On August 14, 2019, WeWork filed the S-1 that was supposed to open its road to a public listing. The document showed revenue of $1.82 billion for 2018, a net loss of $1.9 billion for the same year, and future minimum lease payments of $47.2 billion as of that June.
Six weeks later the IPO was withdrawn, the CEO was gone, and the $47 billion valuation SoftBank had put on the company nine months earlier was heading for $8 billion. Four years after that, WeWork filed for Chapter 11.
I wrote about the same species of failure in The AI Trap, where a company with genuinely excellent technology was still burning cash faster than its business model could ever repay. WeWork is that pattern with a lease agreement instead of a GPU bill. The demand was real. The arithmetic was not built to survive a stall.
This is piece five of The Moat That Broke, a series on advantages that failed and the specific mechanism that killed each one. Nike’s mechanism was a channel giving up its own shelf space. Intel’s was a foundry falling a node behind and never closing the gap. Boeing’s was finance outvoting the engineers who could see the defect.
Peloton’s was a demand spike mistaken for a moat. WeWork’s mechanism is a duration mismatch: money coming in on 30-day terms, money going out on 15-year terms, and a growth rate that was the only thing keeping the two in balance.
When Your Liabilities Are Longer Than Your Contracts
Most businesses that fail on unit economics fail loudly. They lose money on every unit and everyone can see it in the pricing. WeWork’s failure mode was quieter and more dangerous, because for years the reported numbers could plausibly be read as a company still finding its footing rather than a company built on a structural mismatch.
The mismatch, defined
WeWork signed long leases on office buildings, typically for a decade or more, and committed to fixed rent for the full term. It then sold desks and offices to members on month-to-month or short-term memberships. That gap between the length of the liability and the length of the revenue meant to cover it is the entire story.
A landlord bears no risk from a tenant leaving early, because the tenant is still on the hook for the lease. WeWork inverted that, taking on the landlord’s fixed obligation while selling a product with none of the landlord’s protection against churn.
Why growth hides it
A duration mismatch is invisible for as long as new memberships are being signed faster than old ones lapse. Fresh monthly revenue covers the fixed rent bill, occupancy looks healthy, and the trailing numbers show growth.
The company can report an 89% average occupancy rate, as WeWork did in the S-1 for buildings open at least 18 months, and that figure is true and also almost uninformative about what happens if new signups slow down. The mismatch only becomes visible once growth stops, and by then the leases are already signed and cannot be cancelled at the same speed the memberships can.
The story was not decoration, it was the funding mechanism
It is tempting to treat Adam Neumann’s language about “elevating the world’s consciousness” and a valuation built on “energy and spirituality” as marketing excess layered on top of a real estate business. That undersells what the story was doing.
A pure real estate arbitrage business, borrowing long and renting short, would have been priced by investors like a real estate company: modest multiples, heavy scrutiny of lease coverage ratios, a hard look at what happens in a downturn.
The technology and community narrative was what let WeWork be priced instead like a software platform, on growth and total addressable market rather than lease coverage. The story is what let the mismatch go unpriced for years.
A $47 Billion Valuation Built on a $47.2 Billion Liability
WeWork was founded in 2010 by Adam Neumann and Miguel McKelvey, selling shared office space with a community layer on top: events, beer taps, a member app, a brand built to feel like a startup rather than a landlord. It worked.
By the time of its 2019 S-1, WeWork operated 528 locations across 111 cities and had exceeded 662,000 members worldwide, more than half of them outside the United States.
SoftBank started backing the company in 2017 with an initial $4.4 billion investment through its Vision Fund, and kept doubling down. By January 2019 SoftBank had priced a fresh round at a $47 billion valuation. Company insiders were reportedly preparing to sell shares into that price. Then came the S-1.
The filing, released August 14, 2019, showed the mismatch in numbers a public market had to read rather than take on faith.
Revenue of $1.82 billion in 2018 against a net loss of $1.9 billion the same year, meaning the company lost roughly a dollar for every dollar of revenue it booked.
Future minimum lease payments of $47.2 billion as of June 2019, up from about $34 billion at the end of 2018, a jump that outpaced the company’s own revenue growth. Under the new lease accounting standard, the balance sheet carried roughly $18 billion in long-term lease obligations against about $15 billion in matching right-of-use assets. Governance details, including Neumann’s control structure and related-party transactions, compounded the doubts.
The roadshow that followed did not go well. On September 24, 2019, Neumann resigned as CEO under board pressure. On September 30, WeWork formally withdrew its IPO prospectus.
