Four investment professionals. Fifty billion dollars under management. One position worth roughly forty billion.
That is the whole architecture of Vy Capital, the venture firm the Financial Times flagged this week as the fifth-largest disclosed holder of SpaceX equity โ approximately 3.4% of the company, accumulated from a 2016 entry point when the rocket business was valued near $15 billion. The public mark now sits near $1.75 trillion.
Nine years. An enterprise value that compounded roughly a hundredfold. A self-reported internal rate of return of 41% since 2014, with $4.6 billion already distributed back to limited partners. A few dozen employees in total. Four people making the actual allocation decisions. The fund stopped accepting outside capital last year.
Now look at my own dashboard for the past seven days. Nine DeFi lending markets above 90% utilization. Four of them shedding liquidity providers at double-digit rates. A Layer 2 rollup that logged fewer than 400 unique depositors in a month despite $60 million in incentive spend. The macro view reveals what the micro ledger hides: capital is not scarce. It is being compressed into fewer hands, and the compression is accelerating in exactly the place where nobody is watching.
Vy Capital is not a household name. That is a design choice, not an accident. Founded in 2014, the firm runs a deliberately narrow mandate: back Elon Musk's operating companies, and get in early. The disclosed concentration would fail most institutional risk screens โ SpaceX, The Boring Company, Neuralink, plus a $700 million commitment to the 2022 Twitter take-private.
John Hering, who runs the firm, is not a passive allocator. He put more than $100 million into SpaceX in the weeks after the September 2016 Falcon 9 pad explosion, a moment when the company's near-term solvency was a genuine open question. From 2019 onward he worked inside Starlink's early commercial build โ hiring personnel, constructing the unit's financial models. He held a SpaceX employee badge. He now sits on the board of The Boring Company, which recently raised $3 billion with Vy participating.
Read the sequence again. Investor becomes operator. Operator becomes director. Director becomes the largest external shareholder. The governance walls that Sequoia or Andreessen Horowitz maintain for compliance reasons do not exist in this structure. The relationship is the diligence.
The output of that structure is a return profile public markets cannot reproduce. Assets under management moved from $27 billion at the end of last year to $50 billion in June. The firm's letter to investors projects that if its judgments hold, SpaceX will be valued above $10 trillion within five to seven years.
Ten trillion. Hold that number. It is doing more work in this story than any line item on any balance sheet.
The Musk complex is one balance sheet wearing five corporate hats.
SpaceX, Starlink, The Boring Company, Neuralink, xAI, X. On paper these are separate legal entities with separate cap tables and separate investor bases. In practice they share a founder, a talent pipeline, a capital network, and โ critically โ a set of cross-holding investors who sit on multiple boards at once. Vy is the clearest case. It is not the only one.
The $700 million committed to the Twitter take-private is the piece most analysts skip. Twitter had no plausible path to a $10 trillion outcome. That commitment was a liquidity bridge from one Musk entity to another, executed by the same four people who later bought deeper into SpaceX. Follow the direction of capital, not the stated rationale.
I have audited this pattern before. In 2020 I deployed $50,000 of my own capital across Aave and Compound to model what happens when nominally independent lending markets share underlying collateral. The finding was not subtle: correlated collateral produces correlated liquidation. The property that generates yield in a composable system is the same property that generates contagion. DeFi eventually priced what the architecture had always implied โ supply caps, isolation modes, debt ceilings. It took three exploits and roughly $8 billion in losses to get there.
Private capital has not learned it. It does not have to publish the number.
If Starlink's terminal business missed its subscriber trajectory by 30%, what happens to the mark on a $40 billion SpaceX position? What happens to The Boring Company's $3 billion round? To Neuralink's next raise? Nobody knows. None of these marks are marked. They are negotiated, or modeled, or asserted in a quarterly letter to a closed pool of LPs with no exit until an IPO or a secondary sale.
A valuation nobody can redeem is a narrative instrument, not a measurement.
Here is where crypto operators get the story backwards. The prevailing belief inside our industry is that digital assets are the frontier of capital formation โ permissionless, global, continuously priced. That was true in 2017. It stopped being true somewhere around 2021, and the last four years have made it explicit.
I mapped this in early 2024, ahead of the spot Bitcoin ETF approvals. I pulled roughly 10 million on-chain transactions and correlated institutional deposit patterns against realized price stability. The result cut against consensus: ETF inflows functioned as a liquidity sink, not a price driver, over short horizons. Capital arrived, got absorbed, and sat still. Absorption is not discovery. A market can receive billions of dollars and still price nothing, because the incoming capital has no opinion โ it has a mandate. Post-ETF bitcoin stopped functioning as a peer-to-peer payment network and became a line item on a brokerage statement.
Apply the same lens to a $40 billion private position in a $1.75 trillion company.
SpaceX publishes no segment-level financials. Starlink's revenue mix, churn, capital intensity per satellite, ground-station economics โ all of it lives behind a private filing room door. The $10 trillion projection in Vy's letter is not a model output. It is a sales document aimed at the only audience that matters to a firm with $50 billion under management and four decision-makers: the LPs already inside, deciding whether to re-up.
