For most of the last century, Delaware was the default home for American incorporation. That is now changing. After the Court of Chancery struck down Elon Musk’s multibillion-dollar Tesla pay package in Tornetta, citing his undue influence over the board, companies began relocating their legal home—a trend now known as ‘DExit’.By August 2026, about eighty companies had announced plans to leave, most for Nevada or Texas. The Delaware Supreme Court reinstated the package in December 2025, in re Tesla, holding rescission an improper remedy, but it did not dispute the board’s lack of independence, and by then the exodus was already underway.
Texas and Nevada compete on familiar terms, Texas with a new specialist Business Court and Nevada with an even more management-friendly code. Most reincorporations cluster among founder-led firms in technology, crypto, and artificial intelligence, the very companies Delaware’s latest proposal aims to attract. The competition is no longer about incorporation costs but about how much accountability a jurisdiction can avoid. Delaware’s response goes furthest, proposing to remove the person who could be held accountable.
The competition over accountability
The real issue is not that firms are leaving, but what they are moving toward. In 2025, Texas amended its Business Organisations Code to make directors harder to sue, requiring shareholders to hold three percent of shares before a derivative suit, unlike Delaware. Delaware responded with Senate Bill 21, further limiting the liability of controlling shareholders and conflicted boards, later upheld in Rutledge.The shift extends beyond legislation. In the same period, the court reversed the Tornetta rescission, eliminating a key judicial check on controlling shareholders just as statutory protections eased. Each change is justified as promoting predictability or reducing costs. Cumulatively, they describe a market in which states compete by insulating directors from accountability to those their decisions affect.
This revives a perennial debate. In 1974, Cary warned of a ‘race to the bottom’, led by Delaware; Winter responded that market forces would ensure a ‘race to the top’, as investors would avoid states with inadequate protections. DExit puts that conflict to a fresh test, but with a new twist—Authority is shifting from shareholders to the boardroom, with no effective mechanism yet to control it.
Delaware removes the manager
Delaware has chosen a different path. In July 2026, Secretary of State Charuni Patibanda-Sanchez, alongside John Nay of Norm Ai, introduced the Artificial Intelligence Company (AIC), a corporate form in which an AI agent, not a natural person, conducts the company’s affairs. Under the draft Artificial Intelligence Regulatory Sandbox and Company Act, an AIC could own property, enter contracts, and litigate in its own name, while a single human or corporate member retained a liability shield, provided the entity remains adequately capitalised, and its activities are properly recorded. Where Texas and Nevada have weakened shareholder oversight of managers, Delaware removes the manager from the equation, a move currently unmatched by any competitor.
Its architects present the plan as oversight, not deregulation. Autonomous commerce is already under way, they argue; the real choice is between governing it ‘in daylight’ or surrendering it to anonymous infrastructure beyond any court’s reach. The experiment is accordingly wrapped in the language of control. A thirty-month sandbox, eligibility decided by a committee, and dissolution reserved to the Court of Chancery. The design presumes an autonomous system can be switched off as easily as it was switched on. Whether a court can dissolve an entity with active obligations to suppliers and counterparties is a question the draft leaves the experiment to resolve.
Where the race reaches bottom
Set against DExit, the AIC does not break from regulatory competition but extends it by new means. The reincorporation wave asks whether shareholders can still hold human managers accountable; the AIC asks whether a human manager remains at all. The race reaches bottom once a jurisdiction eliminates the human element altogether, for rivals cannot undercut further and are left only to replicate. Nor need the innovation stay within Delaware’s borders. A liability shield holds only as far as other states recognise it under the internal affairs doctrine, and some commentators already doubt whether courts elsewhere would extend such entities the deference given ordinary corporations.
This proposal touches the core of company law. Since Salomon, the corporation has been distinct from its owners, enabling limited liability. The AIC extends that separation, removing not just ownership from liability, but decision-making from any human agent, and the problem runs deeper than identifying a solvent defendant. Fault-based liability assumes responsibility can be traced to a human mind, and English law attributes a company’s intent and knowledge to those who direct it, as in Lennard’s and Tesco.Meridianloosened the test, but attribution still turns on a human’s state of mind. An entity run by an AI model has no such mind, and liability, as the law traditionally understands it, becomes unattributable. What the incorporation race treats as a competitive feature is, in fact, the foundation for corporate accountability.
A target without a defendant
Supporters argue that incorporation itself addresses this concern—give an autonomous system a legal identity, and responsibility has a defined target, enforceable against the AIC and its assets. The point is credible; however, a legal target is not a liable defendant. Damages may be collected while responsibility vanishes, because the member remains shielded and the AI agent has no guilty mind. The structure provides an entity to sue but no one to blame. Winter assumed markets would punish states with weak protections; the AIC challenges that assumption. If investors are drawn to companies that make responsibility easy to avoid, market discipline runs in reverse, rewarding the very erosion it was meant to prevent.
The incorporation race and the AIC belong to the same trend. The competition among Delaware, Texas, and Nevada is less about efficiency than about making it less likely that any individual answers for corporate conduct, and the AIC is its furthest step, making the removal of the human element a defining feature rather than a coincidence. The real risk is not that one sandbox experiment fails, but that competition has produced a model rivals have every reason to replicate. Where market discipline once corrected such races and now rewards them, the correction must come from outside the race, from whether other states recognise an entity with no human principal, and whether a court dissolving one can still reach the creditors and public left holding decisions no person made or can answer for.
Igho Dabor is a Senior Lecturer in Law at the University of Huddersfield.
Share:
With the support of
