6-8x in GPU count by 2028-2029, from 350-600K GPUs to 3.5-4.5 million GPUs per facility, signaling a step-change in infrastructure scale.
- NVIDIA management expects per-gigawatt buildout costs to rise from $50-60B to $80-100B, attributing the increase to rising compute density rather than just inflation or supply constraints.
- Anthropic, historically associated with Google TPUs, is now committing to GPU capacity across Azure, CoreWeave, and SpaceX — representing a meaningful diversification of its compute strategy.
- Google is actively pushing its TPU ecosystem into neoclouds and has secured Anthropic as a customer in a new deal, with hints that OpenAI may also be drawn into the TPU ecosystem despite its GPU-first posture.
Topics
Transcript
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…
Full transcript available for MurmurCast members
Sign Up to AccessMore from HHHYPERGROWTH
Premium: Farther out waves
NVIDIA is expanding its AI ecosystem beyond data centers into physical AI, edge AI, and industrial automation through autonomous vehicles, robotics, and smart devices. The company is anchoring itself as the end-to-end stack across multiple compute tiers, while NVLink Fusion partnerships and a future roadmap including Vera Rubin Ultra and Feynman extend its dominance further.
Premium: Vera Rubin decoder ring
NVIDIA's Vera Rubin platform represents a strategic shift toward Agentic AI at scale, disaggregating workloads across specialized chips and rack systems. Key announcements include a Groq-powered LPX rack for disaggregated inference, a standalone Vera CPU rack for agentic orchestration, and a redesigned MGX modular architecture that dramatically reduces assembly time. NVIDIA is also scaling up supply chain capacity and expanding networking capabilities to support clusters exceeding 500,000 GPUs.
Premium: Modular inference
NVIDIA's Vera Rubin platform represents a major expansion into a modular AI factory system, incorporating 7 chips including the acquired Groq architecture. The platform is expected to begin generating revenue in Q3 2027, with management projecting significant TAM expansion and revenue uplift across multiple dimensions including inference, storage, and CPU workloads.