How much is Google worth?

by John Q on April 27, 2004

According to this report, the widely-predicted Google IPO is likely to value the equity in Google at more than $20 billion – others suggest $25 billion. I immediately wondered whether Google was really worth $25 billion.

I started on a standard financial analysis. Although, as a private company, Google doesn’t have to publish annual reports, it’s been estimated that Google has annual revenues of $500 million and profits of $125 million so that the return on equity is about 0.5 per cent. We can expect that to grow reasonably fast in the next few years, but the scope for expansion in Google’s core business is far from limitless. Most people in the developed world are already online and most of the heavy users already use Google (Eszter might have more to say on this). Moreover, there’s no strong reason to suppose that Google will be around in, say, 20 years time. I find it hard to draw a plausible earnings path that would yield a present value of $25 billion at any reasonable discount rate.

That’s a problem for the investors, though. The Google example started me thinking about the more general problem of economic valuation in the Internet era. I started by looking at this piece by Simson Garfinkelhat tip – Tyler Cowen. As well as reporting potential competition from Akamai (relevant in considering Google’s longevity), Garfinkel estimates that Google operates a network of 100 000 servers, but that clever design allows the use of very cheap computers as servers. Let’s and suppose an average of $500 a piece. This implies that the main piece of capital equipment operated by Google is worth around $50 million[1] – a hefty sum, but a tiny fraction of the estimated equity value (and presumably there’s some debt in there as well) .

Next, it’s of interest to look at capital-labour ratios. Google apparently has about 1000 employees, which would suggest a total labour cost of the order of $100 million per year – a little on the low side as a proportion of revenues of $500 million, but not implausible. On the other hand, the number of employees is minuscule in relation to the valuation above, which implies a capital stock of $25 million per worker. I feel sure that this kind of ratio would imply some pretty strange organizational policies.

Then there’s the question of how much Google is worth in economic terms. I would think the correct answer must be lot more than the present value of its revenues. I use Google all the time, but unless text ads have a subliminal effect for which Google is being paid, I’ve never contributed a penny to its revenues, and quite possibly never will.

The general problem is that, in an economy dominated by public goods, like that of the Internet, there’s no reason to expect any relationship between economic value and capacity to raise revenue. Things of immense social value (this blog, for example!) are given away because there’s no point doing anything else. On the other hand significant profits can be made by those who can find a suitable choke point, even if they haven’t actually contributed anything of value. Assuming for the moment that SCO prevails in its attempts to extract revenue from Linux users, it won’t be because SCO’s code was better than some free alternative but simply because it was widely distributed before anyone found out it was copyrighted.

If the Internet continues to grow in economic importance, the central role of public goods in its formation will pose big problems for capitalism, though not necessarily to the benefit of traditional forms of socialism.

fn1. Thanks to commentators danny yee and thijs for correcting parametric and arithmetic errors in the original version of the draft, and thereby greatly strengthening my point.



Danny Yee 04.27.04 at 10:22 am

$2500 per server is probably an order of magnitude out – I’d be very surprised if it was more than $500 per server. And data centre space and bandwidth are costs, it’s true, but we’re not talking billions here.


thijs 04.27.04 at 11:14 am

Erm, $2500 times 100,000 = 250 million, not 2.5 billion. That might make a difference.


John Quiggin 04.27.04 at 11:30 am

Thanks Danny and Thijs. Combining Danny’s value estimate and Thijs’ correction of my silly arithmetic error, my point is strengthened by a factor of 50. I’ve changed the post accordingly.


Doug 04.27.04 at 12:32 pm

Didn’t we do all this before in 1998? Or 1999 at the latest? There isn’t an inherent value in a company. Not that there aren’t useful metrics for deciding what you personally may want to pay for a stock, but that doesn’t guarantee that other market participants will settle anywhere close to that personal preference.

Why are some categories of stocks valued at low multiples of earnings? Why can some categories get away with multiples of 20 or 50? You can pay a lot for an MBA to tell you the answers to that sort of thing, but it all boils down to what people are willing to pay.

