Home » Internet » Huge Net Capex: One-Time or Permanent?

Huge Net Capex: One-Time or Permanent?

The WSJ describes a key issue facing the big net companies, especially Google: massive (and increasing) capital spending.  Last week, for example, Google stated that 2006 CAPEX would “significantly exceed” 2005’s mind-boggling $800 million, prompting analysts to toss out 2006 estimates in the $1.25 to $1.5 billion range.  Yahoo! spent $400 million last year, up from $250 million the year before.  Amazon’s cranking up spending…

This gives rise to three key questions:

First, what on earth are they (or at least Google) spending that much money on?  I understand the need for “servers,” but why has this need exploded so much in the last two years?  GDrive? Gmail?  Neither of these ideas is revolutionary. Video is a CAPEX hog, but is the difference that vast?  Also, why does Google need $400 million more than Yahoo! and eBay do?  I shudder to recall this, but CAPEX is where companies occasionally stuff expenses they don’t want to flush through their income statements right away (e.g., Worldcom and Homestore).  I have huge respect and confidence for Yahoo!’s financial team, and there is no conceivable reason for Google’s team to play such games.  Without having more detail about exactly what the money is being spent on, however, the possibility of shifting capitalization policies can’t be ruled out.

Second, is this simply an up-front investment necessary to build out global capabilities, in which case CAPEX will moderate in future years, or has run a global Internet operation become much more expensive?  If the answer is the former, today’s CAPEX is largely irrelevant: The companies can afford it, and a future easing back on spending will drive free cash flow growth rates through the roof.  If the spending is permanent, however, these companies (again, Google especially) aren’t going to generate anything like the free cash flow the market is currently expecting.

Third, returning to a point raised in question one, why is Google spending more than twice as much as Yahoo! and more than Yahoo! and eBay combined?  Again, explanations about “storage” and “computational power” just don’t ring true.  eBay and Yahoo! need storage and computational power, too, and, right now, Yahoo! handles more traffic than Google does.  The difference between the CAPEX levels, moreover, is so vast that it seems safe to assume that either Google is overspending (in which case, the question is “on what?”) or Yahoo! and eBay are underspending (in which case, free cash flow is about to take a hit at both of those companies as well).

When Google releases its 10K, we’ll get more insight into where that $800 million went.  But this won’t provide any into where this year’s estimated $1.25 billion is going–or how much CAPEX as a percent of revenue will be required to run these companies for the next 10 years.

15 thoughts on “Huge Net Capex: One-Time or Permanent?”

  1. Henry,

    Would you be able to compare Capex across other companies/industries? Say, mining/gaming/semiconductors, etc. Is google building a spaceship/space elevator? It seems highly concerning that the street and investors know very little about the company, meanwhile expenses and other key financial data are hidden deep with the Google Black Box!


  2. G’s CAPEX will only go higher now. The search geeks are now entering the WiFi industry. Although the investments they have in mind are relatively small ($10-20M per market), reality will show it will cost at least double that amount. They are partnering with EL, but I am not sure EL can commit large funds to these roll outs. Anyway, duplicate their plan in 20 markets, and you will arrive at another $200-400M (if not $400-800M). Of course, the plan may not call for building/ rolling out in all markets in the same year, but this is to give you an idea what Google has already commited to.

  3. Andrew,
    Have you seen the ultimate geek movie, Contact? I have it on high authority they are building the transport in that movie so that Schmidt can visit his father and heal his emotional wounds. If he receives forgiveness, they will no longer make such outrageous claims such as $100 billion in revenue. You see, the only reason those claims were made is because he’s seeking validation from his father.

  4. Wait, I caught Contact last night! does that mean I am GEEK!?

    How much will they end up spending to earn $1 of revenue.

    Economies of scale? ANYONE?

    These guys need to read “Increasing Returns”. All this cash outlay and for what? A product that does not display nor have a customer “lock-in” nor does it disply any of Increasing Retunr principles. Search is easily subsitutable and the technology is as well.

    I just do not see the benefit to the user/stakeholder/shareholder that would justify such cash out flows.


  5. From the WJS article on the GOOG analyst day’s notes (leaked):

    According to the notes, Google believes storing users’ data centrally will make the computing device — such as a PC or phone — less important. That could undercut the advantage that Microsoft has thanks to its dominant Windows operating system, browser software and Office productivity applications. It “suits our strength vis-à-vis Microsoft,” the Google notes say.

    Right, I get it. If they just build a giant free hard drive in the sky, they’ll kill Microsoft because people won’t need Windows and Office anymore.

    SUNW had this dream back in about 1995 coincedentally. The dream was insane back then, and its insane now. It shows a very, very deep lack of understanding of how the world works.

    This would be an idle curiosity except that it appears that GOOG is doing two dangerous things here:

    1. Spending hugely to make this dream work (the costs to even start the “free harddrive in the sky” idea are astronomical).

    2. Once again, removing focus from the business that got them there. In this case, with such an enormous expenditure, the management team would have no choice but to very carefully focus on this project to the detriment of many others.


  6. Google is reaching into to many areas too quickly and in markets that have battle tested, entrenched warriors such as MSFT & EBay. Stealing lunch from Yahoo! Who took their eye of the ball, is not nearly the same as going head to head with super heavy weights. As Ross Perot said to Bill Clinton during a three-way debate in 1992: “Just because you can run the corner grocery store, doesn’t mean you can run the Wal-Mart down the block!”

