Two figures inside the same document refuse to reconcile. A reported holding of $40 billion, taken at a stated 3.4% of the target company, implies that company carries a valuation near $1.18 trillion. Eleven paragraphs later, the same document describes that company going public this year at roughly $1.75 trillion. The spread between those two numbers runs to about $570 billion — more than the annual output of all but a handful of national economies. Both figures appear in a report that is otherwise confident, structured, and dense with specifics.
The company is SpaceX. The holder is Vy Capital, a Dubai-based venture and asset-management firm. And the more dramatic of the two claims — that SpaceX held a public listing at a $1.75 trillion valuation — contradicts the baseline fact that SpaceX has never gone public and has repeatedly signaled no near-term intention of doing so. This is not a rounding error or a currency mismatch. One of the two load-bearing facts is wrong. When that happens in a document people are meant to trust, the correct response is the one I learned at seventeen, hunched over a Solidity codebase at two in the morning: stop reading the narrative, start checking the arithmetic. Trust no one, verify the proof, sign the block.
That instinct is why this story belongs on a crypto desk, even though the entity at its center has no on-chain footprint whatsoever. Vy Capital runs no protocol, issues no token, and touches no blockchain. It is a private-market allocator holding equity in rockets, satellites, and brain-computer interfaces. But the entire institutional crypto narrative for 2026 — real-world asset tokenization, tokenized private equity, on-chain net asset value attestation — is built on the assumption that a firm like this can be pulled onto a ledger and made verifiable. This case is the perfect stress test. It shows exactly where verification breaks.
Context matters here. Vy Capital was founded in 2013 by Alexander Tamas, a former Goldman Sachs banker, and operates from Dubai's financial district. The report describes a firm managing roughly $50 billion in assets with a core team of about four people, up from $27 billion. It reports a cumulative 41% total IRR and $4.6 billion in cumulative distributions. It has reportedly stopped accepting external capital, which converts it from a fee-driven asset manager into something closer to a permanent-capital vehicle. Its portfolio, per the document, includes SpaceX, the Boring Company, Neuralink, and stakes connected to X. The firm maintains an extraordinarily low public profile: no penalties, no regulatory actions, almost no disclosure.
In the tokenization world, this is precisely the profile that advocates want on-chain. BlackRock's BUIDL fund already demonstrated the machinery — permissioned entry, KYC-gated transfers, a compliant wrapper around a treasury product. The pitch for extending it to private equity is straightforward: private equity is illiquid and opaque; put it on a ledger and you unlock transparency, fractionalization, and secondary liquidity. In 2024 I traced one thousand transactions through BUIDL's on-chain settlement layer to map its permissioned entry mechanisms for a technical breakdown. What I found then is directly relevant to what follows. The compliance layer, not the settlement layer, was the actual product. And the compliance layer is exactly where a case like Vy Capital resists verification.
The Arithmetic Does Not Compile
Start with the data integrity problem, because everything else depends on it. The report gives three valuation anchors, and only one of them survives scrutiny.
The 2016 anchor — that Vy invested when SpaceX was valued around $15 billion — is broadly consistent with the public record for that period, which put SpaceX in the low-teens billions. That number checks out. It has internal logic, external corroboration, and a plausible timeline.
The other two anchors fail. A $40 billion stake at 3.4% ownership implies a $1.18 trillion enterprise value. A separate claim of a $1.75 trillion IPO implies that Vy's stake, at the same 3.4%, would be worth roughly $59.5 billion, not $40 billion. The document contradicts itself on its own central holding. If the position is worth $40 billion, the company is worth $1.18 trillion. If the company is worth $1.75 trillion, the position is worth $59.5 billion. Both statements cannot be true.
This is the exact pattern I found in 2017, auditing the Solidity implementation of Golem's token distribution. The whitepaper promised a coherent economy. The contract had three integer overflow vulnerabilities that would have broken the distribution logic entirely. The marketing and the mechanism lived in different universes. I submitted patches through GitHub before mainnet launch, and the episode permanently reordered my priorities: code and arithmetic first, team narrative never. The Vy Capital document reproduces that disconnect in prose. A confident overview, a precise-sounding percentage, and underneath it, numbers that do not compile.
The IPO claim compounds the problem. If the report's most dramatic assertion — a public listing at $1.75 trillion — is false, then the credibility of every derivative figure collapses with it. The 41% IRR. The $50 billion AUM. The $4.6 billion in distributions. These are not independently verifiable. They are claims from a firm that, by its own description, discloses almost nothing. When one load-bearing fact fails, the reasonable posture is to discount the entire dataset until each line is reconciled. I have seen what happens when analysts skip that step. In 2022, after Terra/Luna, I performed forensic reviews of twelve failed DeFi protocols and documented fifteen distinct security misconfigurations that led to exploits. Almost every one traced back to a team that trusted its own narrative over its own reconciliation.
