Corporate structure stories often collapse three different questions into one: who controls the organization, who operates the business, and who holds an economic interest. OpenAI's published structure makes those questions distinct.
The three-layer reading
At the top of the visual is nonprofit control. The operating company sits below that governance layer. Employees and investors can have economic interests without that fact alone answering the control question.
This is why a conventional stakeholder list is a poor representation. Lists encode presence; they do not encode authority.
Four relationships that should never share one arrow
Organization diagrams often use a generic line for relationships with very different meanings. A control map should distinguish at least these:
| Relationship | Question answered |
|---|---|
| controls | Who has formal governance authority? |
| operates | Which entity conducts the business activity? |
| holds an economic interest | Who participates financially? |
| appoints or selects | How is a governing role filled? |
The same organization can appear in more than one relationship. That does not make the relationships interchangeable. A financial interest does not automatically establish board control; a board role does not automatically describe day-to-day product reporting lines.
What the visual does — and does not — claim
The diagram shows relationship categories described on OpenAI's own structure page. It is not a cap table, does not assign ownership percentages, and does not reconstruct every legal entity.
It also does not show every contractual right, subsidiary, governance mechanism, or economic term. Those details would require governing documents and current disclosures beyond a public orientation page. The diagram's claim is intentionally narrower: the published structure describes a nonprofit governance layer, an operating business, and economic stakeholders as distinct categories.
Confirmed: the published distinction between the nonprofit governance layer and the operating business. Editorial simplification: the compact grouping and labels used to make the relationship readable.
How we built the map
We extracted named entities and relationship verbs from the source, then normalized the verbs into typed edges. We did not infer percentages or reporting relationships from prominence on the page. Where the public explanation summarized several legal mechanisms, the visual kept a higher-level node rather than inventing detail.
Each future revision should be processed as a diff:
- identify an added, removed, or renamed entity;
- identify the exact relationship statement that changed;
- determine whether the change concerns governance, operations, or economics;
- preserve the earlier dated version;
- link the changed edge to the new primary source.
This makes the artifact useful during fast-moving coverage. A reader can see whether a story concerns control, a financing relationship, or an executive role before reading a paragraph of caveats.
A better way to cover organization changes
When a board member, investor, or executive changes, update the node at the layer where the change occurred. That prevents a board appointment from being presented as an operating reorganization, or an investment from being presented as a transfer of mission control.
Three examples show why the layer matters. A new executive can change operating leadership without changing formal control. A new investor can change economic participation without obtaining governance rights implied nowhere in the source. A governance amendment can change control even if the product organization looks identical the next morning.
This is also the reason MyMap's org chart maker is only one tool in the workflow. A control map needs typed relationships — “controls,” “operates,” and “holds an interest” — rather than reporting lines alone.
What would trigger a correction
We would revise the diagram if OpenAI changes its published structure, if a controlling document contradicts our edge label, or if we discover that a node combines entities whose separation is necessary to understand authority. A new funding announcement alone does not justify redrawing control; the attached rights and authoritative description determine the edge.
We would not update this page merely because a secondary article uses a different metaphor for the structure. Secondary reporting can reveal a question, but the map changes only after the relationship is supported by a primary source appropriate to the claim.
Citation note
Readers citing this article should cite OpenAI for the underlying structure and MyMap for this particular visual arrangement. If a legal or financial decision depends on the exact entities or interests, consult the governing documents and current disclosures rather than this simplified map.
A precise citation should state the access date because organization structures can change. When embedding the SVG, retain its title, source line, and “not a cap table” limitation. Cropping those labels away turns a bounded explainer into a broader claim than the article supports.