Snap Committed to Spend $1B on Web Services with Amazon | Hacker News Reader

Snap Committed to Spend $1B on Web Services with Amazon

bloomberg.com

48 pointsrobertgk50 comments

$3B over the next five years. Rackspace was worth around $4.6B in August with assets of around $1.6B. Assuming that Snap would require as big an infrastructure as Rackspace, then purchasing Rackspace for $4.6B was surely a better deal, because that's the amount their GC + AWS + Raw Hardware bill would cost, if they decide to build their own infra. Given how much AWS was earning a year back while still being massively profitable, these services cost a few hundred million a year to run at worst.

I have a strong feeling that this has a lot to do with available human capital to build and sustain these services. It even took Rackspace years to have a "Cloud" offering and it was essentially Slicehost with blings.

Come to think of it, it is around $4 per user, per year. With the way the landscape is with everything shifting to (live) videos, it probably is not that much an expenditure. Especially given that Snap will have several times the users in next 5 years, so the cost will look even less. That, and inflation.

Except that's entirely unrealistic as an actual business move. Coming up with $600m per year for operations, is very different for Snap versus coming up with $4.3 billion all at once to buy another business. This is especially true given Snap's (currently) anticipated revenue growth rate.

First, the people funding Snap - the people providing the capital to buy Rackspace - would not look well upon them trying to buy Rackspace. They'd think it was a huge distraction. Snap has some capital now, but they require that for operations, so they'd have to go out and raise a new $4.3 billion for the deal (and they'd suffer a painful dilution because investors would balk).

Second, you can replicate the useful (to Snap) infrastructure of Rackspace, at a fraction of the cost of buying the business as it was valued based on a relatively high multiple.

A year from now Snap should be able to hit at least $2 per user on a quarterly basis, with 200 million or more daily active users (both conservative numbers). At $400m quarterly in sales and growing, versus the ~$150m Google / Amazon costs (which might be lower in the early years of the deal?), it's a bad business move to take a 15% to 20% dilution hit to purchase the overvalued Rackspace business. That bad value ratio gets more and more lopsided if Snap keeps growing in the coming years.

Buying Rackspace to run your infrastructure on? That makes no sense. Customers are using those servers, you'd have to kick them all off in order to reprovision them for your own use.

Otherwise, you'd still have to buy new hardware under the Rackspace moniker.

$1B at AWS, $2B to Google Cloud, ... eventually this adds up to real money.

It only cost SpaceX $1.2 billion to develop a reusable launch vehicle.

and at the end we'll have either one of the coolest pieces of engineering and technology ever, or...throwaway pics.

NB: I really hope Snap ends up doing some really cool stuff with VR/AR and makes my comment seems stupid in hindsight. Maybe if they could live-stream in full 360 VR the launch and landing of a space X rocket...

For context: The US Broom, Brush, and Mop Manufacturing Industry was $2.8B in 2013 [1]. Not everything is as sexy as launching rockets to outer space, but as long as people derive utility from throwaway pics there will be demand for billions of dollars of web services.

[1] http://www.businesswire.com/news/home/20130301005429/en/Rese...

Pretty amusing to compare snapchat to SpaceX.

My argument is by design. Even if SpaceX isn't profitable, for far less than what Snap is spending on infrastructure, they're delivering payloads to orbit. smh.

I don't mean profitability - I just think it's funny to compare a company that makes an utterly unnecessary toy to a company trying to turn us into a multi-planetary species. And the fact that they spend so much more money is even more amusing.

> utterly unnecessary toy

I don't get why people dismiss anything that isn't immediately pragmatic as useless. Snapchat is an addicting communication tool that motivates people to share and stay in touch. Isn't one of the core tenants of advancement as a species cooperation and affection? Isn't improving the quality of our personal relationships something worthwhile to pursue in its own?

Most people dismiss snapchat interactions as fleeting and shallow, but since when are close relationships built on 100% esoteric mind melds? Getting a giggle out of my girlfriend with a silly emoji can still strengthen our bond

>Snapchat is an addicting communication tool that motivates people to share and stay in touch.

So is MySpace, so is Facebook, so is twitter.

>Isn't one of the core tenants of advancement as a species cooperation and affection?

There isn't much evidence that tools to help you isolate yourself into echo chambers are advancements at all.

