Why a studio books "did we build it right" separately from "did we price it right," the accounting logic that makes the dual-entity model legible to an LP.
Disclosure: This describes AOS's own structure. Figures are design parameters for our model, not realized results, and we are pre-launch.
The dual-entity structure has been explained as a chassis. This is the other half: the accounting logic that makes the chassis make sense.
What is an LP actually buying when they back a venture studio? That sounds like a soft question. It is the hardest one in the model, and the answer turns on a piece of accounting most studios never make legible.
A venture studio does two completely different jobs that get casually lumped into one word, "investing." It creates companies, and it invests in companies. Those carry different risks, on different clocks, judged by different evidence. Put them in one book and you get a number an LP cannot interpret. Split them across two entities and the LP can finally see what they are paying for, and what they are not. Our colleagues laid out the two boxes and how they connect in Dual Entity Model. Here we are after something narrower: why the split is not org-chart hygiene but the thing that makes the whole model readable to the person writing the check.
Every studio outcome is really the answer to two separate questions.
Question one: did we build it right? Did the company get made well, did the team form, did the product reach the market, did the demand show up. This is execution risk on creation. It is mostly about operating competence, and it plays out over a build cycle and the months after.
Question two: did we price it right? Given a company that exists and works, did we put capital in at a value that returns the fund. This is investment risk. It is about entry price, ownership, reserves, and exit, and it plays out over years.
Mash those into a single profit-and-loss and you cannot tell a studio that builds beautifully but overpays from a studio that builds messily but buys cheap. They can post the same blended return for completely opposite reasons. An LP underwriting that blended number is flying blind, and sophisticated LPs know it, which is why they discount what they cannot decompose.
So we put creation in its own entity. The Studio entity carries the cost and the risk of making companies: operator salaries, shared services, the build cycles, the concepts that go through the gate and stop. Its job is to convert ideas into investable companies at a known cost and a known hit rate.
The metric that governs it is operational, not financial. How many concepts entered, how many cleared the filter, what each cleared company cost to produce, how long it took. When we say something like 40 to 60 percent of deployed capital goes to operations rather than into the cap table, this is the entity where that shows up, and it is a feature of the creation engine, not a leak from the investment one. You are paying to manufacture investable companies. That line item is the manufacturing.
A blended return answers neither question honestly. It just averages two different bets into one number nobody can act on.
The Fund entity does the other job. Given companies the Studio produced (and, where it fits, companies it did not), the Fund decides entry price, ownership stake, reserve strategy, and follow-on. Its metrics are the ones LPs already speak: IRR, DPI, TVPI, multiple on invested capital.
The clean separation means the Fund's performance can be judged as an investment portfolio, on its own terms, without creation costs smeared across it. An LP in the Fund is underwriting capital allocation and price discipline, full stop. An LP can hold a view on the Fund's investing skill that is uncontaminated by the Studio's building skill, and vice versa. Two bets, two scorecards.
This is the payoff, so here is exactly who benefits.
An allocator's whole job is attribution: knowing why a return happened so they can decide whether it repeats. A mixed book defeats attribution by construction. The dual-entity split restores it. The LP can ask, and answer, two clean questions. Is this team good at making companies? Is this team good at pricing them? Those answers can diverge, and knowing which one is driving results is the difference between an informed allocation and a hopeful one.
It also clarifies where the LP's own capital sits and where it does not. Creation risk can be borne largely by the Studio entity and its backers. An LP can take Fund exposure, the investment risk, without taking the full operating risk of the build machine, if that is the exposure they want. The seam between the two entities is exactly the line an allocator gets to choose to stand on. That optionality is only visible because the seam exists.
Candor again. Two entities is more structure than one. More governance, more accounting, more surface for things to go wrong, real friction at formation. We do not pretend the split is free.
The split also forces honesty that a blended number lets you dodge. If the Studio is expensive and the Fund's entries are mediocre, two scorecards make both visible, where one scorecard would have hidden each behind the other. That is uncomfortable by design. We think discomfort that an LP can see beats comfort they cannot trust, and we would rather argue from the harder, more legible position.
So go back to the opening question. What is the LP buying? With one blended book, they cannot tell you, and neither can we. With two, the answer is finally sayable: a Studio you can judge on whether it builds well, and a Fund you can judge on whether it prices well, scored apart so neither hides inside the other. That is what the structure is for. For how creation risk gets priced down before capital enters, see The Six-Week Decision Cycle; for the alignment side of the Fund's economics, see Equitable Venture.
Nothing here is an offer to sell a security or investment advice.
Field notes on venture building, AI, and capital. No spam, unsubscribe anytime.
By subscribing you agree to receive AOS Insights e-mails. We use your address only for this newsletter - see our Privacy Policy.
