When a founder has an idea for a company that may generate revenue and strives to serve a public good purpose, how does he or she decide whether to originate it as a for-profit or not-for-profit company? As the lines between nonprofits and for-profits continue to blur, this question may be taking on more and more importance. No longer are the organizational structures of each type of company cut-and-dry. Ingenuity in legal structures and changes in tax law now offer more flexibility than ever to combine purposes and mix-and-match entity designs.
Increasingly organizations may have multiple profit distributing and nonprofit entities underneath an umbrella, with a complex network of control, management and ownership rules connecting them. This certainly offers more choice to founders but also may create more opacity in distinguishing the core purposes of and critical differences between a profit distributing company and a nonprofit one. And with this opacity may also come more difficulty in evaluating whether nonprofits are fulfilling their financial, ethical, tax and disclosure responsibilities to retain nonprofit status. A particularly critical need is the ability to know whether evolving hybrid organizational structures that include for-profit and nonprofit arms are contaminating nonprofit mission aims and structural accountabilities that nonprofits have to not distribute profits or confer private benefits (Stein, 2018). As profit orientation expands it also becomes important to consider how that motivation can affect the mission and power dynamics among stakeholders.
Two companies offer unique opportunities to dig into these topics and questions: OpenAI, an artificial intelligence research tank and software maker and Patagonia, a retailer of outdoor sporting apparel and gear. Both companies were founded with one organizational format only to evolve over time into some version of a hybrid nonprofit and for-profit operating structure with different entities and governing approaches than those with which each was founded. Each organization is associated with a charismatic and influential founder and seems to have cultivated radically loyal employee and customer followings. Both companies aspire to deliver massive, bold, arguably impossible missions in furtherance of a global good while operating significant commercialization efforts with a profit distribution aim.
Patagonia and OpenAI have also been broadly covered by a spectrum of media following – and critiquing – their corporate moves and the financial impact of them. All of these similarities make these two brands a good pairing to consider how the kinds of structural blueprints by which they operate might impact their profit orientation and subsequently their ability to satisfy nonprofit mission and governance requirements.
Core areas of exploration
When a market-based business needs money, it can sell products and services or seek investors, among a series of options for generating funds. In these businesses, the companies and their investors expect to make a return on their spend. While nonprofits, like any business, require funds to operate, they receive special tax treatment as compensation for the fact that they are not allowed to operate in the same profit-distribution manner as market-based companies. So those who give money to nonprofits are donors, not investors.
How nonprofits earn the money they do, how they determine profits and losses, whether and how they distribute those and how much tangible and intangible benefit the nonprofit’s leaders and network receive are all essential questions to distinguishing between a company that’s functioning like a nonprofit and one that’s functioning like a for-profit. In the case of Patagonia, the retailer did not begin life as a nonprofit but recently became one. Conversely, OpenAI began as a nonprofit research group that looks increasingly like a for-profit tech start up.
Did Patagonia become a nonprofit simply to help its founding family avoid paying taxes? Should OpenAI have its nonprofit status revoked because of how it’s planning to distribute profits to private investors? How is the difference between appropriate and inappropriate nonprofit revenue generation and profit distribution judged? Where is the line between implicit benefits and explicit benefits to insiders who save billions in taxes or make billions in gains thanks to the nonprofit structure? Is entity flexibility to create hybrid organizations a good or bad thing? What role do nonprofit boards have in setting and answering these questions and does increased profit orientation endanger board controls? How might Patagonia and OpenAI be examples to learn by for the future? This paper will attempt to examine these primary questions and offer some areas for consideration and action.
Patagonia
Founded in 1973 by Yvon Chouinard, Patagonia is a cult brand with over 70 stores across the globe selling outdoor apparel and equipment company for professionals and serious enthusiasts of hiking, mountain biking, camping, climbing and more. Long before it became corporate chic to do so, Patagonia was progressive on environmental causes including creating a sustainable supply chain, recommending reuse of their merchandise and giving back 1 percent of annual profits to environmental causes through its 1% to the Planet Fund. They were early movers who began producing with organic cotton in 1996 and in 2005 committed to reuse and recycle polyester (Balch, 2022).
