Chamath Palihapitiya published something last week that should make every credit union board member sit up straight. Not because it’s about credit unions — it isn’t. But because it describes a world where the financial architecture most institutions depend on breaks down, and then, almost by accident, describes why yours doesn’t.
His argument: AI lowers the cost of disruption so fast that no company can credibly project its cash flows beyond five years. If that’s true, then 60 to 80 percent of the equity value of every public company — the part that lives in years ten and beyond, the so-called “terminal value” — evaporates. By his math, the S&P 500 reprices from $58 trillion to roughly $14 trillion. A 75 percent drawdown. Worse than 2008 by multiples.
It’s a thought experiment. But the math is clean, the historical precedents are real, and 1,500 people bookmarked it within hours. It’s worth taking seriously — not because the exact scenario plays out, but because the mechanism he identifies is already in motion.
The Paradox Nobody’s Talking About
The part of Chamath’s essay that stuck with me wasn’t the market math. It was the paradox buried near the end.
The companies driving AI disruption — Microsoft, Google, Amazon, Meta — are committing what Goldman Sachs estimates at $300 to $500 billion per year to AI infrastructure. Data centers. Custom chips. Cooling systems. Power plants. These are 10 to 15-year bets. You don’t build a $10 billion data center expecting to recoup that investment in three years.
But here’s the problem: if AI itself makes competitive advantages temporary — if the very thing these data centers enable is a world where no business can project cash flows past year five — then the capital expenditure that funds AI becomes irrational. Why invest $10 billion over fifteen years if AI ensures that no one, including you, can defend a market position for fifteen years?
The disruption engine destroys the financing mechanism for disruption.
Chamath calls this “probably self-defeating.” I think he’s underselling it. This isn’t a footnote. It’s the central tension of the next decade of technology investment. And the resolution tells you more about who wins than any analysis of which model is better than which.
How Terminal Value Actually Works (And Why It Matters to You)
I want to make this concrete, because the implications for credit union technology strategy are direct.
When a public company gets valued, analysts project its free cash flow and discount those future dollars back to today. The first five years are grounded in operational data. Everything past year ten collapses into a single number called “terminal value” — a mathematical bet that the company will keep generating cash indefinitely.
For most public companies, terminal value represents 60 to 80 percent of the stock price. When you buy a share of Microsoft at 35x earnings, you’re not paying for what Microsoft earns today. You’re paying for the assumption that Microsoft will still be earning and growing in 2040.
Chamath’s argument is that AI makes that assumption untenable. If AI-driven disruption gives any company a 20 percent annual probability of obsolescence — not unreasonable in fast-moving sectors — then the expected lifespan drops to five years. Discount a five-year cash flow stream at a 9 percent cost of equity and you get roughly 4x free cash flow. Not 35x. Not 22x. Four.
That repricing destroys wealth on a scale that makes 2008 look like a correction.
But here’s what matters for this conversation: that repricing mechanism doesn’t apply to you.
The Structural Immunity of Cooperative Capital
Credit unions don’t have terminal value in the equity-pricing sense. There is no stock to reprice. No public market applying duration discounts. No activist investor demanding you stop investing in year-seven capabilities because the market won’t pay for them.
This isn’t a minor technicality. It’s a structural advantage that becomes enormous in exactly the scenario Chamath describes.
When Chamath says private capital retreats to short-duration assets — 2 to 7x free cash flow, cash-on-cash returns, nothing past the five-year horizon — he’s describing a world where most institutions lose the ability to make long-term technology investments. R&D budgets get slashed. Infrastructure projects get shelved. Every CFO optimizes for the next eighteen months because that’s all the market will pay for.
Your board doesn’t operate under that constraint.
A credit union board can authorize a five-year technology investment — a data infrastructure build, an agent orchestration platform, a compliance automation pipeline — because the people who own the institution aren’t pricing terminal value. They’re pricing current utility. “Does my credit union serve me well today, and do I trust it to serve me well tomorrow?” That’s a fundamentally different question than “What is the net present value of this company’s free cash flow in year twelve?”
The members who fund your institution through deposits aren’t running discounted cash flow models. They’re asking whether the loan rate is fair, the mobile app works, and the person at the branch knows their name. That relationship doesn’t reprice when an AI startup disrupts a SaaS company in San Francisco.
Sovereign Wealth Funds — And Their Community Equivalent
Chamath’s proposed resolution is that nation-states fill the void. Countries with sovereign wealth funds — the US, China, the Gulf states, Norway, Singapore — step in to finance the long-duration infrastructure that private capital abandons. State capitalism, long derided by market orthodoxy, gets vindicated at the precise moment market capitalism’s pricing model breaks down.
