WeWork Didn't Fail at Its IPO. The IPO Did Exactly What It's Supposed To.

On August 14, 2019, a company that had spent years being described as a tech firm filed a 350-page legal document and, in the language of accountants and lawyers, finally had to say what it actually was. The S-1 disclosed a net loss of $1.9 billion on $1.8 billion of revenue for 2018, and another $0.9 billion lost in just the first half of 2019.¹ None of this was new — it had all happened. What was new was that, for the first time, it had to be written down where strangers could read it. Six weeks later the offering was dead.⁵

The story everyone tells is that WeWork picked a bad moment, that the market turned, that SoftBank pushed too hard. It's a story about timing and ego. The real story is duller and far more damning: nothing changed about the company between January and September 2019. What changed was who got to look at it.

A $47 billion number that exactly one investor had ever tested

Here is the fact that reframes everything. WeWork's famous $47 billion valuation was not set at the IPO. It was set in January 2019, seven months earlier, when SoftBank led a $1 billion Series H round at that mark.⁴ That number was never a market price. It was the opinion of a single, extraordinarily enthusiastic buyer, encoded into a funding round and then repeated so often it began to sound like a fact. A private valuation is not a price discovered by competition. It is a price one investor is willing to pay, frozen in time, with no one on the other side to argue. WeWork's $47 billion was that — a number with one author.

An IPO is the moment that number meets resistance for the first time. The S-1 isn't marketing; it's the mechanism by which a thousand skeptical investors get to do, simultaneously and in public, the diligence that one cheerleader skipped. That is the whole point of the document. So when the S-1 'unraveled,' it wasn't malfunctioning. It was working perfectly — converting a private fantasy into a public question and letting the answer come back loud.

The business that calls a lease an asset and the rent an adjustment

Strip away the talk of community and elevated global consciousness, and WeWork did one thing: it signed long leases — often around fifteen years — and re-rented the same space in short, flexible chunks.⁸ That is rent arbitrage, and it is an old, capital-hungry, deeply cyclical business. The danger is structural: the long-term lease obligations are fixed and enormous, while the short-term sublease income evaporates the instant a downturn empties the desks. You owe for fifteen years and collect by the month. In a recession, that gap is the whole company.

To make this look like software, WeWork leaned on a metric of its own design: 'Community Adjusted EBITDA.' It took adjusted earnings and then further subtracted general and administrative expenses and other core operating costs.³ In a 2018 debt filing the company reported $233.1 million of Community Adjusted EBITDA for 2017 — alongside an actual net loss of $933.5 million that same year.³ The metric didn't measure profitability. It measured profitability if you ignored most of the cost of running the business, including the rent that was the business. Harvard Business School's analysis put it plainly: the financial community panned the figure for excluding rent, G&A, and other core operating costs.⁸

When the founder rents his own company its name

The numbers might have been survivable. The governance was not, because it told investors something the financials couldn't: whose interests the company actually served. The clearest example sat in the disclosures. WeWork's parent had acquired rights to the 'We' family of trademarks from WE Holdings LLC — an entity co-managed by CEO Adam Neumann and co-founder Miguel McKelvey — in exchange for $5.9 million worth of partnership interests.⁶ The CEO, in effect, had the company pay for the right to use the name it was built on, to an entity he himself controlled.

The backlash was immediate enough that, in an amended S-1 filed September 4, the company disclosed the arrangement had been unwound 'at Adam's direction,' with the partnership interests returned and WeWork keeping all the trademark rights.⁶ But the reversal proved the point rather than fixing it. A deal that was fine in January and embarrassing in September hadn't changed — only the audience had. Public disclosure has a way of making a thing look exactly as it is.

Wasn't this just bad timing and a bad market?

The fair objection is that markets are moody and the late-2019 window was unkind to unprofitable growth stories — that a better tape might have gotten the deal done. There's something to it. But it gets the causation backwards. The financials in the S-1 were audited and finalized; Ernst & Young signed off on statements dated April 25, 2019.² These were not surprises that bad luck exposed. The same SoftBank that had stamped $47 billion on the company was, by filing time, reportedly pushing to delay the offering — the most enthusiastic buyer in the world wanted the test postponed. And when the test came anyway, the answer wasn't a 20% haircut you'd blame on the weather. It was a reported $10 billion against $47 billion — a collapse of roughly four-fifths.⁵ A bad market trims a fair price. It does not erase three-quarters of a valuation. That gap was never the market's mistake to make; it was the difference between a price one investor imagined and a price the world would pay.

WeWork's S-1 is sometimes called a catastrophe of transparency, as if the company would have been fine had it only stayed quiet. That has it exactly wrong. The document didn't break the company; it described one. Every load-bearing problem — the unprofitable arbitrage, the metric built to hide it, the founder renting the company its own name — existed in January, when the price was $47 billion and nobody outside the building had to look. The IPO that wasn't was never a failed offering. It was a successful audit, and what it found was a number with one author and no defense. Sunlight didn't ruin WeWork. It just stopped agreeing to look away.