In 2022 Chouinard made media headlines when he announced that he was donating his interest in the $3 billion company to a new set of nonprofit organizations. While Patagonia is still a for-profit company, 100% of its voting shares are owned and controlled by The Patagonia Purpose Trust, making it the strategic leader of Patagonia’s operations. All profits from Patagonia then go into the Holdfast Collective, which controls 100% of the non-voting shares and invests the proceeds of the company into environmental causes (Cuofano, 2023). As a 501(c)(4), Holdfast is able to participate in advocacy and is dedicated to “fighting the environmental crisis and defending nature” (Gautier and Bothello, 2022). This new structure, as seen in Figure 1, was only made possible by a 2018 tax legislation that allowed nonprofit foundations to control up to 100% of voting shares in a corporation, up from only 20% in 1969 (Gautier and Bothello, 2022).
With this move, Patagonia’s new oversight structure does not disrupt the profit orientation of the core retail operation which its CEO Ryan Gellert has described as “unapologetically a for-profit business” (Balch, 2022). But it does allow it to generate significant contributions for environmental causes in potential perpetuity. The new Patagonia nonprofit structure is expected to give all profits, estimated near $100 million per year, to these causes, which Chouinard has envisioned as a form of ultimate succession planning to protect the mission of Patagonia (Gelles, 2022).
The backlash Patagonia and Chouinard received in the wake of this news included accusations that this was a blindly self-interested effort to avoid estate taxes which would sooner-than-later become an issue for the family of the 84-year-old founder. Through this route they did, in fact, avoid as much as $700 million in inheritance taxes, though they did pay an estimated $17.5 million or more in taxes on the donated shares to the 501(c)(4) since those classifications of nonprofit are not eligible for donation tax deductions. Despite this tax avoidance value, Chouinard says he was looking for a long time for the right vehicle to ensure that Patagonia didn’t fall away from its mission and that he had never been comfortable being a billionaire in the first place (Gelles, 2022).
Skeptics might ask why the company didn’t give away more profit long before now? Why was the 1% to the Planet Fund not a 10% to the Planet Fund or a 90% to the Planet Fund? One might ask why not make this shift the minute the 2018 laws went into effect instead of waiting until an inheritance of some sort is near given the leader’s age? Whether one believes Chouinard’s actions were to protect his family from taxes or that he was architecting a brilliant succession structure to maintain the right balance of profit and giving, one cannot argue that this new Patagonia structure is innovative and fairly simple in outlining how the entities will serve the company’s stated mission. This level of structural ingenuity offers new opportunity to manipulate entity and governance, directly shaping preferred outcomes in taxation, charitable donations, political and lobby support, all of which may come with both benefits and risks.
OpenAI
If Patagonia is an example of a for-profit company that morphed into a nonprofit over time, OpenAI may be the opposite case of a nonprofit that appears to be transforming into a for-profit in real time. Founded as a 501(c)(3) nonprofit in 2015 by a series of tech luminaries, the most famous of which was Elon Musk, OpenAI declared a lofty 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” (Coldewey, 2019).
Within a year Musk and his financial commitment had parted ways with the organization which would come to be associated with current CEO Sam Altman. In 2019 the first public awareness developed around GPT-2, a natural language model that would ultimately become the product to make OpenAI a household name: Chat GPT. In February 2019 the company made headlines for not releasing the full GPT product, advising that they were concerned by the power of the tool and its ability to generate such realistic content that it could impersonate and fake information. Citing safety, security and its mission to protect humanity, OpenAI did not offer the full product release which made some in the open ecosystem community angry, as they did not believe the company should be withholding important advances, given its research-first ethos (Whittaker, 2019).
Only a month later in March 2019 OpenAI announced that it was changing its organizational structure to become OpenAI LP a self-described hybrid for-profit and nonprofit company which would be hallmarked by a capped-profit company that would focus on taking on investors and generating profit returns. This new entity would be managed by the original nonprofit OpenAI Inc. and profits in excess of 100X would be used by the nonprofit to support education and advocacy though in 2025 that cap is slated to increase by 20% each year (Economist, 2023).
