Means, ends, and OpenAI
OpenAI's likely transition from nonprofit to private control reveals what (we believe) nonprofits are for.
“We shape our buildings, and afterwards our buildings shape us.”
–Winston Churchill, comments to the House of Commons, 1943
OpenAI – the company behind ChatGPT, DALL-E, and other artificial intelligence tools – is oddly structured for a global juggernaut. The organization is held and governed by a nonprofit entity (OpenAI, Inc.), with a for-profit subsidiary (Open AI LLC) that manages a limited partnership (OpenAI LP). The for-profit components of the enterprise also explicitly limit (or “cap”) economic returns to investors. So, it’s a nonprofit that governs a constrained for-profit that runs a limited partnership.
As the company describes these choices on its website:
We designed OpenAI’s structure…as a chassis for OpenAI’s mission: to build artificial general intelligence (AGI) that is safe and benefits all of humanity.
The emphasis on chassis is mine, since it’s an intriguing and useful word. “Chassis,” of course, is most commonly used to describe the “base frame of a motor car, with its mechanism” (OED). In electronics, the chassis refers to the “frame on which the parts of a radio receiver are mounted [including] the assemblage of parts on the frame” (OED). In both cases, the chassis determines the underlying performance, capacity, and behavior of a device before the bells and whistles are attached.
So why build such a convoluted chassis for OpenAI? Because the enterprise operates at a complex intersection of mission, money, and marketplace.
Mission: OpenAI was founded with a mission “to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return.” That suggests decision-makers who value purpose over profit, and therefore a nonprofit entity that is “barred from distributing its net earnings, if any, to individuals who exercise control over it, such as members, officers, directors, or trustees” (Hansmann 1980).
Money: To approach that mission, it needs a whole lot of money. By some estimates, the company is spending $7 billion a year (Sorkin et al 2024). That suggests the need for a for-profit subsidiary that offers at least a little opportunity for investors to earn financial return.
Marketplace: OpenAI is operating in an unknown and unknowable marketplace, where the systems and services require radical new thinking to even guess about actual economic value. That suggests the need for a limited partnership where multiple participants can pool their investments and protect themselves against liability.
OpenAI therefore needs a chassis that’s built for mission alignment, money magnetism, and radical marketplace uncertainty. A governing nonprofit with a for-(limited)-profit subsidiary that controls a limited partnership seemed to offer the best configuration. And yet, OpenAI is signaling a plan to change that chassis (PBS News 2024, Hu and Kai 2024), reducing or removing the nonprofit control.
Why does this matter to arts managers and arts organizations? Because many arts organizations are built on a similar (but simpler) chassis. Their nonprofit governance discourages self-serving behavior and attracts contributions. Their contributed revenue and volunteer/discounted labor allow for more or different programs than the commercial marketplace would allow. And these underlying dynamics shape how the organization moves through the world.
And yet, many arts managers never consider the chassis underlying the “car” they’re driving, and therefore never realize the forces that are tugging the steering wheel in unexpected directions.
My educator/scholar colleague Joan Jeffri once described policy analysis to me in three simple questions: Who chooses? Who benefits? Who pays? The “chassis” of a business entity is an answer to those questions, and an engine for what’s possible and probable for the enterprise.
If OpenAI changes its structure, it will invariably change its decision-making and behavior in both expected and unexpected ways. Let’s hope they bring a deep and careful human intelligence to the next chapter.
From the ArtsManaged Field Guide
Function of the Week: Finance
Finance involves designing, maintaining, and sustaining systems of money and stuff.
Framework of the Week: Capital Continuum
Capital Continuum refers to the various types of durable resources needed to build and sustain an arts enterprise, influenced by the motivations of donors and investors. It emphasizes understanding how different funding sources, from charitable donations to venture capital, align with the goals and risks of an arts initiative.
Image generated by artificial intelligence.
Sources
“OpenAI Looks to Convert from Nonprofit Roots and Become For-Profit Company.” 2024. PBS News. September 26, 2024.
Hansmann, Henry B. 1980. “The Role of Nonprofit Enterprise.” Yale Law Journal 89:835–901.
Hu, Krystal, and Kenrick Cai. 2024. “OpenAI’s Huge Valuation Hinges on Upending Corporate Structure.” Reuters, September 14, 2024, sec. Artificial Intelligence.
Sorkin, Andrew Ross, Ravi Mattu, Bernhard Warner, Sarah Kessler, Michael J. de la Merced, Lauren Hirsch, Ephrat Livni, and Nell Gallogly. 2024. “OpenAI Aims for a $150 Billion Valuation.” The New York Times, September 12, 2024, sec. Business.