Hacker News Reader

Comment by matthewmacleod | Hacker News Reader

Fascinating that IBM's entire lower-end server business is worth about two thirds of Snapchat.

I certainly agree that Snapchat is over-valued.

The problem with being in a mature market is you can't convince investors that huge growth is just around the corner, so they start looking at things like 'profit' and 'dividends'.

And when you have HP and Compaq and Dell producing basically the same thing, you can't charge much of a premium as your customers can easily switch suppliers.

Compaq does not exist anymore.

That only serves to reinforce my point :)

Let's say s/Compaq/SuperMicro/ then

That's because social websites are in the business of taking money from speculators and unsuspecting investors. The valuation doesn't represent their revenue-producing capacity, just the fact that they can be traded by speculators for the (remote) possibility of becoming the next Facebook.

Well, in part though these social networks can theoretically be worth the amounts they state - that is if they're not factoring competition and that users are mobile and can easily start using another service very quickly. This is why governance of the ecosystem is the next competitive advantage layer.

LINE, a free messaging app in Japan, made $194m in a single quarter, and its high margin business continues to grow fast. It's monetized through virtual goods and partnerships. And yet, it's nearly unheard of in the US.

I'm not sure if SnapChat is a winner and if their multi-billion dollar valuation is correct, but there's there's still whitespace in the market and plenty of money to be made.

http://www.techinasia.com/line-app-financials-q3-2013-sees-r...

LINE makes 80% of its revenue from stickers (20%) and games (60%). I can't see stickers as a workable revenue model outside of Asia. As for games, it may work, but I don't think Westerners like the idea of integrating their games into their messaging apps/services. After all, Facebook tried this with games, but nowadays people prefer to play standalone game apps on their mobile devices.

Stickers are as crazy as hats.

And one of LINE's core business is being a (mobile) video game storefront.

For those who did not get the reference, Valve paid out 400K$ in the first week of 2014 [1] to item creators for Team Fortress 2 and Dota 2, which are purely cosmetic video game items ("hats") [2].

[1] http://steamdb.info/blog/43/

[2] http://ca.ign.com/articles/2013/04/16/how-to-make-a-living-s...

> I can't see stickers as a workable revenue model outside of Asia.

Asia is a HUGE market.

It's also notoriously difficult to break into (in the software space) for Western companies. With Japan, there's the isolationist culture, and with China, there's the Great Firewall.

I'm thinking less about Zynga-like in-browser games, and more like Candy Crush (solo game but it's arguably more fun with others, and def faster/cheaper) and Quiz Up (obv a two person game). For the same user, they'll have different friends lists on SnapChat vs Facebook, and may want to play games more with former vs the latter.

Agree about sticker, but other virtual goods are open. Farmville was rumored to be making $1m a day, and I believe it. Who knows what other iterations we'll see.

Disagree with stickers. Everyone I know uses them constantly. A friend of mine has the cat with birthday hat as a tattoo.

What is sticker in this context?

LINE is big here (Uruguay, South America), it's not used in the U.S.?

Not really, nope. Very small user base here comparatively. LINE and KakaoTalk were huge in Korea when I was there last but when I came here no one had even heard of them.

I believe that Whatsapp and Groupme are the two popular apps in the U.S. I personally use both, but prefer Whatsapp and use Groupme only because my workplace does.

WhatsApp is really big in some European countries, especially Spain. But apps provide a non-exclusive messaging market, so I suspect people can fairly easily switch to alternatives.

WhatsApp is huge here as well, but many people started using LINE after the 1 dollar scare (99% of users here literally can't pay because of the lack of international credit cards).

Yeah, I don't see it. This is Dot Com Bubble 2.0. Nothing but hype. I don't care how many teenage eyeballs SnapChat has. You don't suddenly become Tencent or LINE or whatever tomorrow's out-of-the-ass comparison will be. SnapChat is going to have an uphill battle moving their users to anything that is not a simple picture transferring app. And they only have so much time left. If there is one thing teenagers do not do, it is use something past its trend expiration date. The fact that SnapChat hasn't been monetized yet probably means it's already too late and they have no clue where they are going with it, other than hoping to get bought out for an even more outrageous sum of money.

The challenge for SnapChat is to shift from novelty and entertainment to utility. Facebook was fun early on, especially when adding friends and reconnecting with old ones that you hadn't talked to in 10+ years. However, it's shifted to more of a utility for sharing photos and keeping loose tabs on family and friends. SnapChat needs to bridge that, and it can enjoy life far beyond a fad.