Keeping with the pace of extreme change in 2019, only four months later in July came the first $1 billion partnership with tech giant Microsoft (Microsoft, 2019), which checked many mutual boxes for the two companies. First, OpenAI received a lot of the investment value in computing power, which artificial intelligence programming requires in massive amounts. Second, Microsoft and OpenAI aligned around the Microsoft Azure cloud platform as the core of OpenAI’s products and the recipient of co-development of intellectual property. This was a direct competitive stab to both Amazon and Google who are the biggest competitors to both Microsoft and OpenAI in the race for artificial intelligence breakthroughs and commercialization. For those early objectors to OpenAI’s decision not to previously release the full version of GPT this all may have been seen as evidence of the company’s increasing focus on private money-making efforts. Did they hold back the product release until they had a structural vehicle through which to monetize it?
A reported additional $2 billion or more was infused into OpenAI leading up to the now famous ChatGPT-3 launch in November of 2022 which has become the first mass market product to captivate the general public in the complex space of artificial intelligence. Although a free tool, enterprise versions have been launched for recurring fees as has ChatGPT Plus, a $20 per month beefed up version which became available in February 2023. That same quarter, media outlets reported that Microsoft was in talks to close another $10 billion in funding which would yield Microsoft 75% of OpenAI profits until their investment was recouped and then put the Seattle tech company in the minority ownership seat with a 49% stake in Open AI. An additional 49% ownership stake would be distributed across multiple investors with the original OpenAI Inc. nonprofit retaining only 2 percent of the for-profit company’s control (Hoffman and Albergotti, 2023).
Much like 2019, 2023 has turned out to be a pivotal year for the company with a series of white-knuckle governance situations being played out in the media for the world to evaluate. In Q1 2023 while Microsoft was rumored to be considering more investment, OpenAI was rumored to also be working toward allowing employees holding shares to transact those shares for individual gain, just like at any other start up. Once valued at $26 billion, by September 2023 the company was valued at up to $90 billion as they worked toward a transaction that would likely make many of its more than 700 employees millionaires. Projecting to hit $1 billion in revenues this year, the valuation and sale process seems to be functioning quite as it would for any other tech startup in Silicon Valley which values the ability to cash out as a key employee benefit and retention tactic (Seetharaman and Berber, 2023) and quite apart from a nonprofit’s inability to distribute profits to inside individuals.
In November 2023 came the unexpected firing of CEO Altman by the nonprofit board, which still controlled the company on paper, followed by a near unanimous employee outcry to reinstate him, followed by his return and the eventual replacement of almost all of the board of directors. Rumored to be at the center of this series of events is the developing disagreement between Altman and key research employees and the board over how and when to release discoveries and when and how to commercialize given the organization’s original mission to develop research in service to humanity without regard to profits. Also at issue are reported to be Altman’s side projects which included seeking billions for another tech startup that may have been seen as a conflict of interest (Vance et al, 2023).
It is difficult to review with hindsight the chronology of structural, investment and commercial milestones since 2019 and not see a coordinated series of events designed to help OpenAI shed its nonprofit limitations without ceding its nonprofit designation. Doing so enabled OpenAI not only to raise the billions needed to successfully innovate in artificial intelligence but also to ensure that the groups providing those billions would be allowed to profit from those contributions in venture capital-like ways as opposed to tax reduction-like ways. The gains that Microsoft and other investors or shareholding employees will receive from a traditional investment vehicle will certainly dwarf any potential value created by a nonprofit-corporate partnership or a traditional donor relationship.
Curiously, the nonprofit narrative around the OpenAI brand and mission have not seemed to change in step with the company’s structural changes. Altman’s statements and the company messaging continue to center a humanitarian intention to make the world better through and yet protect the world from their inventions. As they have steadily shifted into a profit center in many formal ways, the company has continued to speak like a nonprofit and refer to itself as a nonprofit. Whether an intentional misdirection or simply an effective branding strategy, this approach seems to be insulating the business and Altman from many tough questions in the big media outlets.
