Mark ở vũ trụ khácVIMark học AI
Mark ở vũ trụ khác · Another Universe Anatomy of a Business Failure #9 · Real estate · Fundraising

WeWork

long lease short members

The We Company / WeWork Inc. · IPO pulled Sep 2019 · bankruptcy filing Nov 2023

What happenedIPO filing withdrawn after 6 weeks; bankruptcy filing 4 years later
Shock number2018: a $1.93 billion net loss on $1.82 billion of revenue
Root causelong leases in, short memberships out, growth faster than profitability
Main sourcesForm S-1 (Aug 14, 2019), 10-K, 10-Q, Chapter 11 filing (Nov 2023)

Source: The We Company Form S-1 (Aug 14, 2019), selected financial data; withdrawal statement of Sep 30, 2019; WeWork Chapter 11 announcement (Nov 6, 2023).

The 1-minute recap

Essence

Losses bigger than revenue2018: revenue $1.82 billion, net loss $1.93 billion.
Lease long, sell shortAbout $47 billion of rent owed over many years, while members could leave early.
A story bigger than the numbersThe IPO filing's mission: "to elevate the world's consciousness".
One person, a lot of powerThe founder's shares once carried 20 votes each.
IPO pulled, then bankruptcyFiling withdrawn Sep 30, 2019. Listed another way in 2021. Chapter 11 on Nov 6, 2023.

Source: The We Company Form S-1 (Aug 14, 2019): financial data, lease note, prospectus summary; amended S-1 (Sep 2019) on voting rights; statement of Sep 30, 2019; WeWork Inc. 8-K (Oct 20, 2021); Chapter 11 announcement (Nov 6, 2023).

Part 1

The question

A company with hundreds of thousands of members and revenue nearly doubling each year: why did it still fall? Because each desk it sold did not pay for its own rent, and a story about "community" covered that up for years.

Source: 527,000 members at June 30, 2019 (CNBC citing the S-1, secondary source); revenue and losses: S-1, selected financial data. The answer is Mark's analysis.

Before 2019Grow very fastLease buildings long-term, fit them out, re-let them desk by desk, room by room.
→
Aug 14 – Sep 30, 2019The IPO filing shows the numbersInvestors read the S-1, saw the losses and the lease obligations. The IPO was pulled after 6 weeks.
→
Nov 6, 2023BankruptcyAfter one rescue and one SPAC listing, it still could not carry its old leases.
Part 2

What happened

Up to the peak and down again in about five years. Dates and numbers come from SEC filings and the companies' own announcements.

WeWork's path. January 2019, valued at about $47 billion in a private funding round (press reports). Aug 14, 2019, IPO filing. Sep 30, 2019, filing withdrawn. October 2019, SoftBank rescue package. Oct 20, 2021, listed by merging with the SPAC BowX. Aug 8, 2023, the company itself reports substantial doubt about continuing as a going concern. Nov 6, 2023, bankruptcy filing. Jan 2019valued ~$47B* Aug 14, 2019S-1 filed Sep 30, 2019IPO pulled Oct 2019SoftBank rescue Oct 20, 2021listed via SPAC Aug 8, 2023"substantial doubt" Nov 6, 2023bankruptcy * private funding round valuation, per press reports; not a market price · time scale illustrative
The peak was a number on paper; the bottom was a petition in court. The $47 billion was the price of a private funding round, never tested on an open market: once the IPO filing came out, talk of the valuation fell to roughly $10–15 billion, and then the IPO was pulled.Source: $47 billion and $10–15 billion valuations: Forbes (Nov 7, 2023), CFN, CFO.com (secondary sources); S-1 (Aug 14, 2019); withdrawal statement Sep 30, 2019; SoftBank release Oct 23, 2019; WeWork Inc. 8-K Oct 20, 2021; Q2 2023 10-Q (Aug 8, 2023); Chapter 11 announcement (Nov 6, 2023).
Jan 2019 · The paper peakA private funding round values the company at about $47 billion (secondary source).
Aug 14, 2019 · IPO filing (S-1)2018: revenue $1.82 billion, net loss $1.93 billion. First half of 2019: revenue $1.54 billion, net loss about $904 million.
Sep 2019 · Filing amendedThe founder's voting power cut from 20 to 10 votes per share. The $5.9 million paid for the "We" name is unwound.
Sep 24, 2019 · New leadersAdam Neumann steps down as CEO and becomes non-executive chairman. Two co-CEOs take over.
Sep 30, 2019 · IPO pulledThe co-CEOs: postponing the IPO "to focus on our core business".
Oct 22–23, 2019 · SoftBank steps in$5 billion of new financing, $1.5 billion of existing commitment accelerated, a tender offer for up to $3 billion of shares. The tender offer was ended on Apr 2, 2020.
Oct 20, 2021 · Listed via SPACMerger with BowX, ticker "WE" on the NYSE, about $1.3 billion of gross proceeds.
2021–2022 · Still losing moneyNet loss attributable to the company: $4.44 billion (2021), about $2.03 billion (2022).
Aug 8, 2023 · A warning from insideThe quarterly report states "substantial doubt" about continuing as a going concern; cash down to $205 million.
Nov 6, 2023 · BankruptcyChapter 11 filed in the U.S. Bankruptcy Court for the District of New Jersey, to deal with its "legacy leases".