> SnapChat is going to have an uphill battle moving their users to anything that is not a simple picture transferring app.

SnapChat is kind of being used as a text messaging app already. That's been my observation of young users, and Bryce Roberts mentioned this as well: https://twitter.com/bryce/status/424600001868599296 (not that a VC has a crystal ball, but it summed up my thoughts.). I think older folks have written if off as an Instagram substitute and the young crowd is only interested in it because it's trendy.

I realize there is a bit of snark in that comment but the interesting bit is known as COGS. Part of the "value" of a business is the rate at which it can make money, and the server business is limited in its rate by how many machines you can build and the price you can sell them for. Most internet businesses (like Snapchat) are limited only by the amount of money you can extract from users and the number of users you can sign up. While the sunk costs of hosting servers and infrastructure is non-zero, the marginal cost of an additional customer is tiny compared to the marginal revenue they could generate.

This is the fundamental way in which the information economy is different than the goods economy. And that makes comparing a business in one, with a business in the other, invalid.

And just how much money is Snapchat extracting from its users to justify 3 billion?

Well this estimate http://techcrunch.com/2013/11/19/how-many-users-does-snapcha... is that snapchat probably has at least 10M users. So let's say you wanted a 10% return on your $3B value, you would have to generate $30 in net income per user per year ($300M). If you charge a 'membership' fee of $3/month [1] that gives you $360M a year, leaving $60M/year (or $5M / month "burn" rate).

So would people pay that? I don't know. But it would be the sort of math you would go through evaluating that deal.

To evaluate the server deal you have to look at how many units you think you are going to ship, what your COGS are on them, and your gross margin. You can't really give someone 'free server' for a year because you have money sunk into building it and shipping it.

Further your revenue probably doesn't change if 100 people use a server or 1000 people do.

My point remains that it is not possible to compare 'goods' economy companies with 'info' economy companies.

[1] [Edit: added the RPM rates] Conversely, in advertising terms with 400M snaps per day you would need a RPM of $2.50 to generate $360M/year in revenues.

I highly doubt SnapChat's audience of majority less than 18-year-old population (with most in their early teens) are going to pay $30 a year for a way to send auto-deleting pictures to each other. Especially when there are numerous other free multimedia messaging apps. Getting even 1% of that audience to pay would be impressive.

An RPM of $2.50 is also equally unrealistic for an audience of the mostly non-credit-card holding teen demographic, even more so when they're looking at the autoexpiring content for mere seconds before moving onto reply. Not only that but I'd bet a lot of those snaps per day come from a small subset of the 10M users.

Less than 18 turns 30 in 10-15 years.

And chances of Snapchat existing in its current form in 10-15 years are microscopic.

> So would people pay that? I don't know.

I know. People would not pay that.

The same way you know IBM small server business would be worth more than an app?

Common sense no longer applies here.

No, I have no idea about the relative values of server businesses and apps. But I know that very few people will pay $3 a month for Snapchat.

So lets say you take it as a given that snapchat is mostly teens. They look at their snapchats, there is a way to know they have seen them. Lets say you are the GAP and you want to get the latest "look" out there, so you pay SnapChat $.05 every time Snapchat shows a Snap of someone in the "look". Snapchat has to show 50 "Gap" snaps in every 1,000 to make $2.50 per thousand. Typical user will see them on one in 20 'snaps', if the average really is 30 snaps looked at in a day, that is one per day.

I'm not saying you are wrong, what I'm saying is that I can imagine how they would make it profitable should they need to.

Sure, what you're describing is feasible. I was referring only to the idea of end users paying for the Snapchat service.

What I read out of your comment is that though comparing both kinds of companies is invalid, valuations on information economy companies is at present nonsensical at best.

Not only that, but nobody forked over billions for Snapchat. The company took investment at a price that suggests we can extrapolate a total valuation in the billions, but nothing like that amount of money has actually changed hands. Lenovo is putting up $2 billion in cash for this deal.

Snapchat turned down a $3 billion offer from Facebook. That's what the snarky comments are about.

Again, that's not cash on the barrel head. That's a rumor. Was it $3 billion in Facebook stock? Was it $1 billion plus a performance bonus? Would Snapchat continue to operate independently, or was it a huge acquihire?

We know what IBM's server business is worth. Snapchat's valuation is just speculation.

To be fair, the lower-end of IBM's server business is much like dell; they don't actually manufacture any of the complex bits, they just sell them. Your CPU is made my intel, your ram by Samsung and your raid chipset by LSI, just like if you bought a SuperMicro; In fact, it wouldn't surprise me at all if SuperMicro did more of the board layout work in-house than IBM or dell.