>Getting a giggle out of my girlfriend with a silly emoji can still strengthen our bond

Somehow couples were communicating before snapchat. How do you think people got by before they could send pictures of themselves with oversized tongues and dog ears?

> How do you think people got by before they could send pictures of themselves with oversized tongues and dog ears?

Those were dark times indeed.

Some people even had to resort to using physical means of communication, namely paper and ink. To make matters worse, it left open the possibility that the receiver of your communique may not dispose of it properly. I've heard reports of people keeping such "private messages" around for 30, 40, sometimes 50+ years!

It's worse than that. Snapchat is clearly very pragmatic to its users. There are few things more important to people than communication, and that will always be true. Might as well proclaim the telegraph, telephone call or email to be a ridiculous toy (except Snap is far superior to any of those given its capabilities).

It's the same premise that has people shake their heads in outrage at sports team valuations. It's worldview bias, projecting onto other people what one's own values are.

Communication is obviously not unnecessary, but Snapchat could not exist and everything would be pretty much the same.

Snapchat is clearly in the entertainment category, competing for eyeballs, where the value they add over the next best option is small compared to the rewards they get for being able to advertise to those eyeballs.

This is a problem with most of capitalism but is much more pronounced in ad supported models since there's no way for a slightly inferior product to get an edge by being slightly cheaper.

What makes delivering payloads to orbit qualitatively better than sending pictures to friends?

If snapchat is 100% successful in their current endeavors, you still get to send poop emojis to your friends down the street.

If spacex is 100% successful, you get to send poop emojis to your friends on another planet.

> If snapchat is 100% successful in their current endeavors, you still get to send poop emojis to your friends down the street.

I don't use snapchat yet I still send poop emojis to my friends down the street.

Whether they succeed or fail will have no impact on my poop emoji sending capability.

Who knows what's really going on behind the scenes. Considering the vast amount of information present in images and videos, they can extract all kinds of valuable data from it. Reconstruct user activities, brand preferences etc.. or improve algorithms for object recognition, spatial reconstruction or improvements of camera quality to make them cheaper, etc.

There is a lot you can do with it - if you have the computing power.

Why would they go multi-cloud? ( I'm asking because even very large infrastructure don't use 2 providers )

From a management / ops perspective it's a nightmare to manage and imo adds more problems than it solves.

I think redundancy would be the main reason. They also can use that to leverage for better pricing as well.

I think it's more likely the second reason.

Could be different micro-services. I'm not at Snap's size, but it's common for me to use Google App Engine (which Snap uses) and then also dedicated bare metal servers for fast data processing. What's more, Snapchat predates Google Compute Engine, so it may just be a case that they have some micro-services designed to run on AWS and haven't bothered to move it elsewhere.

Last week everybody was shocked that they're going to spend $400mm/yr with Google... Now the total cloud bill is up to $600mm/yr! I can't wrap my head around these numbers.

You'd probably also have a hard time wrapping your head around their traffic and usage.

I'd imagine so. I'd be curious to see the numbers on their infrastructure. What $600mm is paying for.

and their lack of profit.

At least that can be forgiven if you have accelerating revenue, which can't be said about many startups with big valuations.

User growth is flat though. That would concern me in regards to how much more they can accelerate their revenue in comparison to costs.

I don't think that all the computing power is just for powering their app. They're probably going to extract valuable information out of the vast amount of visual data that is streamed to their servers 24/7.

Let's grant them a period of huge growth and say they go from current 160 million users to even, 320 million after 1 year.

As others point out, they are spending $600 million per year between Google and Amazon. Do they make more than $2 per user per year?

Yes. Snap is at $1.05 per user (daily actives) in the most recent quarter, up from $0.31 the year prior. That's just for the quarter. They can probably hit at least $5 or $6 per user for all of fiscal 2017.

Facebook is at $7.16 per user in the most recent quarter (based on daily actives; it's $4.62 based on monthly actives), and Twitter is at around $2.23 per user based on monthly actives in the most recent quarter (I can't locate recent daily active numbers).

The math here is awkward to visualise because the numbers are multiplying, but does your $5-6 for 2017 overall require them to be above Facebook's current $7.16 in at least Q4 '17? That seems really ambitious, but I might be reading this wrong. For reference, could you give a ballpark of where you see Facebook's $/user for 2017?