The lesson is not that Vy Capital is fraudulent. It is that the document's provenance is suspect, and the crypto industry's tokenization push treats documents exactly like this as settled inputs. An oracle that ingests unverified NAV is not an oracle. It is a rumor with a timestamp.
The Compliance Stack Is the Real Architecture
Assume, for argument's sake, that the holdings are real. The next question for anyone proposing to tokenize them is structural: what actually sits between Vy Capital and its SpaceX shares?
The answer, for any cross-border allocator, is a stack of intermediaries. A Dubai-based manager. A Cayman-exempt limited partnership. Likely a United States special-purpose vehicle. Possibly layered offshore structures. The report explicitly notes that Vy appears to have stopped external fundraising, which reduces its obligations under collective-investment regulations — and, if anything, increases its freedom to remain opaque rather than decrease it.
That opacity is not incidental. It is load-bearing, and it collides directly with United States national-security review.
SpaceX sits inside the defense and space industrial base. Its launch technology and Starlink constellation touch ITAR export controls and fall under the jurisdiction of CFIUS, the Committee on Foreign Investment in the United States. Foreign ownership of sensitive defense-adjacent assets is subject to review, and the report notes that Vy participates in Starlink business development — a fact that pushes it toward, not away from, the ITAR technical-export boundary. A foreign entity holding equity in a company of this profile faces structural scrutiny that no tokenization layer can abstract away.
The probable workaround is the standard one: non-voting equity, a United States SPV, multi-layer offshore wrappers designed to keep the ultimate beneficial owner below the CFIUS threshold. I have audited structures like this before. In 2025, examining the oracle systems behind Fetch.ai's AI agent payments, I found a latency vulnerability in their off-chain computation verification and proposed a zero-knowledge proof integration to restore trustlessness. The structural insight generalizes. Every layer added to hide jurisdiction is a layer added to the attestation surface. You cannot tokenize a claim that has been deliberately engineered to defeat tracing, because the token's value proposition is that the claim is traceable.
Then there is the data-privacy dimension, and it is underappreciated. Vy's portfolio reportedly includes Neuralink, which operates in brain-computer interfaces — a category that generates neural data, the most sensitive class of personal information that exists. Emerging United States state privacy laws, and future federal neural-data legislation, will reshape Neuralink's compliance obligations and therefore its valuation, and therefore Vy's mark on that position. A tokenized share of that exposure would inherit a compliance liability that does not yet have a settled legal definition.
The AML and CFT picture completes the stack. Any manager must run know-your-customer and source-of-funds checks on its limited partners. A capital chain running from the Middle East, through Cayman and Delaware vehicles, into sensitive American technology increases the difficulty of beneficial-ownership tracing — the exact difficulty that sanctions and foreign-investment regimes are designed to overcome. None of this is disqualifying. All of it is unverifiable from the outside.
What Tokenization Actually Requires
Strip the marketing away and on-chain private equity requires four primitives: a custody model, a valuation feed, a transfer-restriction engine, and a redemption mechanism. Measure Vy Capital against each, and the gap becomes precise.
Custody. On-chain, custody is a key. Off-chain, custody of a SpaceX position is a chain of paper: SPV agreements, transfer restrictions, rights of first refusal, board consents. The number of parties who must sign off before a beneficial interest moves is unknown and probably large. A token cannot represent an interest that its holder cannot legally transfer.
Valuation. This is the weakest link. SpaceX has no continuous public market price. Its value is set by periodic primary rounds and by secondary transactions, meaning any mark is a point-in-time estimate, often stale by months. An on-chain NAV attestation would therefore be a periodic human signature dressed as an automated feed. In my 2020 stress test of Compound Finance, I modeled liquidation thresholds for five hundred distinct user portfolios under volatility and published a report predicting the September yield collapse. That model worked because the inputs — interest rates, collateral prices — were continuous and observable. Private-equity NAV is neither. You cannot build a liquidation engine on a price that updates twice a year.
Transfer restrictions. SpaceX shares are almost certainly subject to transfer restrictions that exist precisely to control who can hold them. This is not a bug to be routed around; it is the feature that keeps the cap table clean for CFIUS and ITAR purposes. A permissioned token could encode the restriction, but encoding the restriction means encoding the same opacity the restriction was built to protect. The chain would faithfully record a wall.
Redemption. On-chain redemption assumes atomic settlement. Off-chain redemption of a private interest assumes lawyers, timelines, and gatekeepers. The mismatch is total. The token settles in seconds; the underlying moves in quarters, if at all.
Each of these four primitives is a place where the "tokenize it for transparency" claim inverts. The token does not make the underlying transparent. It makes the opaque underlying look transparent, which is worse.
The Four-Person Oracle
Now consider governance, because it determines how much any of this can be trusted even if the structures were clean.
The report describes a firm managing roughly $50 billion with a core team of about four people. That is extraordinary human productivity by any measure. It is also a single point of failure with no redundancy.
