How Are Private Investments Valued? WeWork's $47 Billion

Private investments are valued by a mark, not a price. The manager holding the asset assigns a number each reporting period, usually anchored to whatever the most recent investor paid, sometimes adjusted against comparable public companies or an estimate of future cash. That number is then applied to every share, including the ones that never traded. It is an estimate wearing the clothes of a fact.

In January 2019, SoftBank put $2 billion into WeWork.¹ It had been discussing a far larger commitment and pulled back; part of what it did write was fresh capital into the company and part bought shares from people who already held them. The terms implied a valuation of $47 billion for the whole business.¹ Nobody had paid $47 billion for WeWork. Nobody ever would. But for most of that year, $47 billion sat inside the quarterly statements of funds that owned a piece of it, and it was the figure the company carried into its attempt to go public that August. The episode appears here as history and as a lesson in method — not as a view on any company, fund or security.

Owners who hold stakes in venture, buyout or real-estate funds meet this machinery every quarter, usually without noticing it. The statement arrives, the number has moved slightly or not at all, and where it came from goes unasked. An accountant could ask thirty questions of that figure. Three decide whether it should change anyone's behaviour.

Who set the $47 billion, and what did that cheque actually buy?

The governing principle: a private valuation is the last cheque, multiplied out. When an investor buys newly issued shares at some price per share, the manager multiplies that price by every share in existence and reports the product. The industry calls the result the post-money valuation — the company's implied value immediately after the new money lands. Every fund holding the company then has a defensible reason to carry its own stake at the same price per share, because accounting rules favour an observable transaction over an imagined one.

Two catches live inside that arithmetic. The first is size. A cheque of that kind buys a small slice of the company; a price paid for a sliver is then treated as a price for the whole. The second is what the sliver consisted of. Late-stage investors rarely buy the same shares founders and employees hold. They buy preferred stock — shares that come with protections, typically the right to be repaid first if the company is sold cheaply, and often the right to extra shares if a later round prices lower. A share carrying insurance is worth more than one without it. Copying the insured price across the uninsured shares overstates the whole.

What changes if the answer is no: if a holder cannot name who set the last mark, how much of the company they bought, and what protections they received, the figure on the statement is not a valuation but a rumour with a decimal point. It repeats, in institutional clothing, the gap between a number that has been printed and a number a second buyer will actually pay. The question for the next quarterly letter is plain: which round is this mark based on, and what did that investor get that we did not?

Why did WeWork's mark take nine months to move?

The governing principle: the accounting standard asks for an exit price on the reporting date — what a willing buyer and a willing seller would agree on that day, in an unhurried sale. The last cheque is evidence toward that estimate, not a substitute for making it. In practice, though, private marks tend to move on events rather than on information: a new round, a sale, a bankruptcy, an auditor who will no longer sign. Between events, the default is inertia. Accounting rules sort valuation inputs into three levels, and private stakes sit at the bottom, Level 3 — the polite term for "valued on the manager's judgement, because no market exists."

WeWork's filing to go public, in August 2019,² told anyone who read it that the business had lost roughly $1.9 billion in 2018 on about $1.8 billion of revenue,⁴ that its leases ran for fifteen years while its members could leave within a month, and that its founder had personal dealings with the company a public board would not tolerate. Revenue was growing; cash was not accumulating — a distinction owners already know how to test, and one that does not need a new vocabulary. None of it was secret to the existing investors. What was new was that a wider set of buyers was now being asked to write the cheque, and declined at anything like the private number. The offering was withdrawn in September,⁵ the founder stepped down before the IPO was withdrawn,³ and in October SoftBank took control through a rescue package that valued the company at roughly $8 billion.⁶

Nine months, $47 billion to about $8 billion.⁷ The business did not lose five-sixths of its worth in that time. The mark had been waiting for an event, and the event arrived.

What changes if the answer is no: if the last event that touched a mark is more than a year old, the sensible assumption is that the number is stale in whichever direction the world has moved — and that the direction is more often down than up, because managers mark upward willingly and downward reluctantly. A quarterly figure that has not moved is not stability. It is silence.

What would the stake fetch if it had to be sold tomorrow?

The governing principle: a mark is a reporting number; a price is what a buyer pays today. The gap between them is the cost of being private. Fund interests can be sold before a fund winds up, through what is called the secondary market — specialist buyers who purchase other people's fund stakes. Those buyers routinely pay less than the reported mark, sometimes slightly, sometimes a great deal, because they are the ones absorbing the uncertainty the mark politely ignores.

The deeper test is older than any accounting standard. A buyer who must be repaid out of the business itself, rather than by the next investor, asks how much cash the thing produces and how durable that cash is. WeWork's own documents had answered that for anyone willing to look past the headline: an earlier bond prospectus introduced a measure it called "community adjusted EBITDA," which stripped out so many ordinary costs that a large loss became a modest profit. The phrase became shorthand for the distance between a company's adjusted profit and the cash it actually keeps. The $47 billion rested on a story about growth. The $8 billion rested on what a controlling owner thought the leases and the cash burn were worth. The company eventually went public through a merger in 2021,⁹ at a fraction of the original figure, and filed for bankruptcy protection in 2023.¹⁰ Each of those later numbers was a price. Only the first had ever been a mark.

What changes if the answer is no: if a holder has no idea what a secondary buyer would pay, the mark should not be borrowed against, counted toward spending, or used to tell the family what they are worth. It belongs in a column headed "estimate" and should be discounted privately, even while the statement reports it in full.

What to do with the three answers

Three questions, asked of every private holding once a year: who set the mark, on what terms and for how much of the company; what event last moved it and how long ago; and what the stake would fetch if it had to be sold. None of them requires a finance background. They require only the habit most founders already have — distrusting a number until someone can explain where it came from.

The WeWork mark was not fraud. It was the system working as designed. The design assumes the last person to write a cheque was right about the whole company, and that the next event will confirm it. Owners holding fund interests are under no obligation to share that assumption. A buyer's term sheet was never the value of a business until the cash cleared; it was one offer, for one structure, on one day. Private investments are valued the same way. The quarterly statement simply prints the offer more often, and for longer, than a term sheet ever did.