Why WeWork Isn’t AWS and the CEO Problem, Cloudflare’s S-1, Contrasting S-1s

Good morning,

Yesterday I mistakenly said that IWG, the shared office space company that is held as WeWork’s closest competitor, lost £3.8 million last year; apparently I can’t read U.K. financial statements properly, because the company actually made £154 million last year. I regret the error.

On to the update:

Why WeWork Isn’t AWS and the CEO Problem

There are three important points of follow-up to yesterday’s article The WeWork IPO (although two of them are more clarifications of points I was perhaps not clear enough about, particularly the last one).

First, my primary point in comparing WeWork to AWS was to emphasize just how valuable it is to convert fixed costs to variable costs; this not only provides benefits to existing businesses, it also capitalizes on and fuels new business creation. However, the comparison should not go much further than that; there plenty of other important differences between AWS and WeWork.

The most important of these is a core Stratechery axiom, the importance of zero distribution and zero transfer costs. AWS can serve anyone anywhere immediately. All you need is a browser and a credit card to sign up. That means that AWS’s fixed costs are not only shared with all of AWS’s customers but also that that customer base can scale more rapidly and more cheaply.

Real estate, on the other hand, is a physical asset. On one hand, this means that while a single WeWork location can achieve efficiencies for all of the workspaces and offices in that location (shared lobby, shared restrooms, shared beer, etc.), there is no obvious shared efficiency across different geographic locations; the company claims that its culture and technology do scale in this way, but I’m skeptical about much benefit beyond brand (which, by the way, will likely lose prestige the more that WeWork expands).

Real estate also has significant transaction costs: not only do companies have to actually move in, WeWork reportedly sources a good portion of its business from brokers to which it pays hefty commissions. These are typically 10 percent of the first month’s rent, but WeWork has at times gone far beyond that. From Bloomberg last year:

This year, WeWork is enlisting another party in the battle: brokers. The company is offering commercial real estate brokers worldwide a 100 percent commission on the first year of rent paid by any tenant who switches to WeWork from a top competitor and signs a lease by October 1. Tenants also get half off the first year’s rent if they sign for at least 12 months. That means, accounting for the discount, that WeWork’s current bonus to brokers is five times the standard commission it typically offers of 10 percent on the first year’s rent.

This gets to the second point of follow-up: much like its free-spending ride-sharing peers, WeWork does not provide nearly enough clarity around unit costs, customer acquisition costs and lifetime value. Exactly how profitable is one workstation? Does that calculation include the cost to fill that seat, including brokerage fees? What is the lifetime value of a member? It would also be useful to have a breakdown of properties by cohort: if locations are profitable after 24 months, why not show that explicitly?

That leads to the last point: I just don’t trust WeWork’s leadership, and the lack of concrete numbers does not help matters. There is too much self-dealing, too little disclosure, and a far too believable story of CEO Adam Neumann convincing private investors to give him money through sheer charisma in a way that will be impossible with public markets. Add on the $700 million he has taken out of the company — hardly a vote of confidence — and I personally would not invest (Per my ethics policy I don’t make investment decisions about individual stocks).

That doesn’t mean the bull case doesn’t exist, and I think it was well-worth describing: so many of the factors, from excess capital to new kinds of businesses to remote work, are driven by structural changes wrought by the Internet. The problem may simply be that WeWork isn’t Internet enough (and seriously, that corporate governance is a joke!).

Cloudflare’s S-1

From the Financial Times:

Cyber security group Cloudflare published plans on Thursday to list on the New York Stock Exchange, joining the raft of Silicon Valley initial public offerings just days after being embroiled in a debate over whether it should offer services to fringe internet forums.

The San Francisco-based company — which provides security for clients’ web services, including protecting them against distributed denial of service attacks — said in its filing that revenues rose around 50 per cent in the first half of 2019 to $129m from $86m the previous year. Net losses over the same period widened to $37m from $32m year-on-year.

I’m not going to get too much into the Cloudflare financials; Alex Clayton, as he so often does, has an excellent breakdown here. I agree with his conclusion:

Cloudflare is the leading cloud-first CDN and network services provider and has major trends moving in their favor — the move to cloud infrastructure and the increased need for secure and performant applications/websites. They also have a fairly efficient business and are ~40% bigger (and growing slightly faster) than their nearest competitor, Fastly, on an LTM-revenue basis. I suspect the public markets will look past the operating losses given their revenue growth, market leadership, and fairly efficient, self-serve model. Similar to essentially all the previous high-growth SaaS IPOs over the past few years, Cloudflare should fare well in the public markets.