If investors bid up Google into the stratosphere, none of the cost-of-equipment, or value-of-labor or whatever-else analysis is going to matter worth a dickey bird. If you think Google is going to be overvalued, by all means try to short it, but I would be it’s going to run like a freight train for at least a year and anyone shorting Google in that time is throwing money out the window. As to whether it will eventually come down, I’m agnostic, but good luck timing the market.


des 04.27.04 at 1:04 pm

This analysis relies on the assumption that Google’s “core business” is search engines. It isn’t – it’s the platform they’ve built to run them on. They have systems no one else has, and presumably the corresponding IP, and the interesting question is what _else_ can you run on a rock-solid hyperredundant distributed megacomputer platform like that, and who will pay for it?

If I were Google, which I’m not, I’d be asking intelligence services exactly what kind of junk they’re planning to run their next-generation snoopware on, for starters.


Jeremy Osner 04.27.04 at 1:58 pm

Thanks for the link, Des — fascinating stuff.


Russell L. Carter 04.27.04 at 3:49 pm

“… and the interesting question is what else can you run on a rock-solid hyperredundant distributed megacomputer platform like that…”

The fact that the question needs asking is a clue… Applications able to take advantage of the scalability gains that GoogleFS achieves through relaxed consistency semantics are not that common. I can think of stuff the NSA does which would be highly applicable, and of course email and search, but beyond that, it’s not clear. In other words, you can’t build scalable systems with transactional semantics on google’s technology. BTW, I’m a bit amused that the perceived value that apparently is going to inflate the IPO is not the demonstrated capability, instead its the potential ones. Because Inktomi has been there, done that, and died.


Dan Goodman 04.27.04 at 5:31 pm

I think you also need to look at what else they own, which includes And at projects listed at


Mark Byron 04.27.04 at 5:35 pm

That would give it a 160 P/E when the average stock is around 25. Income would have to increase six fold before it finishes maturing; I don’t see that happening.

Google’s good, but not that good.


Jack 04.27.04 at 5:57 pm

The return on equity that you calculate would more usually be called the earnings yield. The equity referred to in most ROE calculations is the balance sheet equity which is likely to be much smaller than the market capitalisation in this case. Not that this undermines the main point but someone will be doing some cash flow/return on equity calculations to perform the valuation and extrapolating growth out to five years and working Moore’s law into the cost base. Add to that massive operational gearing assumptions (marginal margins being apparently quite spectacular). I imagine if it is suggested that the company is worth $25bn and investors can get in for $10bn they’re sure to be enthusiastic.

My question is why would anyone sell shares in such a rapidly growing apparently self financing company? On the one hand the sellers don’t need to sell unless the price is very good. On the other maybe it isn’t as good as it looks. There were rumours of an impending float last summer.

Will they time things as well as AOL?


Hack 04.27.04 at 7:10 pm

“My question is why would anyone sell shares in such a rapidly growing apparently self financing company? ”

VCs, eager to claim some good returns after a three year slump. Also they are hoping a succesful google IPO will create a rising tide lifting their other boats.


Russell L. Carter 04.27.04 at 7:35 pm

“think you also need to look at what else they own, which includes And at projects listed at”

Yeah, blogs are a perfect match too, as I realized after I posted the first comment. All these are excellent ways of aggregating eyeballs, but I don’t see much of a unique business beyond selling ads. And there is no area on the lab page in which IBM or Microsoft or even Sun has not already achieved world class competency.

Microsoft has already bought every world class expert that can be bought; it’s nice that Google has developed a great stable, but they’ll never be able to compete against Microsoft’s cash cow.


Quaker 04.27.04 at 7:38 pm

Sure, you scoff now.

You’ll be laughing out of the other side of your face when they introduce their perpetual motion engine.


richard bauge 04.27.04 at 8:14 pm

You laugh now, but people like you also laughed when they said the Segway would change the world. Now look who’s laughing.

Okay, bad example.


