The Forty-Seven-Billion-Dollar Sublease

In January 2019, the Japanese investment fund SoftBank Vision Fund led an investment round that valued WeWork at forty-seven billion dollars. The company, founded by Adam Neumann and Miguel McKelvey in 2010, operated a chain of co-working office spaces — buildings the company leased from landlords on long-term, fixed-rate contracts, then subleased to small businesses and freelancers on short-term, variable-rate memberships. The arbitrage between the long lease (paid in fixed dollars) and the short sublease (collected in variable dollars) was the entire business.

Nine months later, the company filed an S-1 to go public. The filing was withdrawn within weeks. The valuation collapsed. By 2023, WeWork filed for Chapter 11 bankruptcy. SoftBank's investment, which had peaked at over eighteen billion dollars deployed across multiple rounds, was substantially written off. Neumann, who had personally extracted over a billion dollars in cash through the company's various financing rounds, left the company with a settlement payment from SoftBank that observers at the time described as historically unusual in its generosity to the departing founder.

The interesting question is what specifically went wrong, and what the period from 2019 to 2023 tells us about how capital markets occasionally lose the ability to distinguish a real estate arbitrage from a technology company.

The Mismatch. WeWork's central business risk, visible to anyone willing to read the lease schedule, was an asset-liability duration mismatch. The company had committed to multi-billion-dollar fixed-rate obligations to landlords (often ten to fifteen year leases on the most expensive office buildings in major cities). It collected revenue from members on monthly subscriptions that the members could cancel at thirty days' notice. In a stable economic environment, this mismatch was manageable — sublease premiums were sufficient to cover the lease cost plus operating expenses plus a margin. In a recession, or any economic environment that shrank demand for office space, the mismatch became catastrophic: the fixed obligations continued, the variable revenue collapsed, and the company would burn cash at a rate proportional to its scale.

The 2019 S-1 disclosed this mismatch, though it did so in language designed to make it seem less alarming than it actually was. The S-1 described WeWork as a technology company despite the company's operating activity being almost entirely the leasing of physical real estate. The filing repeatedly used the word "community" to describe what was, in financial terms, a subleasing operation. The CEO held multiple roles, controlled multiple voting classes, owned the trademark "We" which the company had purchased from him personally for several million dollars, and had structured the governance such that the public investors would have no meaningful path to influence management decisions. The combination of disclosures — duration mismatch, governance opacity, related-party transactions — was so concerning that the IPO collapsed within weeks of the prospectus becoming public.

The Pandemic Acceleration. What had been a slow-motion problem became fast-motion during 2020-2022. Office demand collapsed across the major American cities. Companies reduced their square footage requirements as remote work became normalized. The implicit contract — that small businesses needed flexible office space and would pay a premium for it — was disrupted, possibly permanently. WeWork was, by this point, attempting to renegotiate landlord leases on hundreds of buildings, sometimes succeeding, often failing. The cash burn was sustained at rates that no amount of new financing could absorb.

The bankruptcy, when it came, was not a surprise. The post-bankruptcy entity continues to operate, in much smaller form, with renegotiated lease terms and a clearer focus on profitable building-by-building economics.

The Lesson Hidden in the Branding. What the WeWork experience demonstrated, with unusual clarity, was the persistent capacity of capital markets to misread the nature of a business when the founder's pitch is sufficiently charismatic and the secular story (in this case, the future of work) is sufficiently exciting. The asset-liability mismatch was visible in 2014. The governance problems were visible in 2016. The pattern of related-party transactions was visible in 2017. The valuation continued to expand. The fund managers continued to participate. The eventual unwinding — bankruptcy, write-down, founder enrichment in inverse proportion to investor return — was foreseeable to anyone willing to read the lease schedule carefully and ignore the founder's tendency to call subleased offices a "physical social network."

This is the residual cost of overwhelmingly enthusiastic capital market periods. The expensive lessons get printed. The pitchbooks get rewritten. The next charismatic founder gets a slightly more skeptical reception. And then, in the next cycle, the lesson is forgotten, the next pitch is delivered, and the schedule of leases is once again neglected in favor of the language of community.