Source: S-1 (Aug 14, 2019), selected financial data; amended S-1 Sep 2019 (via Allwork.space, Fox Business, secondary sources quoting the filing); The We Company release Sep 24, 2019; statement Sep 30, 2019; SoftBank releases Oct 23, 2019 and Apr 2, 2020; WeWork Inc. 8-K Oct 20, 2021; 10-K for 2021 and 2022; Q2 2023 10-Q; Chapter 11 announcement Nov 6, 2023.

Part 1

What happened

$47B on papera 6-week IPOSPACChapter 11
Part 2

Why

lease long, sell shortunit economicsgovernancemyths
Part 3

Work lesson

one-page unit economicsmatch the termsask "what was left out"
Part 3

Why: four layers of causes

Strategy layer: lease long, sell short

duration mismatchS-1, lease note
On the left, WeWork signs multi-year, non-cancelable building leases: at June 30, 2019, total undiscounted fixed minimum lease payments were about $34 billion, plus $13.1 billion for leases signed but not yet commenced, about $47 billion in all. About $23.8 billion of the $34 billion falls in 2024 and later. On the right, members rent desks and rooms with much shorter commitments. When members leave, the building rent is still due. WeWork pays landlords (fixed, many years) ~$34.0B in the lease table · $23.8B of it due in 2024 and later + $13.1B signed, not yet commenced → ~$47B Members pay WeWork (much shorter commitments) the gap: members leave, rent keeps running member bars are illustrative, not to scale
This is the old way of making money in office subleasing, just run very fast. It profits when the building is full and the re-let price is above the rent paid. It hurts when members leave faster than the leases end.Source: S-1 (Aug 14, 2019), lease note at June 30, 2019: total undiscounted fixed minimum lease cost payments of $34,036,725 thousand; the sentence "exclude an additional $13.1 billion relating to executed non-cancelable leases that have not yet commenced". The comment is Mark's.

Unit economics layer: sell more, lose more

unit economicsS-1; 10-K 2021, 2022
Revenue and net loss attributable to the company. 2018: revenue $1.82 billion, loss $1.61 billion (total net loss including noncontrolling interests: $1.93 billion). First half of 2019: revenue $1.54 billion, loss $0.69 billion. 2021: revenue $2.57 billion, loss $4.44 billion. 1.821.61 1.540.69 2.574.44* 2018H1 20192021 revenue ($B) net loss attributable to the company ($B) · *2021 bar truncated
Growing revenue does not prove the model works. The right question: for each building, each desk, after rent and operations, what is left? In 2018 the total net loss ($1.93 billion) was larger than revenue itself.Source: S-1 (Aug 14, 2019), selected financial data (2018: revenue 1,821,751, net loss 1,927,343, loss attributable to the company 1,610,792, in thousands of USD; H1 2019: revenue 1,535,420, loss attributable about 689,700 thousand); 10-K for 2021 (revenue 2,570,127, loss attributable to WeWork Inc. 4,439,027 thousand).