He’s right about the mechanism. But he’s thinking exclusively at the macro level — trillion-dollar sovereign funds, state-backed industrial policy. He misses that the same structural pattern already exists at community scale, embedded in an ownership model that predates AI by a century.
Credit unions are, structurally, the sovereign wealth funds of community finance.
I don’t say that loosely. It’s a description of the capital structure, not a metaphor. Credit unions are member-owned, mission-driven, and insulated from public market repricing. They can make 10-year bets because their capital base isn’t subject to quarterly mark-to-market. They can invest in infrastructure that takes years to mature because their “shareholders” measure returns in service quality, not stock price appreciation.
According to NCUA data, there are 4,539 federally insured credit unions in the United States holding $2.3 trillion in assets. That’s patient capital at a scale that most people — including Chamath — overlook because it doesn’t show up in hedge fund portfolios or venture capital pitch decks.
What This Means for Your AI Strategy
If Chamath’s scenario partially materializes — even a 30 to 40 percent compression in terminal values, not the full 75 percent — the practical consequences for technology investment are immediate.
Your vendors’ runway shortens. AI startups funded by venture capital face exactly the repricing Chamath describes. A pre-revenue company valued at $1 billion on the promise of future market dominance? That valuation collapses when investors stop paying for terminal value. The vendor you signed a three-year contract with may not exist in three years — not because their product is bad, but because the financing model that sustains them breaks.
I’ve watched this cycle before. At Flowroute, we built voice infrastructure in a market littered with well-funded competitors who burned through capital chasing growth. Many of them are gone. The companies that survived owned infrastructure, not interfaces. The same selection pressure is coming for AI vendors, and it’ll be more severe because the AI market is moving faster than telecom ever did.
Long-duration technology investments become rarer — and more valuable. If most companies retreat to 18-month planning horizons, the institutions still capable of thinking in 5-year arcs gain a compounding advantage. The infrastructure they build while competitors optimize quarterly gets deeper, more proprietary, and harder to replicate with each passing year.
This is the credit union opportunity. Not because credit unions are technology leaders — most aren’t, yet. But because they have the capital structure to make the bets that banks increasingly cannot. A publicly traded regional bank repricing to 5x free cash flow will slash its technology budget. Your board, not subject to that repricing, doesn’t have to.
And yet — most credit union boards are squandering this advantage. They’re running AI procurement the same way regional banks do: annual vendor reviews, 18-month RFP cycles, pilot programs that never graduate to production. They’re voluntarily adopting the short-termism of institutions that have no choice. A board that approves a one-year chatbot contract when it could be funding a five-year data infrastructure build is surrendering its single greatest structural advantage — the ability to invest patiently — in exchange for a demo that looks good in a committee presentation. That’s not prudent governance. It’s imitation of a model that Chamath just explained is breaking.
The context layer becomes the asset that matters. In a world where AI capabilities are commoditized and competitive advantages compress, the only durable differentiator is institutional knowledge — the patterns, relationships, and operational context that take years to build and can’t be replicated by a funded competitor with a Claude subscription.
I wrote about this in Article 17 — the Company Context Layer. The five layers of institutional knowledge, from written SOPs to undocumented operational patterns to examiner relationships. That architecture isn’t just a technology strategy. In Chamath’s world, it’s the primary store of institutional value. The thing that survives when everything else reprices.
The Capex Paradox Has a Winner
Chamath frames his essay as a thought exercise, and it’s a good one — the kind that forces you to stress-test assumptions you didn’t know you were making. But thought exercises have a way of becoming investment theses, and investment theses have a way of becoming reality.
The scenario he describes — compressed terminal values, shortened planning horizons, capital retreating from long-duration bets — doesn’t require a 75 percent market crash to matter. Even a partial move in this direction changes the competitive landscape for financial institutions. It rewards patient capital. It punishes short-termism. It advantages institutions whose owners measure outcomes in decades, not quarters.
That’s not a description of Wall Street. It’s a description of the cooperative model.
At Concreit, I watched this play out in miniature. When the 2022 rate shock hit and real estate crowdfunding platforms were imploding — several shut down entirely, others froze redemptions — our member-aligned capital structure held. Not because we were smarter. Because our investors were patient and our model didn’t depend on terminal value assumptions that evaporated overnight. The speculative platforms needed continuous capital inflows to survive. We needed members who trusted the structure. The pattern is the same every time: when the environment gets hostile, patient capital wins.
Credit unions have been patient capital for over a century. The AI era doesn’t change that. If anything, Chamath just made the case for why it matters more than ever.
This is Article 33 of the Runline Insights series. In the next article, we’ll explore the other side of the equation — when AI commoditizes every capability, what exactly survives? The answer isn’t a technology. It’s what your institution knows that nobody else does.
Previously: Article 28 — Seven Moats. Most Credit Unions Have Two | Article 17 — The Company Context Layer