Actually, while I agree with Clayton’s overall optimism about Cloudflare’s financials and growth rate, I think this conclusion undersells Cloudflare’s potential going forward, particularly in terms of laddering-up.

Cloudflare’s explanation of its core offering in its S-1 goes like this:

  • The Internet was designed to be decentralized and resilient, not secure and fast.
  • In response a number of vendors (VPNs, firewalls, routing, traffic optimization, load balancing, etc.) arose to provide on-premises hardware solutions primarily for corporate networks.
  • The rise of cloud computing and smartphones made it vastly more complicated to secure these networks and applications, rendering on-premises solutions obsolete.
  • New security solutions need to be built with cloud and mobile in mind, which means the solutions themselves must be cloud-based and distributed.
  • Cloudflare has built the leading cloud-based and distributed security solution.

This alone is compelling (and is what Clayton was referring to in his conclusion); Cloudflare appears to have significant growth ahead of it, particularly with large accounts that make up an increasing portion of their business.

What is interesting is that in the process of creating this massive distributed network Cloudflare, well, has a massive distributed network. That may sound like a truism, but keep in mind Cloudflare’s network is not made up of special-purpose hardware, but rather generic commodity hardware that provides all of these security solutions via software. That, though, means the Cloudflare can run other kinds of software too — and it doesn’t have to be their own. One of the most intriguing Cloudflare products is called Cloudflare Workers. From the S-1:

Increasingly, developers outside of Cloudflare seek the same flexibility and performance afforded by our serverless architecture. In response, we opened our platform to outside developers with a product called Cloudflare Workers. With Cloudflare Workers, our developer customers can write their own code and deploy it in seconds directly onto our global cloud platform and have it run close to their users.

This is not merely a configuration language, but rich, complex code written in the language of choice by our customers to power sophisticated applications. The efficiency of our platform allows us to offer Cloudflare Workers at prices that are significantly below traditional public cloud computing providers while still maintaining an attractive gross margin. Moreover, applications utilizing Cloudflare Workers exhibit global scalability and strong performance relative to other cloud computing vendors. We expect the use of Cloudflare Workers to grow and to empower an entirely new class of applications.

If, as the saying goes, you can measure greatness by the number and stature of one’s enemies, then Cloudflare’s competitive set is good news indeed: not simply on-premise hardware network vendors, and not simply cloud-based solution vendors, but even the biggest cloud players of all: AWS, Azure, Google, and Alibaba. The company used its disruption of the first set to build the leading contender of the second set, and in the process laddered-up into the scale necessary to compete with the third.

Contrasting S-1s

What makes Cloudflare’s S-1 particularly striking is the stark contrast it draws to WeWork’s filing. Cloudflare is a classic software company, with high gross margins (77.4%), expenses that are dominated by Sales & Marketing that tracks revenue growth, negative churn rates (i.e. the total revenue of a cohort increases over time as increased spending outpaces churn), and an obvious path to profitability presuming growth continues.

WeWork, on the other hand, has gross margins of only 14.4%, and that number is negative if you include depreciation and amortization, as you should (and as Cloudflare does). It has two line items called “Pre-opening Location Expenses” and “Growth and New Market Development Expenses” that are in addition to Sales & Marketing, rival it in size, and are growing as a percentage of revenue. And, while WeWork has a negative churn rate in terms of number of seats per company, the average price per seat for WeWork as a whole is declining (whether this is due to geographic expansion or falling prices is impossible to determine). And, of course, even if — especially if! — WeWork continues to grow it is not at all obvious the company will be profitable.

The issue, as I noted at the beginning, is that WeWork is not truly a software company, and nothing makes that clearer than contrasting it to Cloudflare.

And, I should add, per the concerns I noted both yesterday and today, there is the question of leadership. While many of you may disagree with how Cloudflare CEO Matthew Prince has handled the various controversies that have befallen Cloudflare, the overwhelming response I received to the interview I did with him two weeks ago was an appreciation for the thoughtfulness and seriousness with which he approached those decisions.

To be fair, I have not spoken with Neumann, and Cloudflare executives have preferred shares and have sold stock on the secondary market. At the same time, the amount of self-dealing conducted by the former remains startling; at best it suggests a lack of judgment and thoughtfulness (I described the “at worst” yesterday as “looting a company that is running as quickly as it can from massive losses that may very well be fatal whenever the next recession hits.”). The contrast is sharp indeed.


The Daily Update is intended for a single recipient, but occasional forwarding is totally fine! If you would like to order multiple subscriptions for your team with a group discount (minimum 5), please contact me directly.

Thanks for being a supporter, and have a great day!