OpenAI generates many delicate questions, including might the near unanimous employee support in November be related to Altman’s intention to help employees sell hundreds of millions of dollars in stock as recently as September? If the employees did not rise up to influence the board, their personal interests might not be served. Or questions like whether Microsoft controlling half of the company and receiving 75% of the profits is a recognition of the true value of the company’s worth or is instead a deal that more serves Microsoft’s interests over OpenAI’s? These questions matter a lot if OpenAI is a nonprofit and not much at all if it is not. Reviewing OpenAI’s current structural map, as seen in Figure 2, shows that they have retained the primary controlling entity as a nonprofit. This chronology of profit distribution activity arguably calls Open AI’s nonprofit designation into serious question – in reality if not on paper – as the structure may be completely allowable even if not ethical to the mission of the nonprofit, which explicitly demands not having profit motivations at its core focus.
Themes
The two examples of Patagonia and OpenAI have in common novel organizational structures, a predominant focus on profit generation and claims to better and protect humankind and Earth. Three questions offer rich territory to evaluate these companies and the potential benefits and drawbacks from the strategies they have pursued as well as provide areas for future consideration as other businesses look for models and best practices.
What business model do OpenAI and Patagonia appear to represent and is that model unique to them or available to other nonprofits to follow?
Patagonia appears to qualify as a social enterprise considering that its earned income goes beyond incidental and forms a significant and defining portion of its total activity. And the organization has declared that 100% of generated profits will go to mission-relevant causes, satisfying the social value component of the social enterprise (Worth, 2021). Of the common social enterprise models, the product model seems the most relevant to Patagonia since it produces tangible products available for sale at scale that perform certain functions for their consumers (Besharov, M et al., 2019).
This model is widely available to nonprofits that have in-demand products that can be reliably produced, distributed and monetized. While a promising option, this social enterprise model does require significant investment to thrive and to generate any significant income or profit. Organizations should ensure they can manage and perpetuate such resource needs as production facilities, logistics of distribution, marketing and digital capabilities and order and billing capabilities, plus determine how to handle issues like customer returns or disputes. Nonprofits without sufficient startup resources may struggle to find value in the product model. Those successful with a product model should also take care to continuously evaluate whether their product sales mission is serving their larger public good mission.
On paper OpenAI might claim a similar rationale for being considered a product oriented social enterprise, including that they require significant investment and profit generation to finance their nonprofit mission of advancing artificial intelligence. And they do offer a software product. But by structuring profits to go first to investors like Microsoft, by allowing multiple investors to buy into the for-profit LP, by setting an extremely high profit cap that can rise over time and by leaving the nonprofit entity with only 2 percent of company shares in the for-profit arm, OpenAI signals they are more like a classic for-profit tech startup than a nonprofit social enterprise.
Organizations with a significant intellectual property product that requires extremely high capitalization and organizations competing in tech industries may be well advised to follow the OpenAI model but choose to incorporate as a traditional start up for maximum investment and profit return flexibility. The nonprofit element may attach undue governance hurdles that may not ultimately prove valuable to the business.
At OpenAI and Patagonia, is the nonprofit designation the chicken or the egg and what does that mean for other companies looking on?
As much as a new business striving to emulate OpenAI’s success in the tech market might be advised to incorporate as a traditional startup versus as a nonprofit, there is no denying that the nonprofit designation is providing intangible and competitive value to OpenAI. Brad Smith, the president of OpenAI partner and investor Microsoft, recently confirmed this by asking at an industry conference, “Which would you have more confidence in? Getting your technology from a non-profit, or a for-profit company that is entirely controlled by one human being?” (Economist, 2023). As they both compete against juggernauts like Facebook, Google and Amazon, branding OpenAI as a benign research tank and advocate for humanity, leveraging its nonprofit status is a competitive advantage. For OpenAI the profits seem now to rely in part on the credibility the nonprofit moniker provides, especially as the public grows skeptical and fearful of the unknown power of AI. For OpenAI, the profit mechanism seems primary and the nonprofit designation is secondary.
