*spins roulette wheel* Moelis and OpenAI!

I had so many choices for what to blog about this week.  The dispute about Donald Trump’s Truth Social SPAC?   Chancellor McCormick’s conclusion that the Activision/Microsoft merger might have violated Delaware law?  VC Laster’s Moelis decision?  Musk’s lawsuit against OpenAI? 

I ultimately decided to go with Moelis and OpenAI because they actually are fundamentally kind of the same thing, and this way I kill two birds with one stone.

So, earlier this week, it seemed like the big business law news was VC Laster’s holding in West Palm Beach Firefighters’ Pension Fund v. Moelis, issued last Friday, invalidating the shareholder agreement that Ken Moelis reached with Moelis & Co. when he took it public, and that allowed him to functionally remain in control of the business even when his voting stake dropped below a majority.  VC Laster held that the contract violated DGCL 141(a), which requires that corporations be managed by their boards of directors.

VC Laster recognized that every time a corporation enters into any kind of contract at all, the board’s choice set becomes more limited, but – using a word that I personally had never heard before and don’t know how to pronounce – concluded that there is a spectrum, starting with ordinary commercial contracts and ending with arrangements that ultimately intrude so far into corporate governance that they leave the realm of the commercial and raise a Section 141(a) issue.  Relying on Abercrombie v. Davies, 123 A.2d 893 (Del. Ch. 1956), he explained that the validity of a contract therefore depends on two inquiries: first, is it a governance arrangement at all?  If not, we’re done.  A number of factors go into determining whether the contract implicates corporate governance, including whether the contract involves some kind of commercial exchange, and whether it restricts the actions of internal corporate actors, namely, boards and shareholders.  

If the contract does implicate corporate governance, it will be invalid if it substantially restricts the board’s ability to manage the company.   In this case, the Moelis contract fit both bills: there was no commercial exchange, and the contract limited the board’s choices at every turn.  (My personal fave: The board was not permitted to discard the “Moelis” name without Moelis’s approval, even though name changes aren’t usually subject to a shareholder vote at all).

Among other interesting questions raised by the opinion are – what are the nondelegable functions of a corporate board?  DGCL 141(a) requires that corporations be “managed by or under the direction of a board of directors,” but in practice boards take advantage of the “under the direction of” piece and delegate most of those responsibilities to others.  What are the definitional board responsibilities that must remain with the board itself?  Certainly, merger negotiations fall into this category.  And in the Caremark context, Marchand v. Barnhill’s concept of “mission critical” risks is, functionally, a delineation of compliance responsibilities that are definitionally part of the board’s job and cannot be delegated to others.  The Moelis case also circles around the idea of core board functions that cannot be outsourced without violating Section 141(a).

Laster explains, however, that – though there are still outer limits imposed by the DGCL – many restrictions that would be prohibited in a stockholder agreement may still be permissible if they are contained in the corporate charter, or even in a new preferred stock issuance whose terms become part of the charter by operation of law.  And that’s the thing that grabs me.

Why does it matter whether something is in a shareholder agreement, versus a preferred share issuance?

Well, we can talk about transparency, and procedures for amendment – topics covered by Jill Fisch in her paper, Stealth Governance: Shareholder Agreements and Private Ordering.  I can also point out that now, after Moelis is already listed on the New York Stock Exchange, issuing new preferred stock that undermines the governance rights of existing shareholders may not be possible (does it count as a disparate reduction or restriction on shareholder rights if a shareholder agreement that existed at the IPO stage was invalidated and reconstituted in the form of a preferred share issuance?  I dunno).  But also, as I explain in my paper, Inside Out, preferred share terms are treated as internal affairs matters, subject to the law of the state of incorporation.  But shareholder agreements are ordinary contracts, and subject to ordinary choice of law.  Indeed, it’s not uncommon for shareholder agreements to contain a choice of law provision that is different from the state of incorporation.  (Yeah, that’s right, I answer to “Cassandra” now).

So the difference between putting it in a shareholder agreement and a preferred share issuance may be choice of law.

EXCEPT.

VC Laster, when explaining how to test for whether a shareholder agreement addresses governance matters (again, remember the mere fact that it addresses such matters does not invalidate it; you need to go to the second step of degree of constraint), he repeatedly referred to “internal affairs” and “internal governance.”  He didn’t mention choice of law, but the implication? was kind of? that a contract containing such restrictions might not, in fact, be subject to ordinary choice of law after all?  And might be deemed to be subject to the law of the state of organization? 

Which brings us, hilariously, to Elon Musk’s suit against OpenAI.  The gravamen of which, as I understand it, is that Sam Altman had an idea for developing AI for the benefit of “humanity,” pitched this to Musk, and Musk donated a bunch of resources while Altman organized multiple (Delaware) entities to effectuate the vision.  Generally speaking, those entities consist of a Delaware nonprofit corporation, OpenAI, Inc., which has a number of Delaware for-profit LLC subsidiaries.  Musk now advances various California breach of contract and analogous claims because, in his view, OpenAI, Inc. and its subsidiaries have failed to operate in accordance with the original humanity-focused set of promises. “This case is filed to compel OpenAI to adhere to the Founding Agreement and return to its mission to develop AGI for the benefit of humanity,” is an actual allegation at ¶33.

I will leave it to the nonprofit folks to talk about the standing of a donor to sue when a nonprofit fails to operate in accordance with expectations, though the fact that the complaint cites the relevant statute as “E.g. Cal. Bus. & Prof. Code § 17510.8” does not inspire confidence (the “eg” means “for example,” which I interpret as, there is no direct statute on point).  I will leave it to the contract folks to talk about whether promises of benefitting humanity are enforceable, or whether you can even cobble together a contract from the multiple conversations, emails, and organizational documents Musk cites.

Instead, I’ll talk about the fact that, in addition to disgorgement and damages, Musk seeks specific performance in the form of:

1. An order requiring that Defendants continue to follow OpenAI’s longstanding practice of making AI research and technology developed at OpenAI available to the public, and
2. An order prohibiting Defendants from utilizing OpenAI, Inc. or its assets for the financial benefit of the individual Defendants, Microsoft, or any other particular person or entity;

In other words, Musk claims that he has a contract, formed under California law, that allows Musk to dictate the governance choices of a Delaware organized (nonstock) corporation, and further, that a California court should order that this Delaware corporation conduct itself in accordance with his contract. 

Now, as a nonprofit, OpenAI is subject to jurisdiction not only of the Attorney General of its state of organization, but also the Attorney General of the states where it operates.  And I have no idea whether it has violated the legal rules governing nonprofits in those states – that’s up to the AGs.

But Musk is not invoking AG authority; he’s claiming a contractual right to dictate the governance of a Delaware-organized (nonstock) corporation.  So, especially in light of Moelis, if OpenAI wanted to take this as a serious threat, it could file a declaratory judgment action in Delaware to the effect that these “contracts” – if they even exist – are in fact contracts concerning internal affairs matters governed by Delaware law.  And, under Delaware law, at least at the corporate (OpenAI, Inc.) level, they are illegal intrusions into the board’s authority. (Though, I suppose, it might have difficulty getting personal jurisdiction over Musk in Delaware these days)