OpenAI: The Compute Bet

OpenAI

By AKA Studio

What’s inside

• NVDA NVIDIA: Backing $105B of financing for OpenAI's Ohio campus; the substrate for every model.
• TSM TSMC: Fabricates the chips behind every OpenAI training cluster; no substitute at the leading edge.
• AVGO Broadcom: Custom accelerator and networking partner as OpenAI pushes toward its own silicon.
• MU Micron: Memory is the binding constraint on the long-context, agentic models OpenAI keeps shipping.
• MSFT Microsoft: Deepest commercial partner; OpenAI models drive its enterprise AI story.
• ORCL Oracle: Contracted data-center capacity for OpenAI-scale training and inference.
• AMZN Amazon: Agreed to invest $50B, much of it tied to cloud capacity for OpenAI.
• CRWV CoreWeave: Pure-play rented GPU capacity; trades tick-for-tick with OpenAI news.
• VRT Vertiv: Cooling and power gear for the dense clusters OpenAI's training runs require.
• GEV GE Vernova: Generation and grid equipment needed to energize OpenAI-scale campuses.
• CEG Constellation Energy: Firm round-the-clock generation is what AI campuses of this size contract for.
• AEP American Electric Power: OpenAI's note says the first 800MW in Ohio comes largely from existing AEP infrastructure.
• NOW ServiceNow: Workflow layer where OpenAI-style agents get deployed inside large companies.
• PANW Palo Alto Networks: OpenAI's own safety warnings make enterprise defense spend the mirror trade to agent rollout.
• CRWD CrowdStrike: Rogue-agent incidents push buyers toward runtime detection of autonomous behavior.

Why it could work

If OpenAI keeps raising tens of billions and spending it on chips, clouds and power, the suppliers it pays capture the value whether or not OpenAI itself ever lists.

What could break it

Repeated safety pauses, a delayed IPO or a shift of the lead to rivals could slow the spending that every holding here depends on.

For learning only. Not investment advice.