People layer: too much power in one place

governanceS-1 and amended S-1

20 votes per share

The founder's high-vote shares held control. The September 2019 amendment cut this to 10 votes; he still kept a majority.

The "We" name for $5.9 million

The company issued $5.9 million of partnership interests to the founder's private entity to acquire the "We" trademark. Unwound in the amended filing.

Related real estate deals

The founder pledged to give the company any profits from real estate transactions he had entered into with the company.

Source: S-1 (Aug 14, 2019) and amended S-1 (Sep 2019), via Allwork.space, Fox Business, Axios, Business Insider quoting the filings (secondary sources). This is what the filings themselves state; the sources used here contain no court ruling or finding of wrongdoing against any individual.

Incentive layer: easy money rewards speed

incentivesSoftBank 2019–2020; 8-K 2021; 10-Q 2023
A loop: a big story pushes the valuation up, a high valuation pulls in capital, capital leases more buildings, more buildings raise revenue, rising revenue feeds the story. No step in the loop asks whether one building makes money. a big story high valuation equity and debt lease more buildings →→→ revenue grows → feeds the story no step asks: does one building make money?
This loop runs well while money is cheap and easy. SoftBank's October 2019 package was $5 billion of new financing (mostly debt: $1.1 billion senior secured notes, $2.2 billion unsecured notes, a $1.75 billion letter of credit facility). New money buys time; it does not fix unit economics by itself.Source: SoftBank Group release, Oct 23, 2019. The loop is Mark's analysis.

After 2019, same model

Listed via SPAC in October 2021 with about $1.3 billion raised, yet lost $4.44 billion (2021) and about $2.03 billion (2022).

2023: members leave, cash runs low

The Q2 2023 10-Q cites rising "member churn", weaker demand than planned, $205 million of cash.

What had to go: the old leases

At the bankruptcy filing, CEO David Tolley said plainly that the aim was to deal with its "legacy leases".

Source: WeWork Inc. 8-K (Oct 20, 2021); 10-K for 2021 and 2022; Q2 2023 10-Q (Aug 8, 2023); WeWork Chapter 11 announcement (Nov 6, 2023).

Part 4

Fact check: what's true and what's myth

14 things often said about WeWork, checked against SEC filings and the companies' announcements. Click each row for the reasoning.

Two $47 billion figures with two different meanings. One is the private funding round valuation of January 2019, per press reports. The other is total rent owed per the S-1: $34.0 billion in the table plus $13.1 billion of leases signed but not yet commenced. The first "$47B" · valuation, Jan 2019 private round (press) The second "$47B" · rent owed (S-1) $34.0B in the table+ $13.1B not yet started
Same number, different nature. One is what people thought the company was worth. The other is how much rent the company had promised to pay. Only the second one does not disappear when the story stops sounding good.Source: valuation: Forbes (Nov 7, 2023), CFN (secondary sources); rent: S-1 (Aug 14, 2019), lease note at June 30, 2019.
SOLIDIn 2018, WeWork lost more than its revenue

S-1, selected financial data: revenue $1.82 billion, total net loss $1.93 billion.

SOLIDThe IPO filing's mission was "to elevate the world's consciousness"

The line sits in the S-1 prospectus summary (cited by CNBC and Fortune; WeWork's own 2019 newsroom post says the same).

SOLIDThe company paid $5.9 million for the "We" name to the founder's private entity

As stated in the S-1 (paid in partnership interests, not cash). The September 2019 amendment says the issuance was unwound "At Adam's direction". Secondary sources disagree on the date it was returned.

SOLIDIn 2023, WeWork itself wrote "substantial doubt" about its survival

Q2 2023 10-Q (Aug 8, 2023), citing losses, cash needs and higher member churn. The interim CEO called it a technical accounting determination.

CAUTION"A $1.9 billion loss" or "a $1.6 billion loss" in 2018?

Both are in the S-1. $1.93 billion is the total net loss; $1.61 billion is the part attributable to the company after noncontrolling interests. Press reports picked one or the other, so they look contradictory.