Keith 04.27.04 at 9:37 pm

Iv’e heard rumors that Google is working on an OS, which would efectively turn the Internet into a single giant super computer with gigaquads of memmory and thousands of chained processors, that anyone could access, search and use for free. How that would effect their ecomic outlook, ic an’t say, not having much of a head for numbers but i’d guess it’d be pretty valuable.


Ian Whitchurch 04.27.04 at 11:44 pm

OK, disclaimer : I work in the stock market, as a promoter, but I deal with oil stocks, not techs.

What is missed is that Google have the Holy Grail of advertising – knowing exactly what mindshare a product has, as compared to what it had last week.

Lord Lever once said ‘Half the money I spend on advertising is wasted. My only problem is which half ?’

Google can answer that question for you. Say you are Sony, and you want to know which of your stable of pop stars are gaining and losing popularity in which market.

Google, by keeping track of how many users are asking about them and where, can tell you

But it will cost.

Google knows how the world thinks. And thats worth – literally – billions and billions.

Ian Whitchurch


Harry Clarke 04.28.04 at 1:20 pm

This is a nice puzzle. Goggle is probably worth more than $25 billion to society — we use it all the time. But the privately appropriable value is perhaps $2.5 billion not $25 billion. Goggle is drenched in public good value but short on private value. So why do the Wall Street hardnoses value it so high?

You can say investors are stupid or, as Doug says, value is independent of present value calculations but this seems to me a cop out. You can explain any apparent weird valuation this way.

There is no stock market boom so a ‘greater fools’ theory doesn’t work. What’s going on? Why should the market overvalue it on the basis of any plausible present value calculation? Are the Wall Street hardnoses wrongly internalising social values that sound economic theory says they should ignore and if so why?

Good questions (I think) but, sorry, no answers.


Doug 04.29.04 at 8:59 am

I like Ian’s thinking on this, and the idea about distribute computing is also very interesting. Both going to show that John’s valuation is more a starting point than a finishing point.

And if Ian can make a living promoting stocks, then the ‘Wall St hardnoses’ are not as hardnosed as Harry seems to imply. Surely there will be a stampede for Google. The questions for me are how far it will overshoot, how management will react to a market cap that’s probably way above a sustainable level, and when the price will settle down. Answers to those (and some starting capital) would be a good way to make some nice cash.


Seth Finkelstein 04.29.04 at 2:14 pm

Google insiders reportedly are quite concerned with the lessons of Netscape, one of which is Beware Microsoft.

Part of the reason for the IPO now may be that Google is arguably at peak value before Microsoft can mount an attack.


Harry Clarke 04.29.04 at 2:28 pm

Ian’s point is fine if Google can hang onto an exclusive access to this info. But entry seems more than possible — it seems likely.

Seth’s explanation doesn’t seem right on the same grounds. If investors anticipate entry they will downgrade the stock. It is not true that the only group who fear entry are insiders? I think to the contrary everyone assumes some type of entry will occur.

Its not just what Goggle can do but what Goggle can do that others (in the long-run) can’t.


Smoof 04.29.04 at 11:16 pm

Unfortunately, Google has gone and made this whole discussion moot: Google sets $2.7 billion IPO


Harry Clarke 04.30.04 at 1:26 am

Smoof, I don’t think the article you cite makes the discussion moot at all. It says Google seeks to raise $2.7 billion not that this is the value Google is assigning itself.

The last three paras make the point clear — the number of shares to be sold is not set so you cannot guess what the market value will be. The article suggests figures between $16-$24 billion.

The puzzle remains.


Bill Mann 04.30.04 at 3:10 pm

File me under Google-skeptic as well. I don’t think people have even delved into the Dual Class structure of the company yet.

Unfortunately your analysis is a little off. For return on equity you don’t use the stock market value, you use shareholder equity, an account on the balance sheet. SE is the combination of retained earnings, par stock, foreign currency translations, etc.

In the filing they list shareholder equity of $727 million, giving it an annualize ROE exceeding 33%. This amount will drop dramatically by virtue of all the unneeded cash the company is going to generate from the IPO.

Comments on this entry are closed.