Hacker News Reader

I thought his AWS comparison was more favorable than it seems you did. I remember launching a startup in 2009 and our head investor (who was a CFO at a bespoke insurance firm) was floored when he saw our financial docs and that we were able to make our infrastructure expenses OpEx instead of CapEx.

As AWS got its start making it easy for startups to get access to infrastructure, the bull case for WeWork is that startups choose that environment over signing a traditional lease. It’s not that a startup can’t work out of a coffee shop, just like we could have theoretically run our own email servers. But a physical location and address has advantages and WeWork opens that to more people with less friction than the pre-WeWork options.

But that’s the bull case. I’m not a financial analyst and I’m not your financial analyst so this isn’t financial advice, but my risk profile tells me to pass on this opportunity...

> [He] was floored when he saw [...] that we were able to make our infrastructure expenses OpEx instead of CapEx.

Can someone ELI5 to me why spending a dollar on OpEx is better than spending a dollar on CapEx?

You'd likely be able to expense the dollar (deduct 100% of it against your taxes in the current year), versus having to capitalize the CapEx (deduct perhaps 20% of it against your taxes in the current year, via depreciation, while carrying the remaining 80% on your balance sheet for the next ~4 years in this stylized example).

The real magic, though, is flexibility: you don't have to bunch all of your server purchases into an annual (or worse) exercise, execute a buildout at a data center, and then somehow allocate them to internal teams consuming them prior to the next capacity planning exercise. You just fire API calls, and a bill arrives at the end of the month. The bill may be a lot of money, but the amount you were spending on planning, building, operating, fixing, and decommissioning your infra was a lot of money, too, split between big spiky equipment purchases and employee salaries with a dusting of leases and electricity on top.

Am I understanding that right, that there are major inefficiencies induced merely by how you have to expense capital purchases?

In short: yes. A $1k monthly expense is a lot easier to book than a $12k purchase that depreciates over time.

And from an investor point of view, if the business fails you don’t have to worry about the sell off / write down of any hardware that was purchased to get the company started.

The simplification of OpEx vs CapEx is worth a decent premium, especially in the early days when no one knows if the company really has a chance and everyone is scrambling to figure out what you’ll be when you grow up (aka product / market fit).

Oh then you'll love finding out how much of the overall economy is guided by tax code idiosyncrasies - read up on stock buybacks vs. dividends for extra horror.

Do not forget that "capital" purchases can also subject to local property taxes (called business personal property). You can imagine a large farm that pays property tax based on the farm size, but a manufacturer with large capital equipment on a small lot would not pay so much property tax. To make sure that all capital is taxed, you have to pay taxes on Capital purchases, at the market value of the item.

If you can lease it instead, the tax burden is placed on the lessor. In Houston, that's around 3% of the value of the property.

The second paragraph beginning "The real magic, though, is flexibility: " describes practical benefits that are not just due to accounting.

Right, but it sounded like that flexibility was at least partly about not needing to buy the computational resources on a rigid schedule, which, in turn, is due to capex accounting requirements.

It’s also never a convenient time to buy a new physical server and set it up in a colo facility while managing DR. Especially when your bread and butter is a CRUD app.

As the article says, with OpEx, your costs scale with your revenues. So when you have $0 revenues, you have $0 costs. You only pay for what you use.

Instead, with CapEx, you go and spend $10 million on a data center buildout (raising a lot of money from investors which is work too) with $0 revenues coming in at the time, that's real money with no guarantee of any return.

For one an OpEx is 100% deductible whereas a CapEx is something that is amortized over time for tax purposes so the former gets you a better tax result

Because it isn't spend a dollar on opex and spend a dollar capex; it's spend $30 on opex to rent a server for a month or spend $1000 on capex to buy a server, plus $15 a month in opex. You can start and scale a business with a lot less capital if you convert to an opex model.

AWS was a novel offering. WeWork appeals to dad followers and people so green that they don't know Regus and others exist. It might be attractive from a SV VC social perspective, but it's not meaningful as an business-model proposition. It's like opening a chain to compete with McD and Wendy's with the value prop "we have an efficient quick-servoce hexagonal burgers". The valuable part isn't new, and the new part isn't changing fundamentals.

Success frequently hinges on execution, not changing the fundamentals. The entire premise of the free market is that market competition rewards those who execute the fundamentals the best. If changing fundamentals was a requirement for success most businesses would fail against the incumbent.

After all, there are plenty of competitors to McDs and Wendy's, most of who are doing quite well even though there's a McDs on nearly every corner and all burgers basically follow the same recipe.

(That being said, I think WeWork isn't executing very fell. I'm partial to Cross Campus if I must use a hipsterish co-working space, and Regis if I want to use a professional temporary working space.

While I'm not a fan of WeWork, it's not just newbs.

They've been useful to allow established companies to expand and "open an office" in a given city with a handful of employees without the time, expensive, and effort of a long term lease. My company has done it at least twice for a year+ before they grew it into a full office.