CAUTION"WeWork was once worth $47 billion"

That was the valuation of a private funding round in January 2019 (press reports, secondary source). After the IPO filing, talk of the valuation fell to about $10–15 billion, then the IPO was pulled. An open market never paid $47 billion.

CAUTION"WeWork owed $47 billion in rent"

Close on the number, wrong on "owed". S-1: $34.0 billion of undiscounted fixed minimum lease payments, plus $13.1 billion of leases signed but not yet commenced. It is paid over many years ($23.8 billion falls in 2024 and later), not debt due now.

CAUTION"The IPO failed only because of Adam Neumann"

The governance problems were real and sat right in the S-1. But losses larger than revenue and $47 billion of rent were in the S-1 too. In 2021–2023, after he had left the CEO job, the company still lost money and then went bankrupt.

CAUTION"SoftBank pumped in $9.5 billion to save WeWork"

The package was $5 billion of new financing (mostly debt), $1.5 billion of an existing commitment accelerated, and a tender offer for up to $3 billion of existing shareholders' stock. SoftBank ended the $3 billion tender offer on Apr 2, 2020.

CAUTION"WeWork went bankrupt because of COVID"

The pandemic did hurt: 2021 revenue ($2.57 billion) was below 2020 ($3.42 billion). But the big losses date from 2018, before the pandemic. In 2023 the company cited member churn, office oversupply and rising competition.

CAUTION"After bankruptcy, WeWork disappeared"

Chapter 11 is a restructuring, not a shutdown. WeWork kept operating and emerged from bankruptcy in 2024. The 2024 date was not checked against court records for this piece, hence the caution label.

FALSE"Shared office space was a bad idea from the start"

The demand was real: about 527,000 members in mid-2019 (CNBC citing the S-1). In 2023 WeWork itself cited rising competition in flexible space, meaning many others do the same thing. What failed was not the idea but the structure: long lease commitments, short member commitments, and growth faster than profitability.

FALSE"Listing in 2021 saved WeWork"

After the SPAC merger (Oct 20, 2021), the company still lost $4.44 billion in 2021 and about $2.03 billion in 2022, then wrote "substantial doubt" (Aug 2023) and filed for bankruptcy (Nov 2023).

FALSE"WeWork was a tech company"

The biggest obligation in the S-1 is rent, not software or servers. Its economics are those of a landlord re-letting space, with the same vacancy risk.

Source: S-1 (Aug 14, 2019) and amended S-1 (Sep 2019); SoftBank releases Oct 23, 2019 and Apr 2, 2020; WeWork Inc. 8-K Oct 20, 2021; 10-K 2021, 2022; Q2 2023 10-Q; Chapter 11 announcement Nov 6, 2023; secondary sources named in each row.

Part 5

Verdict

Scores are Mark's assessment based on the sources in Part 8, not data from a study.

Avoidable
4/5
Warnings in advance
5/5
Transparent records
4/5
Lessons kept
2/5
Done right
  • The S-1 disclosed enough for investors to see the problem
  • The market rejected the $47 billion price within 6 weeks
  • In 2023 it wrote its own "substantial doubt" warning before the fall
Not done right
  • Grew fast before one building proved profitable
  • Too much power in one person, related deals with the company itself
  • After 2019 kept the lease-long, sell-short structure

The idea was not wrong. The math was: selling short what you lease long, then using a story and fresh capital to avoid looking at that math.

More: why "warnings in advance" gets 5/5

Every number that sank the IPO was in the S-1 itself: losses larger than revenue, about $47 billion of rent owed, voting power concentrated in one person. No investigation or leak was needed. You only had to read it.

The lesson is that for years before, private funding rounds still paid high prices for the same model. The good story ran ahead of the numbers, until the numbers had to be printed for the public to read.

Source for "More": S-1 (Aug 14, 2019), financial data and lease note; amended S-1 (Sep 2019); the rest is Mark's comment.

Part 6

Work lesson: separate a good story from unit economics

For any "go big" proposal: opening a branch, launching a product, signing a long contract, hiring a new team.

