How to keep your company from being corrupted as it grows

Executive overview

Most founders will be ousted from their own company. According to Harvard Law School, only 20% of venture-backed founders are still CEO three years after going public — yet lawyers, VCs, and bankers all tell each founder they will be the exception.

The problem is not personal weakness or greed. It is structural: standard governance documents legally require boards to accept the highest acquisition bid, even from a company that would destroy what you built. Eric Ries calls this financial gravity — the force that no one controls but everyone obeys.

The fix is ethos plus integrity: internal alignment around a real purpose, paired with governance structures that encode that purpose so it survives any management change.

Trustworthiness is the most underrated asset in business — and it must be built into the structure, not just the culture.

Why this happens to almost everyone

• The board of Vectura, a UK inhaler company, accepted a Philip Morris takeover bid because their charter legally required it; the company was written down $900M and dissolved within three years.
• Standard Delaware incorporation language ("any lawful act or activity") defaults to shareholder primacy under today's law — maximise returns, full stop.
• Shareholder primacy is less than 40 years old; before the 1980s, corporations were legally required to declare a beneficial public purpose.
• Founders are ground down in stages: at incorporation, at each funding round, and at IPO prep — each time told it's either too early or too late to add protections.
• Success does not protect you; it makes you a target.

The harder is easier principle

• If you commit to quality, safety, or ethics upfront, you earn trust that lowers customer acquisition costs, increases loyalty, and speeds up internal decision-making.
• Cloudflare gave away SSL encryption for free — hurting short-term conversion — but increased top-of-funnel by an order of magnitude and is now a $70B company.
• Groupon let executives erode its "one email per day" promise through ROI experiments; it eventually sent eight emails a day and destroyed the brand.
• When principles have a real cost, that cost is also a signal: it is a deposit in the culture bank.
• The Todd Park rule: only make deposits, never intentional withdrawals.

Defining your purpose (ethos)

• Write down who you would rather die than betray — customers, employees, users — and make that the explicit, legal purpose of the company.
• "Mission-driven" is usually mission-hopeful: a candy coating on an extractive engine. Real mission drive means you cannot profit by betraying the stated purpose.
• Audit every OKR, bonus target, and performance metric: is there any path to making money by violating a core principle? Close those gaps.
• Google's "Don't Be Evil" was a slogan with no enforcement apparatus; quarterly reporting, by contrast, has an unbelievably expensive apparatus behind it. Seriousness is measured by apparatus, not intention.
• Mission-aligned organisations operate like a flow state: alignment removes the need for meetings, the invisible leader (Mary Parker Follett's term) guides every micro-decision no manager will ever witness.

Governance structures that protect the mission (integrity)

• Public Benefit Corporation (PBC): a two-page filing in Delaware that replaces "any lawful act" with your specific stated purpose. No meaningful downside; all major AI labs are incorporated as PBCs.
• Director's oath: a board-level Hippocratic oath written into the charter — a precondition of serving as a director.
• Mission guardian: an entity whose sole job is keeping the company mission-locked. Options include:
  • Founder control (useful bridge, but fragile and exhausting long-term)
  • Nonprofit foundation as majority owner (Novo Nordisk model — 100+ years, 6× more likely to reach year 50)
  • Perpetual purpose trust / long-term benefit trust (Anthropic model — trustees appointed by AI safety experts with no equity, plus a "purpose protector" who can sue trustees if they deviate)
  • Employee ownership trust or cooperative (John Lewis, Mondragon)
• Anthropic's Long-Term Benefit Trust has no economic dimension — trustees are accountable to the mission, not to returns. This is what enables Anthropic to refuse model releases or turn down a $200M government contract.
• Costco's governance fortress lets it reinvest in customer experience over shareholder payouts; it resists activist pressure because the board sees itself as a bulwark, not an amplifier of financial gravity.

What early-stage founders should do now

• Register as a PBC immediately and write a real mission into the charter. Test it adversarially: can you think of a way to make money while violating this statement? If yes, tighten it.
• Add a director's oath to the charter as a condition of board service.
• Understand founders preferred shares and mission-protective provisions; use them to build board control, then plan to transition to a mission-controlled structure (not just founder-controlled).
• Pledge equity and future revenue to a nonprofit or purpose trust in the charter — even if you do not boot up the entity until Series C.
• If you are not yet a founder: ask any employer whether their mission is in the legal charter. The question alone propagates upward and creates organisational pressure to act.

The Anthropic story

• Dario Amodei was a first-time founder; no top-tier VC wanted in; ChatGPT had not yet been released.
• From inception, Anthropic incorporated as a PBC and wrote the right to enact a long-term benefit trust into every legal document — implemented at Series C.
• Directors on the for-profit board are appointed by and accountable to outside AI safety trustees who hold no equity.
• Anthropic's competitive advantages — inference cost, product velocity, talent density — all trace back to mission alignment attracting the best people and enabling fast, unambiguous decisions.