Google, the OS company

The bottom line: Google is now an OS company.

The fact that Google's recently-announced OS products are aimed at mobile devices and social networking sites is interesting, and I'll talk about the impact of that below. But it's secondary. I think the big, really important change is that Google has now jumped with both feet into the middle of the operating system world. That potentially has huge implications for the industry.

The impact will depend a lot on how Google follows up. If it pours substantial energy and resources into its OS offerings, it will be extremely bad news for Microsoft and other companies trying to charge money for their own platforms. On the other hand, if Google doesn't make a serious long-term commitment, it will embarrass itself deeply. This isn't like launching a new web application -- an OS has to be complete, and it has to work properly in version 1, or there won't be a version 2.


What they announced

It's kind of ironic. For years after Google became a prominent web company, people speculated about whether or when it would create its own OS. The logic was that Microsoft has its own OS, and Google was challenging Microsoft, so Google would create its own OS too. But then as the years went by and it didn't happen, people moved on to other subjects. The speculation died out. But one of my rules about the tech industry is that "obvious" things happen only after everyone in the industry has written them off. So I guess Google was due.

The company has been creeping toward the OS space for a while. Google Gadgets is an API to create small applications that run in web pages, and Google Gears is code that lets web apps run offline, making it easier for them to challenge desktop applications. But they were both relatively low-profile (or as low profile as anything Google ever does). But in the last couple of weeks, Google made two much more assertive announcements:

--OpenSocial is an effort to create a shared platform for applications that can be embedded within social websites (link).

--The Open Handset Alliance is an effort to create a shared platform powering mobile devices (link).

Although they're aimed at very different parts of the industry, they're both efforts to create a standard platform where there was fragmentation; and they're both alliances of numerous companies, with Google providing most of the code and the marketing glue. I think there's a recurring theme here.


Details on the Open Handset Alliance

Open Social was covered very heavily when it was announced a couple of weeks ago, so I won't recap it all here. If you want more details, Marc Andreessen did an enthusiastic commentary about it on his weblog (link).

The OHA announcement was today, and I want to call out some highlights:

--It's built around a Linux implementation called Android. Android will be free of charge and open source, licensed under terms that allow companies to use it in products without contributing back any of their own code to the public. This will probably annoy a lot of open source fans, but it's important for adoption of the OS, as many companies thinking about working with Linux worry that they will accidentally obligate themselves to give away their own source code.

--Google is creating a suite of applications that will be bundled with Android, but they can be replaced freely by companies that want to bundle other apps, according to Michael Gartenberg (link). There is a lot of speculation, though, that if you bundle the Google apps you'll get a subsidy from Google. The folks over at Skydeck estimate the subsidy could be about $50 per device (link). That might not sound like huge money to you and me, but keep in mind that mobile phone companies routinely turn backflips to squeeze 25 cents out of the cost of a phone. When you sell millions of phones a year, it adds up.

--A huge list of companies participated in the announcement. That's not as impressive as it sounds; when you have a well-known brand, a lot of companies will do a joint press release with you just for the publicity value. But a few stood out:

Hardware vendors. Samsung, Motorola, LG, and HTC all endorsed the OS. HTC and LG gave particularly enthusiastic quotes. The first three companies have all been playing with Linux for some time, so I wasn't surprised. But HTC is another matter -- it is the most innovative Windows Mobile licensee, and Microsoft must be very disturbed to see it blowing kisses at Google.

(A side comment on Motorola: For a company that said it wanted to consolidate down on a small number of platforms, Motorola is behaving strangely -- it jumped all over Symbian a couple of weeks ago, and now is supporting Android as well. I think it has now endorsed more mobile operating systems than any other handset vendor.)

Operators. Participants in the announcement included NTT DoCoMo (a long-time Linux lover), KDDI, China Mobile, T-Mobile, Telecom Italia, Telefonica, and Sprint. That's a very nice geographic spread, and ensures enough operator interest to make the handset vendors invest.

--Google claims all Android applications will have the same level of access to data on the phone. That's pretty interesting -- most smartphone platforms have been moving toward a multiple-level approach in which you need more rigorous security certification in order to access some features of the phone. I'll be interested to see how the security model on Android works.

--We'll get technical information on the OS November 12, and the first phones based on Android should ship in the second half of 2008.

--Although Android's first focus is mobile phones, the New York Times reports that it can be used in other consumer devices as well (link).