Four steps: write the economics of one unit on one page; match the term of money going out with money coming in; ask what an adjusted metric left out; give the person who holds the numbers the right to say no to the storyteller. 1 · one unit, one pagein − out = ?per building, client, order 2 · match the termshow long do we commit?how long do they? 3 · what was left out?"adjusted" metricsvs real profit and loss 4 · someone says "not yet"split the rolesstoryteller ≠ approver Doubling revenue does not answer "does each unit make money".
Each step patches one of WeWork's failures. A loss on every dollar of revenue, leasing long and selling short, a "community" story drowning out the math, and power concentrated in one person.Source: the failures come from the S-1 (Aug 14, 2019) and amended S-1 (Sep 2019); the steps are Mark's suggestions.
One page per unitOne client, one store, one project: money in per month, money out per month, what is left. Without this page, don't talk about "scaling".
Match the termsHow long is the contract we sign, how long do clients commit? If 20% of clients leave next month, who pays the gap?
Ask "what was left out"Whenever you see an "adjusted", "before costs", "excluding..." metric, ask: what is the real number after all costs?
The storyteller is not the approverThe person most in love with the idea should not be the only one signing off. Someone must be allowed to say "not yet".
Check your own team: 5 questions for the next proposal

1. What is our unit (a client, an order, a branch)? How much does each unit make or lose after all costs?

2. How long are we committing to spend, and how long are clients committing to pay?

3. If you strip the adjectives (community, revolution, mission), which numbers are left on the slide?

4. Which costs were left out of the proposal's best-looking number?

5. Who on the team may say "not yet", and when did they last say it?

We have decided to postpone our IPO to focus on our core business, the fundamentals of which remain strong.The We Company's co-CEOs, statement of Sep 30, 2019. Four years later, the company filed for bankruptcy.
Part 7

Read next

The We Company Form S-1 (Aug 14, 2019)Read the summary, then jump straight to the financial data and the lease note. They tell two different stories.
WeWork Inc. Q2 2023 10-QA company writing about its own possible end. The "going concern" passage is short and clear.
SoftBank releases of Oct 23, 2019 and Apr 2, 2020A rescue announced, then partly withdrawn five months later.
Anatomy of a Business Failure #10: FTX (2022)When one person both holds the money and keeps the books. The series finale.
Part 8

Primary sources

SEC filings, the companies' own announcements and bankruptcy records first. Press coverage is a secondary source, labelled as such.

  1. The We Company, Form S-1, Aug 14, 2019: prospectus summary, selected financial data, lease note at June 30, 2019. sec.gov
  2. The We Company, amended S-1, Sep 2019 (voting power 20 → 10 votes, "We" trademark deal unwound, profits from related real estate deals). Read via quotations in Allwork.space, Fox Business, Axios (secondary sources). allwork.space
  3. The We Company, release of Sep 24, 2019 (Neumann steps down as CEO) and statement of Sep 30, 2019 (S-1 withdrawn). Read via TechCrunch, AP (secondary republication). techcrunch.com
  4. SoftBank Group, release of Oct 23, 2019 on the WeWork financing package. copy on sgx.com
  5. SoftBank Group, "SoftBank Group Announces End of WeWork Tender Offer Because Closing Conditions Not Met", Apr 2, 2020. group.softbank
  6. WeWork Inc., Form 8-K, Exhibit 99.1, Oct 20, 2021 (merger with BowX completed). sec.gov
  7. WeWork Inc., Form 10-K for 2021 and 2022. 10-K 2021 · 10-K 2022
  8. WeWork Inc., Form 10-Q for Q2 2023, Aug 8, 2023 (going concern). sec.gov
  9. WeWork Inc., Chapter 11 filing in the U.S. Bankruptcy Court for the District of New Jersey, and announcement of Nov 6, 2023 (statement by CEO David Tolley). The original documents could not be opened for this piece; only the date, court and main point are used, no figures from the petition.
  10. Secondary sources: CNBC (Aug 14, 2019, 527,000 members; the mission in the S-1); Forbes, "WeWork's Rise To $47 Billion, And Fall To Bankruptcy: A Timeline" (Nov 7, 2023); CFN and CFO.com (IPO valuation talk of $10–15 billion).