In distributed-systems terms, this is a cluster with no consensus mechanism and one signing key. Every investment decision, every mark, every attestation flows through a handful of individuals using subjective judgment. There is no on-chain governance, no multi-signature quorum, no independent validator set — and there cannot be, because the underlying assets are not native to any chain. The "consensus" is a human one, enforced by relationships and reputation rather than by cryptography.
This matters for tokenization in a way advocates rarely confront. A blockchain's guarantee is that the state transition is verifiable by anyone. A four-person firm's guarantee is that the state transition is verifiable by no one outside the four. When you tokenize the firm's holdings, you do not import the blockchain's guarantee. You import the firm's. The chain settles; the oracle decides. And here the oracle is four people, sitting in Dubai, holding an unverifiable cap table.
The key-person risk compounds the attribution problem. If the reported returns are driven by access to one founder's network — access to Musk-affiliated companies, to primary rounds closed to outsiders — then the "track record" is not a repeatable process. It is a relationship. Relationships do not port to smart contracts. They do not fractionalize. They cannot be wrapped in a compliant token and sold to a secondary buyer, because what the buyer would be purchasing is proximity, and proximity does not transfer.
Attribution and the 41%
Which brings the analysis to the return figure itself, and to a statistical trap the crypto industry keeps falling into.
The report cites a 41% total IRR across funds that grew AUM from $27 billion to $50 billion. Impressive on its face. But an IRR is only as honest as its attribution, and attribution requires seeing the whole portfolio — which nobody outside the firm can do.
Consider the concentration. If a single position — SpaceX — carries the portfolio, and if that position was marked up on the back of primary rounds rather than realized liquidity, then the headline return reflects unrealized paper gains concentrated in one asset. Strip that line item out and ask what the remaining portfolio returned. The report itself flags the possibility that the residual book is unremarkable. That is survivorship masquerading as skill: the winner is visible, the base rate is not.
I learned to distrust unattributed headline numbers during the 2020 DeFi summer, when yield figures were quoted without their underlying risk. My Compound stress test was an exercise in refusing to take a headline APR at face value and instead reconstructing what it implied about liquidation risk under stress. The same discipline applies here. An IRR without attribution is a whitepaper number. It looks precise and proves nothing.
None of this requires alleging misconduct. It requires only the skepticism that any auditor would apply: show me the losing positions, show me the marks, show me the realized versus unrealized split, and show me the reconciliation between the $40 billion stake and the $1.18 trillion implied valuation — or the $59.5 billion stake and the $1.75 trillion implied valuation. Until the document reconciles with itself, the returns are unreconciled too.
The Blind Spot: Opacity Is the Product
Here is the contrarian reading, and it is one the tokenization narrative refuses to accept.
The prevailing assumption in RWA circles is that private markets are opaque because they are old-fashioned, and that blockchains will fix the opacity by default. That gets the causality backwards. The opacity is not a defect of the structure. It is the structure's reason for existing.
Offshore SPVs, non-voting equity, multi-layer wrappers, and closed fundraising do not exist to inconvenience analysts. They exist to satisfy competing constraints at once: to keep foreign ownership below CFIUS thresholds, to respect ITAR controls on sensitive technology, to manage tax exposure across jurisdictions, and to protect a cap table from unwanted parties. Every wrapper is a solution to a real legal problem. Tokenizing the position does not dissolve those problems. It stacks a ledger on top of them and calls the combination "transparency."
That is the blind spot. Tokenization advocates measure success by whether an asset appears on-chain. The real measure is whether the claim behind it can be independently verified — and for a position engineered around non-disclosure, the answer is no. The chain would display a clean, compliant, fractionally-sold token representing an interest that no outside party can confirm exists, is unencumbered, or is correctly priced. That is not transparency. It is opacity with a settlement layer, which is more dangerous precisely because it looks trustworthy.
The 2024 BUIDL analysis taught me the same thing from the opposite direction. BUIDL works because the underlying is a treasury product with continuous, observable pricing and a simple compliance wrapper. Push the same machinery onto a private, defense-adjacent, cross-border equity stake and the compliance wrapper must hide more than it reveals. The middleware is not the obstacle to adoption. The middleware is the product, and the product is designed not to be verified.
The Oracle Is the New Single Point of Failure
The forward-looking risk is not the blockchain. It is what feeds it.
As institutions push tokenized private markets, the attestation layer becomes the critical dependency — and it is a human layer wearing technical clothing. A signature from a four-person firm. A stale NAV from a primary round. A cap table no regulator can penetrate because the structure was built to prevent exactly that. The chain will faithfully record all of it, immutably, forever.
The vulnerability forecast is uncomfortable. The next generation of tokenized-asset failures will not come from broken consensus or reentrancy bugs. They will come from faithful, immutable, cryptographically-secured records of claims that were never true in the first place. The ledger will be perfect. The truth behind it will not be. And when the two diverge, the industry will discover that it spent a decade hardening the settlement layer while leaving the input layer — the oracle, the attestation, the human signature — exactly as fragile as it was in 2017, when a whitepaper promised a coherent economy and the contract underneath it would not compile.