What it means to the mobile industry

It all depends on the quality of Google's work and the depth of its commitment. If Android has technical or performance problems, it could sink like a stone. If it doesn't have enough drivers or has poor technical support, the handset vendors will avoid it. If the developers can't create good applications, users won't want it. This is a very different business for Google -- handset vendors and operators will not tolerate the sloppy, indifferent technical support that Google provides for its consumer web apps.

If, on the other hand, Google's platform really works and the company invests in it, I think it could have some very important impacts.

Impact on Windows Mobile: Ugliness. The handset companies endorsing Android are also Microsoft's most prominent mobile licensees. I doubt any of them are planning to completely abandon Microsoft (they don't want to be captive to any single OS vendor), but any effort they put into Android is effort that doesn't go into Windows Mobile. So this is ominous.

The whole mobile thing just hasn't worked out the way Microsoft planned. First it couldn't get the big handset brands to license its software, so it focused on signing phone clone vendors in Asia, thinking it could use them to pull down the big guys. But Nokia and the other big brands used their volume and manufacturing skill to beat the daylights out of the small cloners.

Now Google is coming after the market with an OS that's completely free, and may even be subsidized. This will put huge financial pressure on not just Windows Mobile, but all of Windows CE. Even if Microsoft can hold share, its prospects of ever making good money in the sub-PC space look increasingly remote.

Impact on Access: Ugly ugliness. How do you sell your own version of Linux when the world's biggest Internet company is giving one away? I don't know.

Impact on Symbian: Hard to judge. Symbian is the preferred OS of Nokia. As long as Nokia continues to use Symbian, it stays in business. The question is how much it'll grow. After years of painful effort, Symbian just managed to get increased endorsements from Motorola and Samsung. Now Google is messing with both of them. Japan has been a very important growth market for Symbian, now Android is endorsed by both DoCoMo and KDDI. All of that must feel very uncomfortable. If nothing else, it's likely to produce pressure on Symbian to lower its prices. And Symbian should be asking what happens if Android turns out to be everything Google promises -- a free OS that lets handset vendors create great phones easily. It's not fun competing against a free product that's been subsidized by one of the richest companies in the world (just ask Netscape).

Maybe if Symbian agrees to enable Google services on its platform it can get the same subsidies as Android does. It's worth asking. If not, maybe Symbian should be looking for other places where it can add value in the mobile ecosystem.

Impact on mobile developers: Potentially great. Mobile developers have suffered terribly from two things: They have to work through operators to get their applications to market, and they have to rewrite their applications dozens of times for different phones. If Android produces a single consistent Java environment for mobile applications, that would be a big win. And if it can open up the distribution channels for mobile apps, that would be great as well. We don't have enough details to judge either outcome yet, and the app distribution one depends on business arrangements that may be outside Google's control.

Impact on Apple, RIM, and Palm: Probably none at all. A lot of the coverage of Android is positioning it as some sort of challenger to iPhone and RIM.

I don't buy it.

Apple, RIM, and Palm all make integrated systems in which the software and hardware are coordinated together to solve a user problem. Android, by contrast, is only an operating system. It's plumbing, not the whole house. Unless Google's handset licensees magically develop the ability to design for users -- a feat equivalent to a giraffe sprouting wings -- their products won't be any better as systems solutions than they are today. The OS hasn't been the thing holding them back, and changing OS won't alter the situation.

Android puts interesting financial pressure on Microsoft, but it doesn't directly solve any compelling user problems. If it eventually drives a great base of mobile applications, that might eventually be attractive to some users. But in that case the systems vendors could just add a copy of Google's application runtime (it's open source, they can grab it anytime they want). Or they could host their devices on Google's plumbing. Palm and RIM might both benefit if they could transfer engineers away from core OS and toward adding value that's visible to users.


Impact on the tech industry: This isn't just about mobile phones

I have no access to Google's internal thinking, but even if it sincerely believes it's only doing a mobile phone OS, I don't think it can or will stop there. Technology products often develop a momentum of their own, no matter what was intended at the start. The lines between the computing and mobile worlds are breaking down already, and if Google creates an attractive software platform that's free of charge, that platform will inevitably get sucked into other types of devices. I'm not saying that Android is going to end up in PCs, but if it's functional and well supported I think it could end up running on just about everything else that has a screen.

Besides, if you look across all of the recent Google announcements, I think it's clear that Google has a larger agenda: It wants to break down walled gardens, because they interfere with Google's ability to deliver its services. It has even developed a standard methodology for attacking them: Create a consortium so you don't look like a bully, and fund an "open" alternative to whatever is in the way. They are doing it to Facebook, and they're doing it to Windows Mobile. Google doesn't even have to make money from the consortium, as long as it clears the ground for its services to grow.

