Cyrus Reed: Iris, hey — did you catch the Anthropic news this week, because I've been stuck on something and I can't shake it.
Iris Holm: The IPO filing stuff? Which part.
Cyrus Reed: Picture the Anthropic press team — and I want you to actually picture this — they hit publish on two announcements in basically the same news cycle. One says the company lost more than eighty billion dollars in the first half of 2026. The Wall Street Journal has that number. And the other one says — wait, no, the other one says they're committing five percent of their shares to a brand new Public Benefit Fund.
Iris Holm: At the same time?
Cyrus Reed: October 18, 2026. Same cycle. And the IPO — Al Jazeera was still calling it 'looming' as of October 21st, so we're not even post-IPO yet, this is all pre-public.
Iris Holm: So they're pledging equity in a company that hasn't gone public, while disclosing losses that most companies never survive. That's — the contradiction is structural.
Cyrus Reed: And Anthropic is a public benefit corporation, which means there's a stated mandate here, not just a gesture — but I keep turning over the same question: if you're burning eighty billion in six months, when does that five percent ever have real value to give away?
Iris Holm: That's exactly what we need to work out.
Cyrus Reed: But that's — okay, the five percent, I keep trying to picture it as something real. Like, what even is that?
Iris Holm: Someone promises you five percent of their house someday. But they're currently spending twice what they earn. And whether that ever grows to twenty percent — entirely their call.
Cyrus Reed: Wait — the twenty percent, that's not a floor, that's a — what is that exactly?
Iris Holm: Ceiling. The scaling mechanism could bring total equity contributions to around twenty percent as valuation milestones are hit. But the board retains discretion at every step. So it's not a commitment to reach twenty — it's permission to consider reaching twenty.
Cyrus Reed: So both pledges — the Public Benefit Fund and the separate thing where Dario Amodei and the co-founders personally pledged to give away eighty percent of their own equity — both of those depend on a business model that's currently losing eighty billion in six months?
Iris Holm: That's what the Wall Street Journal reported, and yeah — the math is genuinely unresolved. Claude is the commercial product the whole valuation rests on. And the IPO timing? As of late September the debut could occur as soon as late September. That window passed. Still looming.
Cyrus Reed: So they're not committing wealth — they're committing the speculation that wealth eventually shows up.
Iris Holm: Frankly, that's the cleanest way to say it. The pledge is real as a legal structure. The value behind it isn't.
Cyrus Reed: And that's where the Founder LLC comes in — because the speculation, the whole structure, gets locked in before any public shareholders can say a word about it. The Information broke the story first, Reuters confirmed it September 29th. Dario Amodei and the co-founders get extra-powerful voting rights through this LLC vehicle, and that class doesn't even start to sunset until two or fewer founders or successors are left.
Iris Holm: But they're a public benefit corporation. Doesn't that require some accountability to — I don't know, the public?
Cyrus Reed: That's exactly — wait, that's the paradox, right? Yes, the public benefit corporation designation requires pursuing social good. But the Founder LLC means the people operationally defining what 'social good' means are insulated from the shareholders who'd otherwise push back. Anthropic's own IPO filing actually says the governance structure could lead to decisions that conflict with shareholders' short-, medium-, or long-term financial interests. They wrote that in.
Iris Holm: They disclosed it. That's — the disclosure doesn't fix it.
Cyrus Reed: No, it doesn't. And Sam Altman at OpenAI is doing a version of this too, right — the Public Wealth Fund proposal, distributing AI profits to U.S. citizens, outlined in a 2026 policy document. But that's also still a proposal. Not a charter, not enforceable. Both companies are citing the Effective Altruism movement as the philosophical backbone here, but EA prescriptions haven't been — I mean, nobody has translated them into something with actual teeth.
Iris Holm: The framework is real. The enforcement isn't.
Cyrus Reed: That's the gap. And here's — actually, think about what that looks like from the outside. A shareholder at IPO reads the prospectus, sees the public benefit mandate, feels good about it. Has no lever to pull if Dario Amodei decides public benefit means something narrow, or shifts priorities entirely. The provision about removing co-founders who quit or sell too many shares? That's internal discipline. That's not external accountability.
Iris Holm: Right. They've built a structure that lets them claim the mission while removing the mechanism to enforce it.
Cyrus Reed: And whether this is new — that's a question we should sit with, because there's a whole history of voluntary pledges that looked like this and didn't deliver proportional redistribution. That one lands harder once you see the pattern.
Iris Holm: The pattern is what breaks the clean version. Picture someone — a retiree in Akron, reads on October 18th that Anthropic is pledging equity to the public. Spends an afternoon trying to find the charter. The eligibility criteria. The governing body. There isn't one. No document tells her whether she qualifies, who decides, or when.
Cyrus Reed: Wait — there's genuinely no governing body for the Public Benefit Fund yet?
Iris Holm: Not a public one. Unresolved questions include who's eligible, how transferred equity gets taxed, whether corporate boards or independent institutions control the resulting wealth. None of that is settled. The fund exists as a legal entity. The mechanics don't.
Cyrus Reed: Okay but — so that's the Giving Pledge problem, right? Like, that's exactly what happened there. Billionaires commit to give half their wealth away, they define where it goes, deployment ends up slow and founder-controlled, sometimes unrealized entirely. We've watched that play out.
Iris Holm: Exactly the analogy. Sovereign wealth funds actually work — Norway's oil fund works — because there's a governing structure external to the people who created the wealth. Anthropic's structure inverts that. Dario Amodei's team writes the definition of public benefit. Circular.
Cyrus Reed: So they're — wait, are they writing the rulebook before regulators can even show up?
Iris Holm: That's the unresolved question. They deliberately skipped image and video generation — framed it as safety restraint, mission purity. But that self-imposed restraint is also what they're offering as a substitute for external regulation. Proactive responsibility, or regulatory capture in advance. I genuinely can't tell you which, and I don't think the sourcing here is solid enough to assert it.
Cyrus Reed: And the UBI speculation — all that online commentary about direct cash transfers — Anthropic hasn't confirmed any of that, the public benefit corporation statute doesn't require it, and yet that's the image people are carrying around.
Iris Holm: The woman in Akron finds no mechanism. Just a pledge. That's what voluntary looks like from the outside — and voluntary without enforcement is the only accountability tool currently on offer.
Cyrus Reed: And that's — there's this image I can't shake. There's a future Anthropic shareholder meeting, right, somewhere after the IPO, and someone stands up and asks why equity that was reserved for the Public Benefit Fund wasn't used to — I don't know, offset some fraction of an eighty-billion-dollar loss. And Dario Amodei is in that room. The Founder LLC is in that room. And they get to answer that question on their own terms. No external mechanism. Just them.
Iris Holm: And we still won't know — after that meeting — whether the pledge was ever a commitment or just a bet they made when the math looked different.
Cyrus Reed: Yeah. That's — that's actually where I land. Not angry about it, just... not sure.