That said, the IBM has... much more market share than I thought in the server market[1] so even if it's mostly a sales game, it's not a small sales game.

[1]http://www.forbes.com/sites/chuckjones/2013/08/29/ibm-regain...

"IBM's System z mainframe running z/OS experienced its third consecutive quarter of growth, increasing revenue 9.9% year over year to $1.2 billion, representing 9.8% of all server revenue in 2Q13."[0]

IBM makes some big systems. To me, it's not terribly surprising they'd be willing to part with their lower-end units. Mainframes are where the money is for IBM.

[0]http://www.idc.com/getdoc.jsp?containerId=prUS24285213

>IBM makes some big systems. To me, it's not terribly surprising they'd be willing to part with their lower-end units. Mainframes are where the money is for IBM.

Yup. even if that's only 10% of their revenue, I bet it's a lot more of their profit. Margins on assembling servers are driven down by massive competition, and by the fact that it's really difficult to 'differentiate' x86; your product is the same microchips as the competition on very similar boards in very similar metal boxes. (Which isn't to say the margins are zero, it's just that the customer has a lot more negotiation ammo. Your sales person has to work a lot harder.)

This situation is very good for the consumer, of course; they can reasonably play one supplier off of five. But it isn't as great for the suppliers.

The value is probably closer to reality. Don't know what is Snapchat really, but other hip online whatever (instagram, tumblr, facebook, ...) are insanely overvalued for some weird reason.

I thought Marc Andreessen's take was pretty plausible:

"""The bull case on Snapchat is that there's a company in China called Tencent that's worth $100 billion.

And Tencent is worth $100 billion because it takes its messaging services on a smartphone and then wraps them in a wide range of services—things like gaming and social networking and emojis, and video chat—and then charges for all these add-on services.

And it has been one of the most successful technology companies of all time and is worth literally $100 billion on the Hong Kong Stock Exchange.

Maybe that's [CEO Evan Spiegel's ] plan. Maybe Evan's plan is to transplant the Tencent business model into the U.S., which nobody has actually been able to do yet."""

IBM is already IBM, and they're in a market with diminishing margins. Snapchat... Who knows?

Not to mention that the server business brings in $5 billion in actual real revenue yet is being sold for 1/2.

Revenue matters, but the valuation is driven by the amount of profit. Margin in the server business must be very low (5%) and hence the valuation is a multiple of profit and not revenue.

I'm sure the profit margin is low for IBM but will be workable for lenovo, but startup valuations are based on rainbows and unicorns.

do you have a source for the 5% number? I mean, I agree that x86 servers are a low-margin business, but 5% is lower than I'd expect.

I should have qualified that number right from the beginning as a back-of-the envelope estimation:

If we assume a 10x multiple to profit and a 5bn revenue, it results in 5.4% margin. If we just say it was a 8x multiple, then it is 6.75% It goes to 7.7% margin if we make the multiple to be 7x

Consider IBM is lucky to fool Lenovo into it. It puzzles me what's in the world has Lenovo been thinking to spend that much money on that. Different from thinkpad, x86 server business is already doomed. Might Lenovo see a private data center future than public cloud dominance?

How is the x86 server business doomed? Do you think there will be an abrupt shift to ARM servers?

I believe they are referring to a shift to cloud-based servers that usually aren't stock IBM/HP/Dell, etc... machines, but rather custom builds for the cloud.

I wouldn't go so far as to say doomed, but it will be a shrinking business until is his a steady level.

Lenovo has done a pretty good job with the PC/workstation side of IBM that it bought a few years ago. There's no reason to think that they wouldn't do just as well with the servers. Just because IBM doesn't want to spend time for the small (to them) profit in servers doesn't mean that there isn't some real money to be made in that market.

Huge cloud providers like Amazon do utilize custom built servers, but most enterprise "private clouds" will still be buying Dell or HP servers.

We use computers to help us solve so many more problems today then we ever have before and we have much higher expectations of IT-system today. So, the metal scraps used for this keeps piling up in data centres all over the world.

Anecdotal observation: Many of my clients avoids using cloud services often cited reasons: cost and worries over ownership of data.

Based on the my observations my prediction is that traditional server vendors will keep on growing their business for many years to come but it's very likely that cloud services such as PaS/IaS will grow at faster pace ...

Do your clients have highly variable workloads? I'm curious how cloud hosting can be price competitive with dedicated or colocated servers.