In contrast, the structural changes Patagonia made place the profits into reliance on the mission. Governance and program definitions have made the distribution of those profits to mission-relevant environmental causes the only purpose for the core business, removing any conflict. Here, the profit mechanism is secondary, simply a form of fuel that defers to the nonprofit designation, which is primary. As nonprofits and nonprofit watchdogs consider how to innovate in entity structure the question of which motive is driving which decision becomes essential. Is profit raising critical to creating mission results or will the mission fail without the credibility offered by the nonprofit badge? A well-meaning and ethical team would ask themselves these questions and choose the right path but just as possible is the team that might plainly know the trojan value of a nonprofit label and set up a structure that will satisfy, if not exactly sincerely fulfill, the legal and tax requirements of a nonprofit simply to receive its goodwill value. As legislative and legal structural choices become more flexible, this may be an expanding area of concern.
What might the Patagonia and OpenAI examples suggest about the shifting power and profit dynamics between founders, boards, investors and the public that the nonprofit world should watch?
A nonprofit purist may look at the Patagonia example and feel it represents a ‘good’ level of profit orientation with ‘good’ outcomes since the profits will go toward environmental causes clearly tied to its mission. That same purist may look to OpenAI and see it as a ‘bad’ reorganization that exploits the nonprofit label to further its ability to monetize its research and inure private benefit to stock holding employees, investors and founders. In either case both outcomes are made possible by recent legislative adjustments that make complex hybrid for-profit-nonprofit entity constellations legal and feasible.
The kinds of radical mash-ups in profit generation and nonprofit mission that are now possible will increase the innovation that founders try to bring to their organizational structures. Some of that innovation will couple with ‘good’ motivations and some with ‘bad’ driving requisite outcomes that will subjectively be judged by the beholder. Important for the nonprofit governance community will be to consider how to judge these emerging structures with more than sentiment and how they might be evaluated and overseen by legal, tax and ethics bodies.
The OpenAI case also sharply points out how easily board power can be sidelined. A charismatic founder with a series of vocal stakeholders with monetary interests (employees and investors) was able to leverage public pressure and the media cycle to influence the board to reverse its decision to fire the CEO, to achieve the CEO rehire, to largely resign its seats, to be replaced by new members from highly commercial sectors and to almost certainly add a board role for the largest profit-seeking investor. This chain of events has nothing to do with the legal structure on its face but the structure enabled the expansion of the stakeholder pool to grow to include many of these actors who did not exist at the nonprofit’s formation.
Nonprofit boards and founders looking to create or expand profit centers might look earnestly at this example to anticipate how profit motivations and the awareness of those that may develop in the media and the public sphere could affect their own decision making. Watchdog groups might consider developing criteria to imagine organizational vulnerability to board unseating or potential for overpowering the board for certain kinds of high profile or highly profit-oriented nonprofits. The power dynamics exist arguably due to the profit incentives but those incentives are made possible by the allowance of hybrid structures.
Call to Action
Nonprofit organizations must have funding to carry out their mission. For-profit companies are able to employ all the features of the market economy to support their objectives and so, too, should nonprofits. Participating more fully in that market economy will have meaningful consequences, so deciding to pursue profit generation and structuring an organization to do so at scale while remaining a nonprofit should be treated as a very delicate and conscious choice. This is a choice that can change the balance of power between entities and decisionmakers. This is a choice that can unduly influence outside actors to participate in or interfere in the nonprofit’s organizational work. This is a choice that can motivate founder and board actions.
Profit generation and distribution can be both an enabler and a detractor from a nonprofit’s core mission as demonstrated by the two company examples. Patagonia’s reorganization put power into the hands of the nonprofit to activate the company’s retail profits against its mission while OpenAI has steadily reduced the influence of the nonprofit over its for-profit activities. In both cases the sheer designation of nonprofit is proving a material branding value to each company though it remains to be seen whether both will ultimately live up to the legal, tax and ethics standards that being a nonprofit entail.
Neither company would be able to pursue its current profit strategy without the ability to innovate through nonprofit and hybrid entity structures. While The Harvard Business Review has reported that similar hybrid companies known as shareholder foundations have been functioning in Europe for decades (Gautier and Bothello, 2022) these kinds of combination businesses still seem to be new to the United States. Now, during this period of trial and development, is the right time for the nonprofit industry and professional standards thinkers to be evaluating companies like OpenAI and Patagonia, publishing case studies, identifying benefits and challenges, developing best practices and flagging areas of caution. There is much power to be gained as organizations strike an optimal balance between profit generation and nonprofit mission accountability that could help expand resources to drive important outcomes. But now is the time to seek transparency and drive standards before nonprofit status becomes less a method of organizing with a clear set of mission values and practices and more like a shiny, hollow brand badge.