Take a lesson from evolutionary history. The most successful animals are not those that adapt to the environment; they are the ones that reshape the environment to match their needs. I think that's what Google is doing. It's going to use open source and alliances to suck the profitability out of anybody who creates a proprietary island that it can't target.

It'll be interesting to see if and how Google applies this principle to the upcoming frequency auction in the US.

Or to anyone else who gets in its way.

18 comments:

raddedas said...

"Impact on mobile developers: Potentially great... If Android produces a single consistent Java environment for mobile applications, that would be a big win."

ROFL! With two JVM vendors on the initial roster this seems unlikely, not to mention the minimal chances of the other hardware specs standing still enough to eliminate screen size issues, data entry variations, etc. You'd have to assume that every other platform in the industry died as well, not likely for some time even if it is phenomenally successful. Nice thought though.


I was also intruiged by this paragraph in your conclusion: "It wants to break down walled gardens... It has even developed a standard methodology for attacking them... They are doing it to Facebook, and they're doing it to Windows Mobile." I think you're right but I'm intruiged as to why you think Windows Mobile, with negligible market share (especially outside the US enterprise market) is so critical to the existence of walled gardens. The gardens appear to belong to the Operators right now and it's surely they who control whether those walls stay up - inflicting collateral damage on Microsoft in a non-core market would doubtless please Google but it can't be their main purpose...


In general though, nice overview as always :)

"It's going to use open source and alliances to suck the profitability out of anybody who creates a proprietary island that it can't target." - spot on.

Anonymous said...

I just want to add one thing, forget about J2ME for a second, speculation on my part but imagine if they enabled developers to create applications in web scripting languages (like Nokia's S60 will enable with Widget support) and then use Google Gears to let those applications run offline? Then give adsense as a way to pay for those applications and you're golden!

Anonymous said...

I think the biggest issue is whether they'll get significant manufacturer support. If there's no phones it wont go anywhere. Moto, Samsung and LG all have multi platform strategies and I can't see them throwing that away.

That said HTC could have an impact if they're looking to open up a new business area, but on the other hand we're probably talking about 10-20 million devices at most by 2009. I don't think that is enough too have a serious impact.

The point about subsidised phones is interesting, but then its not just Google that could do this - any service provider could...

Bojan said...

Yeah, Stefan Constantinescu I see what are you saying... but mobile phones are online all the time!!!

Anonymous said...

Good write-up Mike. A few thoughts in no particular order...

- On Motorola's supposed desire to focus on a limited number of platforms: Much of the Tapwave development team was picked up by Motorola shortly after they folded, though I was told they weren't working on anything mobile-related.

- Unless I'm getting my facts wrong, don't forget that Android is the brainchild of the brainchild of the Danger hiptop. That was a great total solution (hardware and OS/GUI), hindered mainly, IMO, by its closed development and reliance on T-Mobile as the only US network. I'm anxious to see what comes out of the SDK next week. I'm expecting that this isn't just "plumbing" but GUI and user experience guidelines as well. That said, I'm not sure how all of that will jive with their comments about the platform working on a wide variety of hardware configurations (screen resolution differences, with/without thumbboard, etc.). As I've said many times before, I think the best platforms are the ones where the hardware and UI are designed in sync. This is what contributes to the excellent user experiences of the Sidekick, the Treo, and the iPhone, and part of what makes the Java and WM user experiences poor.

- I have to wonder if the timing of the announcement isn't meant to deflate the iPhone balloon a bit. That's the new king of the hill right now. If the timing estimates are right, we could see the Google OS covered on CDMA and GSM carriers by the end of next year with developers having had several months to get familiar with it, whereas the iPhone will still be stuck on the one GSM carrier.

- What a poor reflection of Access this is. They're shooting for similar timeframes, but Google will have a look at the SDK next week. Meanwhile, the last we've heard about Access' plans, their OS/GUI still seemed like a rather cobbled together approach.

- I think you underestimate the impact to Palm. They'll be playing the Apple game with their own proprietary OS running on their own proprietary hardware. They've got the edge in terms of carrier support, but can they compete with Apple's marketing muscle and will their Linux-based OS and GUI measure up? If the Foleo was any indication, I'm not holding out much hope. For their sake, I hope their development is farther along than it seems, their GUI is better than I expect it to be, and their SDK comes out sooner rather than later.