It's hard to make copies of users, not so hard to copy hardware.

Valued, not worth.

Same as Tesla being worth one and a half Chryslers.

Tesla definitely has more potential upside than Chrysler.

That's definitely debatable. Chrysler sells more cars in a modest town than Tesla does in the entire world.

The question is what is the longer term stability and volatility of that value? In both cases. Also, how much of that value is perceived and how much is calculated.

I came here to say exactly that. It's absolutely crazy.

Eh, Snapchat is worth more than half of Sun Microsystems.

Snapchat is better than IBM's entire lower-end server business. I know this is not a popular opinion, but I think what Snapchat has done is spectacular and what IBM has done is pretty meh.

Do you have particular insights into IBM's low end server business or is it because you use snapchat and have never used an IBM server?

Walk in to any datacenter. HP and Dell absolutely dominate. Third place is Sun. Fourth is Supermicro.

Very, very infrequently you'll see an old IBM server.

>Walk in to any datacenter. HP and Dell absolutely dominate. Third place is Sun. Fourth is Supermicro.

This is also what I thought. (though, places I've worked, ODMs and supermicros dominated, followed by dell, with hp dead last, but yeah, the quote from IBM never even made it to technical evaluation where I would see it.)

However, I went looking for a source[1] before I posted further up this thread. Apparently, IBM sells more servers than dell or hp (but not more than both put together)

Hm. reading further, I see numbers between 10% and 30% of IBM's server revenues being the high-end stuff they are keeping. So that could bump them below both dell and HP, in terms of just x86 servers, but even knocking 1/3rd off, they still sell a lot more x86 than I thought.

[1]http://www.forbes.com/sites/chuckjones/2013/08/29/ibm-regain...

iDataPlex clusters are somewhat more common in HPC environemnts. They are x86-based.

I'm not sure where you make up your data from, but if that's the case, you might be going into a lot of low end data centers.

http://slashdot.org/topic/datacenter/idc-server-market-slowi...

Obviously my data is anecdotal, but I really haven't seen much IBM hardware deployed in communal datacentres (e.g. Peer1 or similar).

If IBM's selling that kind of volume, it could be to governments and other large enterprises with private datacentres.

Then again, that's revenue, so for every IBM server there could be five commodity Dell units in the rack and they'd have the same net cost.

This is true, and I believe you are exactly correct. IBM is catering to private data centers and not ones that you or I would be renting a rack in.

I evaluated using their servers and saw nothing special at all in the market place from IBM. I see Snapchat as something special.

Snapchat is an entertainment product. It could go away tomorrow and the world would be just fine. On the other hand, losing all our server products would be catastrophic. Fortunately, this will never happen because servers are important. Should they become scarce, they would quickly become a very lucrative business resulting in an increasing supply.

IBMs servers may not be as good as the competition, but to compare them to an entertainment product is silly.

You can easily replace an IBM server with any other comparable x86 machine. You can't easily replace Snapchat and the user networks it hosts with another similar service.

My point is that as a class of product, Snapchat could disappear and not be replaced and no critical infrastructure would halt nor any innovation or science be stopped. At worst you would have some grumpy people.

It's silly to compare this to a product class that is crucial to our society and claim that it is better. The relative worth of one server to another is one thing, but the class of server products is infinitely more important than the class social picture sharing products.

Would you also say Facebook could disappear tomorrow without "any innovation or science be stopped" ? That seems like a silly way of determining if something has value or not.

That's ridiculous - you just said "Snapchat is an entertainment product, it could go away tomorrow and the world would be just fine" - IBM's servers could go away tomorrow and the world would ALSO be just fine - unless you LITERALLY mean they disappear without time to replace or back them up, but I don't think that's your point.

If ALL Entertainment products were gone tomorrow, the world would NOT be just fine it would be SHIT. Just like if ALL servers went away, which is what you jumped to.

Snapchat, incidentally, is not an entertainment product anymore than texting is an entertainment product. It's an entertaining communication product.

I'm not immersed in either business but I think you're making your argument too black-and-white. Snapchat is doing something valuable (I don't exactly know what) but I do not think it's better than what IBM is doing, IBM has been a powerful innovator in the computing industry for a long time and there's a lot of infrastructure powered by IBM's technology, even indirectly. I'm sure Snapchat is indirectly influenced by IBM - so writing either one off as better or worse is silly I think.

Different? Definitely. Providing value in totally different but important wasy? I'm sure. Again, I don't know either business in or out so these are more musings than battle-won opinions.

Solid point

Reply on news.ycombinator.com