Reference List
Axios. How OpenAI board is structured. https://www.axios.com/2023/11/18/how-openai-board-is-structured
Balch, O. (2022, April 11). Ryan Gellert: ‘At Patagonia we have to be greener than green – and make a profit’. Reuters. https://www.reuters.com/business/sustainable-business/ryan-gellert-patagonia-we-have-be-greener-than-green-make-profit-2023-04-11/
Besharov, M et al. (2019). The many roads to revenue generation. Stanford Social Innovation Review, (Fall).
Coldewey, D. (2019, March 11). OpenAI shifts from nonprofit to ‘capped-profit’ to attract capital. TechCrunch. https://techcrunch.com/2019/03/11/openai-shifts-from-nonprofit-to-capped-profit-to-attract-capital/
Cuofano, G. (2023, December 13). Patagonia organizational structure. Four Week MBA. https://fourweekmba.com/patagonia-organizational-structure/#:~:text=Patagonia%20has%20a%20particular%20organizational,defining%20the%20company’s%20strategic%20direction.
Gautier, A. and Bothello, J. (2022, October 10). What happens when a company (like Patagonia) transfers ownership to a nonprofit? Harvard Business Review. https://hbr.org/2022/10/what-happens-when-a-company-like-patagonia-becomes-a-nonprofit
Gelles, D. (2022, September 14). Billionaire no more: Patagonia founder gives away the company. The New York Times. https://www.nytimes.com/2022/09/14/climate/patagonia-climate-philanthropy-chouinard.html
Hoffman, L. and Albergotti, R. (2023, January 9). Microsoft eyes $10 billion bet on ChatGPT. Semafor. https://www.semafor.com/article/01/09/2023/microsoft-eyes-10-billion-bet-on-chatgpt
Microsoft News Center (MNC). (2022, July 19). Microsoft [Press release]. https://news.microsoft.com/2019/07/22/openai-forms-exclusive-computing-partnership-with-microsoft-to-build-new-azure-ai-supercomputing-technologies/
OpenAI. Introducing ChatGPT Plus. OpenAI Blog. https://openai.com/blog/chatgpt-plus
Patagonia, Inc. Owned and operated. https://www.patagonia.com/where-we-do-business/owned-and-operated.html
Seetharaman, D. and Berber, J. (2023, September 6). OpenAI seeks new valuation of up to $90 billion in sale of existing shares. The Wall Street Journal. https://www.wsj.com/tech/ai/openai-seeks-new-valuation-of-up-to-90-billion-in-sale-of-existing-shares-ed6229e0
Smith, J. (2022, September 18). How Patagonia surfed around death and taxes: Founder Yvon Chouinard plans for a future without him. Sierra. https://www.sierraclub.org/sierra/how-patagonia-surfed-around-death-and-taxes
Stein, J. (2018). New model for oversight of commercial activities by nonprofits. Fordham Law Review, 86(4), 2029-2070.
The Economist Newspaper Limited. (2023, November 21). Nonprofit motives: Inside OpenAI’s weird governance structure. The Economist. https://www.semafor.com/article/01/09/2023/microsoft-eyes-10-billion-bet-on-chatgpt
Vance, A., Ludlow, E. and Savov, V. (2023, November 20). Nearly all of OpenAI staff threaten to go to Microsoft if board doesn’t quit. Yahoo! Finance. https://finance.yahoo.com/news/openai-staff-threaten-microsoft-board-144547958.html?guccounter=1
Whittaker, J. (2019, February 17). OpenAI built a text generator so good it’s considered too dangerous to release. TechCrunch. https://techcrunch.com/2019/02/17/openai-text-generator-dangerous/
Worth, M. (2021). Nonprofit management: Principles and practice (6th ed.), 204. Sage Publications.