Buying Amazon to own Anthropic gets you a specific number of dollars per $10,000 — and it is smaller than you think. How to size a proxy stake, and how to read a fund's premium or discount to NAV before you pay it.
Suppose Lessons 2 and 3 have done their job: you have a sourced valuation with an honest error band, and you have decided you want exposure. The next question is the one that almost never gets asked out loud — how much of the private company do you actually get?
There is a simple way to answer it. Take the value of the stake the public company discloses, divide by the public company's market value, and multiply by your investment. The result is dollars of the private company per $10,000 you put in.
Across the current map — 20 covered companies, of which 17 have any public route at all — that arithmetic produces numbers most people find deflating.
| Public company | Holds | Disclosed as | Per $10,000 |
|---|---|---|---|
AMZN | ANTHROPIC | $122.3 billion | $454 |
GOOGL | $124.3 billion | $296 | |
MSFT | OpenAI | $11.9 billion | $31 |
AMZN | ANTHROPIC | $8.0 billion | $30 |
Read the right-hand column as a fraction of $10,000. Even the largest entries leave the overwhelming majority of your money invested in something else entirely.
This is why "buy X to get exposure to Y" is usually bad advice rather than a clever trick. If you would not own the proxy on its own merits, the sliver of Y inside it will not rescue the position — and if you would own it anyway, the Y is a rounding error in your thesis.
Registered funds are the one route with no accreditation gate and meaningful concentration. The catch is the price relative to the fund's own stated net asset value.
| Fund | Covered holdings | Per $10,000 | Premium / discount |
|---|---|---|---|
VCX FUNDRISE INNOVATION FUND, LLC · as of 2026-06-30 | 6 | $874 | -20.83% |
DXYZ Destiny Tech100 Inc. · as of 2026-06-30 | 7 | $1,928 | -10.73% |
BST BlackRock Science & Technology Trust · as of 2026-06-30 | 7 | $1,683 | -10.39% |
A negative figure is a discount: the market values the portfolio below what the manager says it is worth. Given what Lesson 3 measured about how marks behave, a discount is not automatically a bargain — it can be the market pricing in a markdown that has not been filed yet.
The exposure column and the premium column work against each other. A fund with heavy covered exposure and a wide premium can deliver less value per dollar than one with modest exposure at a discount. Read both, never one.
A premium is only meaningful against a NAV the fund itself stated, recently, in a machine-readable filing. Where that does not exist, the entry says so:
| Fund | Why no premium is shown |
|---|---|
| BOT | the last filed NAV is 209 days older than the price, past the 120-day limit — the difference would measure the gap between the two dates more than any premium |
| RVI | the fund has not tagged a net asset value per share in XBRL. Several of these funds report NAV only inside a combined shareholder report covering three trusts at once, and picking the wrong column would publish another fund’s NAV as this one’s — so the figure is left to the reader and the filing is linked |
| ARKVX | not applicable — the fund is bought and sold at NAV, not on an exchange |
| BSTZ | the fund has not tagged a net asset value per share in XBRL. Several of these funds report NAV only inside a combined shareholder report covering three trusts at once, and picking the wrong column would publish another fund’s NAV as this one’s — so the figure is left to the reader and the filing is linked |
| BTX | the fund has not tagged a net asset value per share in XBRL. Several of these funds report NAV only inside a combined shareholder report covering three trusts at once, and picking the wrong column would publish another fund’s NAV as this one’s — so the figure is left to the reader and the filing is linked |
| PWRL | the last filed NAV is 178 days older than the price, past the 120-day limit — the difference would measure the gap between the two dates more than any premium |
28 funds are withheld from the map altogether for the same class of reason. A blank with a stated cause is a finding; a plausible-looking estimate in its place would be a fabrication.
3 of the 20 covered companies have no measurable public route: no public company discloses a stake, and no fund we can price holds them.
This list is the most practically useful part of the whole map. If a company is on it, then for an unaccredited investor there is no position to take — and any product offering you one deserves the full weight of Lesson 1's warnings. Knowing that saves more money than a good valuation does.
Full map, including every disclosure we read and every fund we priced: /private-exposure.
You buy $10,000 of a large public company because it owns a stake in a private company you want. The disclosure implies about $400 of that stake per $10,000 invested. What have you actually bought?
A closed-end fund holding private positions trades at a 20% discount to its own stated net asset value. What is the most complete reading?
For several funds we refuse to publish a premium or discount at all. Why is that better than estimating one?
The full exposure map: each covered private company, every public company and fund that holds it, and the dollars you get per $10,000.