  7. How about them data containers that were rumored in the google labs basement ??

    Each node could stuffed and placed on a truck or dumped in a warehouse with zero or very little cost for maintance. Once moblized will cannamlize the online storage and infrastuctre issues. This permits a data on demand service via any method, desktop or mobile device.

    Secondly why is google so interested in MDS ?that’s because the power of the net will be harvested mostly via handheld devices. But for this to work, they still dark fibre optics meshed across an atm cloud to make it happen.

    ROI vs capex will become a mute point as toc will be further driven down and making yahoo and msft and others to play the same game- which as the figures speak for themselves they are not doing – correct ?

    I think that a secondary layer of google networks and online storage is the next wave emergence and that’s why they have dear old vint cerf too !!

    Just think about it- a landscape being formed which bumps off yahoo / msft and other players !!

  8. In case anyone missed it, I’d imagine a possible online data storage space (if this is in fact on the horizon) could be one substantial source of increased capex spending at Google:


    As for the comments above describing an online storage space as “insane”, I have to say that I believe it is the author of that comment that lacks an understanding of where tech is heading. It is not a question of IF online data storage and resultant mobile accessibility will happen, but rather WHEN will it happen. The answer to WHEN is intimately contingent upon a few fundamental prerequisites such as development of sophisticated online data security, protection, and authentication systems. The other factor to consider is whether online data will be housed in a centralized location (ex. Google’s servers) or in a decentralized (ex. bit torrent type) online space. I am very much inclined to say that the decentralized option offers much more powerful possibilities – but this is for another discussion.

    When data can be confidently stored and protected in the online world, I would venture to say that we will see rapid movement in that direction allowing for all of the mobility and access benefits associated with the online world.

  9. GOOG is cash flush. They must have a build it, rather than buy it mentality. I know they admire buffett, but I am not so sure they are thinking of investments in relative terms. Eitherway it should be fun!

  10. As a shareholder of Google, I personally would like for them to increase CAPEX exponentially in the future. They have the intellectual capital, and if I ran the company, I’d throw unlimited money at their brilliant engineers. I could care less about google’s advertising business as I strongly believe their “20% time” will produce the innovation’s that will make this company the most valuable tech company on the planet by 2010.

  11. Ben:

    The concept of the Hard Drive In The Sky has been around for about two decades now, and so far there has been zero demand for it.

    Every year technology makes the problem easier and easier to solve (faster networks, cheaper hard drives, SANs, etc. etc.), and every year there’s still no demand for it. “Sophisticated online data security, protection, and authentication systems” have been around since the 80s. Every year we see better and better solutions. Easier. Cheaper. Brain-dead simple to use. Zero demand.

    The problem is, the above technology factors make the current paradigm of the Hard Drive on Your Desktop even cheaper and better still. P2P networks are only good for illegal content (a centralized/cached server setup is more efficient, and please do the research and the math before parrotting the latest kewlshit you read in Wired).

    My PC hard drive works fine. Is there something wrong with yours? Anybody elses?

    GOOG is putting their money on a solution in search of a problem. Building the solution with today’s great technology is like winning the latest video game with the cheat codes on–and just as pointless.


  12. Many companies are one-hit wonders. The problem is that it may take decades of CAPEX spending before those insiders flush with success and desperately seeking a second act may realize it . . .

  13. Slide8 “We are spending 500m+ in CapEx and innovating in products like automatic machine translation in order to make this happen.”

    Henry –So the Capex is NOT double what the others are spending :)-

  14. Hilarious. No offence, but it is truly priceless watching non-techies discuss subjects about which they know next to nothing.

    Lets examine the main free search product. Currently Google run about 100k servers distributed over their many datacentres. Obviously, this number will grow over time as new DCs’ are added, and existing ones beefed up, but lets take 100k as a nice easy number.

    As they are partway through a major “infrastructure rollout” codenamed Bigdaddy, which is expected to hit every DC, that’s 100k new boxes. Even assuming they kept the unit cost down to their historic figure of $1000, that’s $100m right there. That figure has 2 low assumptions in it too.

    And that’s just 1 service, albeit a fair sized one. There’s also Image Search (a separate service, but fairly storage heavy), Gmail, with its commitment to keep increasing storage capacity in perpetuity (over 2.7Gb and rising), Froogle, Google Groups (or Usenet to some of us), News, Local, Maps, Base, and the biggest daddy of them all : Dark Fibre

    They’ve been buying it up for years, and they don’t peer with anyone. Ever (Ok, not quite true, but a definite anomaly – you just don’t do that). They took over a quarter million sq ft in NY, and I wouldn’t be surprised to see another 2 – 3 million taken around the world, in addition to the “DCs in a container” they have been building up.

    They bid for the contract to supply free WiFi to the whole city of San Francisco – OK, they’ve partnered with Earthlink, but hey, those guys have actually done it (Anaheim and Philadelphia, I think). And successfully! Pretty much a partnership made in heaven, and if the SF rollout is successful, just expect it to go global, quickly.

    That’s estimated to cost $18m to cover 49 measly sq mi. From what I could find, the US had 92,500 sq mi of urban land in 2000 (!), which works out just shy of $34b to do the lot.

    Still wonder where the cash is going?

Leave a Comment