SoftBank, already the largest shareholder and now the only investor still willing to write a check, completed a rescue financing in October that valued the company at $7.5 billion to $8 billion, an 83% collapse from the January mark. By SoftBank’s own later reckoning, its cumulative investment in WeWork by that point had reached $18.5 billion.
WeWork went public anyway, eventually, through a SPAC merger with BowX Acquisition Corp that closed on October 21, 2021, at a $9 billion valuation, a fraction of the 2019 peak.
The stock traded as a public company for two years while the underlying mismatch never actually went away, it had simply been repriced downward.
On November 6, 2023, WeWork Inc. and hundreds of affiliated entities filed for Chapter 11 in the U.S. Bankruptcy Court for the District of New Jersey, listing total debts of $18.65 billion against total assets of $15.06 billion. One of its first motions in the case sought to reject 69 unprofitable leases outright, and the company disclosed it was considering exiting more than 30% of its wholly owned portfolio.
Why it broke
The mechanism is specific. It is not “the company spent too much” or “the founder was reckless,” although both are true as colour. WeWork financed a long-duration, illiquid liability, a 10-to-15-year lease at fixed rent, using a short-duration, highly liquid source of revenue: a membership cancellable with 30 days’ notice.
That structure has exactly one stable state, continuous net growth in new memberships large enough to keep covering leases signed years earlier.
The moment membership growth flattens or reverses, as it did in the 2020 downturn and again as the post-pandemic return-to-office pattern proved uneven, the fixed side of the ledger keeps billing while the flexible side stops paying. No version of this structure survives a demand shock, because the whole model is a bet that demand never shocks.
What WeWork Never Got Wrong
The lazy version of this story treats WeWork as incompetent across the board, and that is not accurate. The demand for flexible, professionally managed office space was real in 2019 and it is still real now.
WeWork’s product, the physical experience of the space, the design, the operations of running hundreds of buildings to a consistent standard, was genuinely good, and it is a reason the brand survived rather than being liquidated.
After emerging from Chapter 11 debt-free in June 2024, the reorganized company kept operating, and by 2026 it reported around 146 U.S. locations and was investing more than $80 million a year back into its portfolio. The category WeWork invented did not fail. The balance sheet that tried to fund it did.
How to Avoid Funding a Long Liability With a Short Asset
1. Match the duration of what you owe to the duration of what you are owed
Before you sign any obligation measured in years, ask what revenue is contractually guaranteed to cover it for that same length of time. If the answer is “revenue we expect to keep renewing,” you have a duration mismatch, not a funded liability.
I laid out the broader discipline for this in the Unit Economics Playbook: every unit of growth should make the next unit of growth cheaper or safer, never more exposed.
2. Price the downside case, not just the growth case
WeWork’s occupancy numbers were real, but they described a world of continuous growth. Model what your fixed obligations look like at flat or negative growth for 18 months, not just at the trend line. If that scenario is fatal, the business is not resilient, it is a timer.
3. Do not let a compelling narrative substitute for a coverage ratio
A strong story about mission or platform potential is not a bad thing to have.
It becomes dangerous the moment it is used to wave off a question a lender or investor would otherwise ask about lease coverage, churn exposure, or contract duration. If your story is the reason nobody is stress-testing your liabilities, that is a warning sign about the story, not a compliment to it.
4. Separate the product decision from the financing decision
WeWork’s coworking product worked. Its financing structure did not. Those are two different judgments and they deserve two different reviews. A good product financed on the wrong terms can still fail completely, and a founder who only ever hears “the product is great” from investors may never get asked the financing question at all.
5. Build in an exit ramp before you need one
Part of what made WeWork’s 2023 filing so damaging was that unwinding bad leases required bankruptcy court lease rejection, a blunt and public instrument. If your business carries long fixed obligations, negotiate break clauses, shorter renewal cycles, or graduated commitments while you still have leverage, rather than renegotiating from visible distress.
A Schedule Is Not a Moat
WeWork’s stated moat was never really the app, the community events, or the brand. Those were real but replicable.
Its actual, unstated moat, the one investors were paying for at a $47 billion valuation, was the claim that growth would never stop long enough for the lease side of the business to catch up with the membership side. That is not a moat. It is a schedule.
The company that emerges from this is smaller, debt-free, and profitable on operations, which suggests the underlying product was sound all along. What killed the first WeWork was not the idea of flexible space. It was a balance sheet that borrowed like a real estate company and sold like a subscription app, and never reconciled the two.
A moat measured in growth rate rather than in structural durability is not a moat. It is a countdown.
This teardown is part of The Moat That Broke. The series closes with The Five Ways a Moat Breaks, which sets this case beside the other eight and turns all nine into one audit you can run on your own business.
