WeWork Wasn't a Money Machine. It Was a Money Machine Run in Reverse.

Picture two contracts on the same desk. One runs fifteen years, signed in ink, with a landlord who will collect rent whether or not anyone shows up. The other can be cancelled by next month. WeWork's entire business was the bet that the second contract - the cancellable one - would reliably cover the first. It signed long-term leases of up to fifteen years and rented the same space back by the month.⁵ That is not a clever arbitrage. It is a maturity mismatch dressed up as a tech company, and it was visible in the company's own filings from the start.

The official story was that WeWork was a software-margin business hiding inside a real estate shell - that occupancy and scale would eventually flip the losses to profit. The real story is the opposite: scale was the disease, not the cure. Every new location added a fifteen-year liability faster than it added month-to-month revenue, so growth widened the gap it was supposed to close.

The number that broke the model was in the prospectus

When WeWork filed its S-1 in August 2019, the document carried its own indictment. Against $47 billion in future lease obligations, the company had committed future revenue of only $4 billion.¹¹ Read that ratio slowly. For every dollar of revenue WeWork could count on, it owed nearly twelve dollars in rent it had already promised to pay. The asset side was a hope - members who could leave at will - and the liability side was a contract. You cannot occupancy your way out of a liability twelve times larger than your committed income, because filling the desks doesn't shorten the lease; it only matters if the rent you charge clears the rent you owe, and it never reliably did.

Three years of hypergrowth, zero operating leverage

A real money machine gets cheaper to run per unit as it grows - the fixed cost spreads, the margin widens. WeWork did the reverse and proved it with its own ledger. Total expenses ran at 190% of revenue in H1 2019, almost exactly the same as the 190% it ran back in 2016.³ Revenue had exploded from $436 million in 2016 to $1.535 billion in just the first half of 2019, yet losses kept pace, swelling from $430 million in 2016 to a $690 million net loss in that same half-year.¹ Three years of breakneck growth bought the company precisely no operating leverage. The line that should have bent toward profit stayed flat as a ruler, because the lease obligations grew in lockstep with the footprint.

Even the per-unit improvements WeWork could genuinely point to never escaped this trap. Its adjusted location operating expenses fell from $27.80 per square foot in Q4 2019 to $22.97 in Q4 2021 - a real efficiency gain.⁸ And it didn't matter. Total annualized location operating expenses still climbed to roughly $3.1 billion in 2021 and $3.3 billion in 2020, up from $2.4 billion in 2019, because the company kept adding square footage faster than it cut the cost of each foot.⁸ Getting cheaper per desk while losing more in total is the signature of a model where expansion is the problem.

When the metric hides the building, look at the building

Because the GAAP picture was so ugly, WeWork led with a number of its own design: 'Community Adjusted EBITDA.' The trick was in what it removed. The metric stripped out not just interest, taxes, depreciation, and amortization, but also rent, tenancy expenses, utilities, building staff salaries, and amenity costs - the actual costs of running a real estate business.⁴ It was, in effect, profit before the cost of having buildings, sold by a company whose entire product was buildings. It hadn't even been invented for the IPO; it first appeared in an earlier bond offering, and was later quietly renamed 'contribution margin' after the SEC pushed back.⁴ The metric was a tell. When the headline number you're shown subtracts the one cost that defines the business, the business is telling you where the body is buried.

The valuation followed the same logic as the metric: it described a story, not a market. The famous $47 billion came entirely from SoftBank's $2 billion Series H round in January 2019 - one investor, setting one price.⁶ Three years earlier, the pre-SoftBank Series F round had valued the company at $16.9 billion.⁶ No independent public-market buyer ever marked WeWork near $47 billion; the IPO was withdrawn in September 2019 amid weak investor demand and a dwindling implied valuation, rather than priced at anything close to that figure. The number was a belief held by one party, not a value confirmed by many.

But couldn't enterprise tenants and longer terms have fixed it?

The fair objection is that the model wasn't doomed by physics - it was doomed by execution. Sign more enterprise clients on multi-year deals, narrow the maturity gap, push occupancy higher, and the mismatch shrinks. There's truth in it: a tighter enterprise mix genuinely reduces the cancellation risk on the revenue side. But it cannot close the real wound, which is price. WeWork's whole appeal to members was sub-market, flexible space - charging less per foot than a conventional lease while paying above-market rents for prime, built-out locations.¹⁰ Analysis of the all-in cost structure found that WeWork's sublease pricing would have needed to be roughly twice the market rate to cover its true costs.⁹ Doubling your price is not a tweak to unit economics; it is the abolition of the value proposition. The flexibility customers paid for was the same flexibility that made the revenue unreliable. You could fix the mismatch only by becoming a normal, expensive landlord - which is the one thing WeWork's entire pitch was built to avoid.

The ending was written in the opening contracts. WeWork filed for Chapter 11 in November 2023 with $18.65 billion in debts against $15.06 billion in assets, dragging close to $16 billion in long-term lease obligations to the grave - the same liabilities that scale was supposed to outgrow.² It emerged in June 2024 having shed over $4 billion in funded debt, mostly by tearing up the leases it could never have covered.⁷ That is the quiet confirmation of the whole thesis. WeWork didn't fail because it grew too fast or spent too freely. It failed because it ran a money machine backwards: it paid long, charged short, and called the difference a community.