- We shouldn't expect Apple to stand still. If nothing else, I'm hoping that this nudges them further in the direction of opening up the platform for development among hobbyists. Their upcoming signed approach smacks too much of a pay-to-play approach with costly up-front signing/testing fees and a need to sell apps through the iTunes store. I applaud Google's desire for Android apps to be installed via wireless, flash card, or USB. That's the pro-user/developer approach I want to back.

Anonymous said...

Oops. Meant to say that Android is the "brainchild of the father of the Danger hiptop."

BANUDK said...

intresting article as always.

But I think Google playing OS game is totally different then they launching web applications.
key thing in getting device manf support .unlike PC world most cost for building a device goes to integrating HW With SW .As HW spec of devices frequently changes keeping low level SW in tune with it is most pain and costly excercise for device manf . Google OS will not solve that problem .
And also funding a full feldged OS development and keeping it compettaive is expensive business and giving it free and hope to make money on ad platform integartion is far fetched and risky stratergy .
If i were investing I wouldnt be happy after this announcment as it significantly increase fixed R&D cost with no return in near term future .

gibtang said...

If Android produces a single consistent Java environment for mobile applications, that would be a big win."

I wouldn't be too optimistic about that. 1 big factor in device fragmentation is screen size followed by different JVM implementations. Unless all the phone manufacturers produce phones with a standardized screen size, say 240 x 320. Device fragmentation will still be an issue.

Also, there is still nothing to prevent Nokia, LG, Motorola etc to put their own implementation of the JVM on the phones and use it. With their own implementation, there will not be a consistent J2ME environment although that is what I wish for.

niko said...

"How do you sell your own version of Linux when the world's biggest Internet company is giving one away?"

I do think you answered yourself, although not clearly, in your write-up. Google aims at being a concurrent of Apple, Palm, etc. Both companies are selling (or willing to) their own Linux. So how to sell Linux? Have it integrated into some *shiny* hardware.

Google, instead of going down the software/hardware couple route, it goes the Windows CE route; only one software that would go into plenty of different hardware... but it does it for free.

Good idea or bad idea???

Douglass Turner said...

Mike, I just don't see this OHA thing going anywhere. I think Google's partners have all the commitment of a Facebook friend.

Let's cut to the chase, are these OHA operators actually going to open their networks to developers in a manner similar to the PC-Internet. I see nothing in this announcement that indicates that. Unless that happens this is a non-event.

Am I missing something here?

Cheers,
Doug

Michael Mace said...

Thanks for the cool comments, everyone.

One update before I respond to the comments: Richard Windsor of Nomura just published his take on the announcement. He believes Android is aimed at mainstream phones rather than at smartphones, because the midrange phones are the ones where Google can't deploy any of its services today. So he sees the main impact being on the embedded phone operating systems (Nokia Series 40 and friends) rather than Windows Mobile and company.

I don't think Google is that finely focused, but it's a really interesting perspective.

He sees Nokia's Ovi services as a big loser in this announcement because Google's phones apps will compete with Ovi. I'm not sure I agree with that one either, but it's worth passing along. Windsor is a sharp guy and an original thinker.


raddedas said...

>>With two JVM vendors on the initial roster this seems unlikely

But Google has been quoted in several places saying that one of their main goals is to produce a single consistent platform. I agree with you about all the barriers to all this, but they're not stupid. At least I think they're not stupid. Anyway, I really want to see that SDK they're promising for next week. That'll tell us a lot.

Speaking of which, I'd love to know why they did the announcement this week and then held back the SDK for another week. Trying to milk publicity out of the announcement, maybe? But Google doesn't need to do that. Weird.


>>I'm intruiged as to why you think Windows Mobile, with negligible market share (especially outside the US enterprise market) is so critical to the existence of walled gardens.

I think the target isn't Windows Mobile, it's Windows. In my opinion, one major intent of this announcement is building a big high wall around Windows, constraining it so it can't spread into adjacent markets. I didn't make that clear in my post.

But yeah, I agree that the operators' gardens are also a target. But Google will need to generate user demand for its phones in order to use Android to pull down the operators -- and that needs (in my opinion) compelling devices based on the Google OS. Won't happen from the usual phone hardware crowd; if they knew how to build compelling mobile data devices they would have shipped them already.

Now, if Google buys a bunch of spectrum in the US, and then deploys a lot of Google phones on that network, that would put some very interesting pressure on the operators.


Stefan said...

>> imagine if they enabled developers to create applications in web scripting languages (like Nokia's S60 will enable with Widget support) and then use Google Gears to let those applications run offline?

Nice. Sounds like bad news for the mobile ambitions of Silverlight and Air. I really, really want to see that SDK.

Adobe is moving too slowly on Air, but that's a subject for a different post.