There were already other suppliers that would allowed companies to do this.

AWS in the early days wasn’t that different than many other cloud providers of the day: you had Rackspace, Linode, Slicehost, etc. And then there’s the oversold shared hosting...

WeWork isn’t a brand new offering; Regus has been around for awhile. But it has mindshare and that has value.

The question is will it be an AWS or pets.com. I won’t answer that for you, but I’m not putting my money into its IPO... ;)

> AWS in the early days wasn’t that different than many other cloud providers of the day

I'm sorry, but this just isn't true. The very first product launched under the AWS banner was S3, which was fundamentally different than anything else on the market. There were tons of companies offering shared and VPS hosting, but nobody offering an API-driven, usage-billed storage service like S3.

A few months later they launched EC2, which was absolutely comparable to the various VPS hosts (if differentiated somewhat by its API-driven use model). But S3 was first, and it made a loud and clear statement that AWS was aiming to build something unlike anything else on the market at the time.

SQS was introduced before S3, but the analogy might still be valid.

S3 was (and is) amazing. But for me AWS as an infrastructure replacement started when EC2 was launched (at least in the context of the current discussion relative to WeWork).

> so this dream WeWork has of gaining corporate partners seems unlikely in the next 5 years, if a recession is to hit as everyone claims. These corporate players will probably have excess office space of their own, never mind paying for more.

What if the corporate partners are partners in flipping excess space in the case of a recession?

This is absolutely spot-on. AWS doesn’t replace email servers, as one commenter suggests. It offers compute infrastructure, a new need that grew alongside a truly novel business. Where’s the novelty of WeWork? Even now Wework has real competition, maybe not from IWG, but from myriad commercial business spaces that are just regearing for co-working — an immediate threat that further damages the AWS analogy.

I dont agree. Its actually the exact same thing. Before AWS you did not time share servers. Servers would be idle for long amounts of time if the capacity requirements were spiky. You have the same situation here, but for work areas/meeting capabilities. Going from a single tenant fully leased capability to a shared workspace model allows for much higher utilization of the underlying assets.

How quickly people forget.

Before AWS, companies did IN FACT time share servers. The whole mainframe model was that IBM owned the server and you only rented a couple of hours on it.

But before WeWork, people also rented office space. WeWork is not the first company to offer rentable office space to workers. They're not breaking new ground here.

> companies did IN FACT time share servers.

Ok. It was so uncommon that I didn't encounter it for the 15 years preceding AWS. It's like pointing to a car that runs on vegetable oil. Yes there are IN FACT cars that run on vegetable oil, but nobody does it statistically. This is much different than the chance that a business (with income) is renting commercial space. Yes your local mom and pop pool company doesn't rent an office (and that's a non-trivial amount of businesses, as I've worked for some), but they aren't going to rent space in WeWork anyway.

>> companies did IN FACT time share servers.

> Ok. It was so uncommon that I didn't encounter it for the 15 years preceding AWS.

That just shows that.. you may not have enough experience. Time sharing was the norm before personal computers.

Some good reading here: https://en.wikipedia.org/wiki/Time-sharing

After the computer revolution of the 80s and 90s, timesharing was reserved for only the largest mainframes (you might see a bunch of people huddled around a large cage in an Exodus colocation). I stand by the fact that it was exceedingly rare.

Sorry, but this was the dominant computing model equivalent to renting servers at AWS. I'm not even talking about "website hosting at $5 per month". If a business wanted powerful machines, you rented them.

Companies didn't own mainframes, generally. Your understanding of "exceedingly rare" is incorrect. The statement that AWS invented the rented server model is also incorrect.

I got my start in web development by renting a shared-hosting account on https://www.dreamhost.com/ .

Shared Hosting is a different model than that being discussed here. Shared hosting never scaled based on usage. To carry the analogy forward that was the same as renting a traditional office space where you had to guess what kind of capacity you needed, and had to limited access to services.

Shared hosting frequently scaled based on usage! They often had limited disk space, bandwidth, and sometimes even CPU cycles, and either you could pay to increase the limit or they'd bill you for overages.

For an extreme example, here's a shared hosting company that's been around since 2002: https://www.nearlyfreespeech.net/services/hosting

That's not really the same thing as the AWS model, thoughm is it? A Linode or Digital Ocean VPS has multiple price levels, but "choose the price level that most closely matches the resources you think you're going to need" is not the same as "metered usage billing."

And Dreamhost predates AWS by a full 10 years.

With a cloud provider you have a lot more behind the scenes control over how an asset is subscribed and provisioned. Now, If we squeeze everyone into a wework with the lights off...

Narrative similarity only works in hindsight. i.e., if A succeeded, and A looks similar to B, which was hugely successful. Then A is claimed to be always having the success factor because it is similar to B.

To predict the success by drawing similarity is always fool's game.