The arithmetic behind $10 trillion is not absurd, and that is precisely what makes it dangerous. Starlink has a plausible path to becoming the physical settlement layer for a world in which machines transact with machines. In 2026 I worked with a decentralized AI agent cluster to architect a zero-knowledge settlement layer for machine-to-machine payments โ 50,000 transactions per second, sub-penny fees, creditworthiness verified without exposing proprietary models. That project pushed me toward a conclusion I did not expect: the demand side of the next cycle is not retail speculation. It is autonomous software paying for bandwidth, compute, and verification.
Every one of those transactions needs a physical layer. A satellite constellation is an excellent physical layer. So the bull case for SpaceX at $10 trillion is not a meme. It is a defensible thesis about who owns the rails of machine commerce.
A defensible thesis and a defensible valuation are different objects. One is an argument. The other requires disclosure.
Now decompose the 41% internal rate of return. Since 2014, $4.6 billion has been distributed to investors against a current asset base of roughly $50 billion. The overwhelming majority of that headline return is unrealized. IRR on a portfolio with no realized exits is an artifact of the mark, not a record of cash. I ran the identical exercise on TerraUSD in 2022 โ separating what a protocol claimed from what it had actually settled โ and the gap is where the entire system lived and died.
To be precise, because the distinction matters: the realized component is small, the claimed component is large, and everything not yet converted to cash is a promise with a discount rate attached.
A few dozen employees managing $50 billion is roughly a 2,000x assets-per-head ratio. It reads as capital efficiency. It also reads as operational fragility โ no independent risk function, no research bench, no compliance layer. In a bear market, the funds that blow up first are the ones with the thinnest middle.
Private marks do not stay private. They propagate. Secondaries reference them. Fund-of-funds NAVs inherit them. Pension allocations and endowment spending rules are calibrated against them. A mark of $40 billion on one name, at one firm, with $50 billion total, is an 80% single-position concentration โ and that concentration sits inside a broader stack of institutions that treat it as a diversified asset class.
The contagion pathway is identical to the one I mapped in the 2020 DeFi stress test. The only difference is latency. On-chain, a correlated liquidation cascade resolves in minutes and everyone sees it. Off-chain, the same cascade resolves over two or three quarters and nobody sees it until the marks get revised.
Watch the direction of capital while all this happens. Dozens of Layer 2 rollups are slicing the same small user base into thinner and thinner fragments, competing for liquidity that does not exist in sufficient quantity to serve any of them. On the other side of the wall, private capital is doing the exact opposite: compressing into a handful of names held by a handful of firms. Two failure modes, opposite signs, one root cause. Capital goes where the incentives point, not where the risk-adjusted return is. In DeFi the incentive was a token subsidy, so liquidity fragmented. In private markets the incentive is access, so liquidity concentrated. Neither outcome was chosen deliberately. Both were produced by the structure.
The same logic explains why Aave and Compound's interest rate curves have drifted so far from anything resembling market-clearing prices. When the benchmark for risk-free yield is itself a governance parameter, the resulting rate is a policy output. Private marks work the same way. The mark is a policy output. It moves when the firm decides it moves.
Here is the blind spot the FT story buries.
The story everyone will tell is that a concentrated, relationship-driven fund beat the market. The story that matters is that a single unreported mark now anchors a material share of global private capital allocation.
Forty billion on one name. Fifty billion under management. By every disclosed signal, the other positions are smaller by orders of magnitude. When an entire portfolio depends on one company's continued ability to raise at ascending marks, the fund is not diversified. It is a levered expression of a single narrative, wrapped in a limited partnership agreement.
I have run this post-mortem. In May 2022 I spent four weeks reverse-engineering the TerraUSD death spiral and quantified the drain rate. The protocol's reserve funds could not cover even 1% of redemptions during a high-volatility window. The mechanism was not exotic. It was a structure whose solvency depended on the continuation of its own promotional loop. Anchoring was the product. Reserves were the marketing.
Nobody is claiming SpaceX is Terra. The satellites are in orbit. The revenue is real. But the funding mechanism carries a structural resemblance worth naming: a valuation whose integrity depends on the next round confirming the last round. That is true of every private mark ever recorded. It is more true when the mark is enormous, the holder is opaque, and the underlying cash flows have never been independently audited at segment level.
The crypto market, for all its ugliness, remains the only large asset class where this failure mode gets detected in hours instead of quarters. On-chain transparency is brutal, inefficient, and permanently embarrassing. It is also the reason my 2020 liquidity warning landed three months before the first major exploits, and the reason a 40-page post-mortem on an algorithmic stablecoin ended up cited by three regulatory bodies. Code does not lie, but it often obscures intent โ and markets with no code obscure both.
Watch three things. Whether SpaceX ever publishes segment-level Starlink economics โ the integrity of the mark depends on it, and no IPO proceeds without it. Whether secondary market depth in SpaceX shares widens or narrows, because a spread is the earliest honest price available to anyone outside the cap table. And whether Vy's $50 billion becomes diversification or stays an 80% bet on one founder's execution.
The narrower question belongs to the rest of us. If the most concentrated, least transparent capital structure in modern finance is now underwriting the physical rails of machine commerce, what exactly is our transparency for โ if not to be the counterweight?