Scott R said...

>>I'm anxious to see what comes out of the SDK next week. I'm expecting that this isn't just "plumbing" but GUI and user experience guidelines as well.

Good point.


>>If the timing estimates are right, we could see the Google OS covered on CDMA and GSM carriers by the end of next year with developers having had several months to get familiar with it, whereas the iPhone will still be stuck on the one GSM carrier.

Good point, although I bet the one (or two) Apple phones by that point will be more compelling than half a dozen devices from LG and Samsung. It's going to be a very interesting fight.


>>I think you underestimate the impact to Palm.

I agree that Palm has plenty of challenges, I just don't think Google's actions make them much worse. Palm's problem is not that its strategy is necessarily wrong, it's that it hasn't been executing well on that strategy.


gibtang said...

>>Unless all the phone manufacturers produce phones with a standardized screen size, say 240 x 320. Device fragmentation will still be an issue.

I don't agree. If you write the API to take into account multiple screen sizes from the start, it's not a big issue. Mac OS did that in the 1980s.


>>there is still nothing to prevent Nokia, LG, Motorola etc to put their own implementation of the JVM on the phones

Yup. All the more reason I want to see that SDK. What is the default API set of this OS?


Douglass said...

>> I think Google's partners have all the commitment of a Facebook friend.

Nice line!

I don't know how Google expects all of this to play out.

If they want to break the operators' power, at some point they need something that's so compelling to users that the operators can't avoid offering it.

But that's assuming they expect this announcement to be the one that breaks the operators. Maybe this one is aimed at the OS vendors, and the frequency auction will be the tool to deal with the operators.

The one thing I'm sure of is that we haven't seen the full picture yet.

Anonymous said...

Michael Mace said:

'Now, if Google buys a bunch of spectrum in the US, and then deploys a lot of Google phones on that network, that would put some very interesting pressure on the operators.'

I think the US may be the least important market for a company like google to own there own specturm. That is because of the internetwork roaming already commonplace and the fact that everything works on an airtime model. They may find this a market that they could effectively operate as an MVNO similar to how virgin mobile and other operate.

This could be much a much more difficult acheivement in europe. I am thinking largely on the assumption that google will be looking to intergrate VOIP(as in Gtalk voice chatting) into there platform. The european operators survive on a per minute termination charge model. This does not translate easily if voice is not being dialed in through the traditional networks; it could be worked right into bundled airtime minutes in the US on the same basis as PSTN calling. Also the sort of domestic roaming(with companies like US cellular, alltel, verizon all promoting the same combined coverage map) common in the US is not common in europe; therefore european operators have much higher wholesale roaming that could make use of these networks much more expensive for a potential google MVNO than in the US.

Michael Molin said...

Hello Michael,

A platform is a real thing i.e. hardware by definition. Programming is using hardware functions of a processor or a family of processors. What is Microsoft Windows 3.0 as a 'PC platform for everyone' without Intel's 80286. And the Internet that has been developed by users on a x86 architecture.

The same is for Google - Intel is offering the Menlow platform with a Silverthorne processor now. That's all this is about - a x86 architecture for mobile devices. Plus WiMAX supported by Menlow platform now and as 'the only' connectivity for Moorestown SoC platform.

Michael Molin

-
GeneTechnics - http://geocities.com/gene_technics/

Michael Mace said...

Thanks, Michael. Good points. The WiMax connection is especially intriguing.

This is one of the reasons why I want to see the Android SKU. How closely will the apps be tied to the hardware? For that matter, how closely will they be tied to Linux?

One way for Google to provide app security would be to limit the applications to Java only, in a totally sandboxed environment. That could also enable Google to port the SKU to other mobile operating systems.

We should find out this week...

Michael Molin said...
This comment has been removed by the author.
Michael Molin said...

Actually, this defines the future. The concept opposite to Java. As we know, Apple is already at Intel's platform. Anyone else? Cray, if someone remember about the supercomputers. And IBM is a research institute for the industry. Telecom is a part of the society, so there is no need in Java when the mobile devices platform has been matched with the desktop platform.

Michael Molin said...

What a strange interface Blogger has. I used to register on a site and then write a comment but not upside down. Otherwise they should be 'as it is' initially.

Anonymous said...

I've just seen the Steve Ballmer (Microsoft CEO) and Nigel Clifford (Symbian CEO) speeches about Android (http://www.weshow.com/us/p/22898/microsofts_ceo_discusses_googles_smartphone) and they really don't seem to be worried about Android, as we can see in their smiley and calm faces.