I'm really curious what their bill looks like. 1.000x m3.large, petabytes of S3 storage, loads of traffic.

I can't imagine what the numbers are. Is there any big player which has public numbers about their infrastructure? (I mean in detail - which services, how many instances, traffic, ...)

I am very curious why they didn't azure as their cloud partner ? What was wrong about it ? (I am geninunely asking, I am really curious about technical capability of azure vs aws vs gcp vs oracle)

In my experience, Azure is a complete pain in the ass. It's just always something with them.

Yesterday for instance we finally got to the bottom of a slow page problem. The issue? Azure CDN was stalling up to 2 seconds at times. Creating a new CDN profile fixed it.

They're expensive, way more than GCP. It's slow - everything is just slow.

I know several companies that chose to pay GCP vs use free Azure credits because it's just "problematic".

And I say this as a fairly big supporter of MS overall, and I really really wanted to like Azure.

GCP just blows everything out of the water. Though it would be nice if they had a Aurora or Postgres RDS equivalent. Automatic DNS integration with GCE would be handy, too.

The primary reason would probably be that Azure was the unproven new kid [1] on the block at the time these decisions were made at Snap.

[1] Yes Google Compute Engine (GCE) is even newer than Azure, but Snapchat used (and probably still uses?) Google App Engine not GCE, which didn't even exist when Snapchat started.

If the Spectacles and other similar things really catch up, they might need some storage space and computing power.

https://www.spectacles.com

I'm curious, at that scale why don't they build their own infrastructure? Surely it would cost less?

My guess is that the human capital they need for such a build-out is hard to come by and mostly already working for Google Cloud and AWS.

It's not like they don't have the option to build it themselves in the future. This provides a lot of flexibility with very little liabilty to get their platform massive.

Besides they aren't like Facebook where they will grow their hardware slowly as they ramp up, Snapchat is already incredibly popular. It would be a massive undertaking to move everything to the data center. The experts who are good at that kind of thing are all already gainfully employed.

> This provides a lot of flexibility with very little liabilty to get their platform massive.

This is completely untrue. They are committed to spending that money with at least Google (whatever they don't use in infra, they have to make up with a true up payment). I don't know the AWS terms, but they would have to be successful at a level that is completely unrealistic for this to be financially sound (considering their growth numbers show users have tapered off).

https://twitter.com/dannysullivan/status/827274299929026560

"Snap’s daily active user grown went from single digit to flat last year"

If user growth is flat, and you're not yet profitable, the only nob you can turn is more ads or making ads more expensive for ad buyers. Neither option will endear Snap with either party.

It's much easier to go belly up with chapter 7 and not have to sell millions and millions (billions?) in physical assets then to simply say "sorry, we're out of business" to the contracts for future obligations.

It costs time, and results in instability for their user base in the interim.

They could do, and it might save them a few hundred million dollars, but the risk of project failure (lateness, budget overrun, etc) would be a lot higher. For example, committing $1bn to AWS with with a 5% risk of failure is a lot more sensible than committing $750m with a 25% risk of failure doing it yourself. It's entirely possible Snap's board would have refused to let them do it themselves if the risk was high enough.

I doubt their concern is saving money in this case, it's more about time to market for new products/features. If they can get a head start by spending more up front and focus entirely on their product, they may be able to grow significantly faster than if they kicked off a multi-year effort to role their own infrastructure.

I don't think they could build their own infrastructure cheaper than Google or Amazon. Both companies have very large scale and have been doing this for quite a time, have custom hardware etc.

For deals of this size, it might make sense for both Google and Amazon to put their margins quite low - as they know the alternative is that Snapchat just builds their own infra. For Google there could be also the PR factor involved.

It would definitely cost less, and datacenters are readily available, already running (i.e. zero actual project risk).

My guess is that they hope to use Google's TensorFlow and/or a lot of the GPU instances from AWS - TensorFlow needs a special chip and GPUs become outdated in 1 year, so it makes more sense to lease them as-needed.

can anyone give a comparable to how much Facebook/Google/Twitter spend on cloud infrastructure per user per year? these numbers are harder to find as they build their own datacenters/staff them/etc.

Reply on news.ycombinator.com