Google Acquires Chip Maker Startup Agnilux

bobwrit points out a story at PC Magazine, from which he extracts "Google has purchased Agnilux, a secretive chip house made up of engineers who architected the heart of the iPad, then left the company. Reuters' PEHub reported the story Tuesday night. A Google spokesman also confirmed the acquisition to PCMag.com. 'We're pleased to welcome the Agnilux team to Google, but we don't have any additional information to share right now,' a Google spokesman said Tuesday night via email."

Comments

Lawsuit in the oven

[alfredos]

I can feel a lawsuit coming...

Re:Lawsuit in the oven

[TubeSteak]

I can feel a lawsuit coming...

Why?
If those guys didn't get sued after leaving Apple and starting their own company, why do you think Apple would have any legal grounds to come after them now?

Re:Lawsuit in the oven

[nurb432]

Why is it any different then any other day? Lawsuits are just the name of the game now.. no matter what a company does *someone* is going to sue..

Re:Lawsuit in the oven

[naplam33]

I'd guess at least some of their contracts at apple included at least one or two pretty strict clauses they can try to enforce as soon as they can prove what they're working on at Google.

Re:Lawsuit in the oven

[TheRaven64]

Because they already started their own company, called PA Semi, and Apple bought it specifically to get their hands on the development team? I'd have expected Apple to insist that the employees had signed contracts that would prevent this kind of thing. I'd also expect Google to be a bit wary of buying a company founded by the same people who just quit the company that had just bought the last company they'd founded...

Re:Lawsuit in the oven

[alfredos]

Why?

Apple probably doesn't care if they go on their own way and have a great success, say, making chips for controlling A/C units. But going to Google, whom they see as a big ship in collision course with them, can't end up in a happy "we're all friends" ending, can it?

Re:Lawsuit in the oven

[LordNimon]

I'd have expected Apple to insist that the employees had signed contracts that would prevent this kind of thing.

There's no way Apple could do that. PA Semi employees didn't have any such contract (I presume), and when Apple bought them, there's no way they could have forced the employees to sign such a contract. I've never heard of a company being purchased on the condition that the employees sign away their rights before the purchase.

Re:Lawsuit in the oven

[Clandestine_Blaze]

I thought such non-compete clauses were void in the state of California? I don't see how they could get sued.

Re:Lawsuit in the oven

[maxume]

It's quite likely that the price was a trifle (In Google's eyes).

And Google probably roped them into being employees at least for a few months.

And maybe they just wanted some particular technology.

Re:Lawsuit in the oven

[ezzzD55J]

because they have money now :)

Re:Lawsuit in the oven

[larry bagina]

Aside from the fact that non-compete agreements are unenforceable in California, their employment contract (if any) was with (and written by) PA Semi.

Re:Lawsuit in the oven

[mjwx]

I can feel a lawsuit coming...

Why?

Because that's the kind of business Apple is, they don't innovate, they litigate. It's been this way for a while, the look and feel lawsuit is a great historical example.

Re:Lawsuit in the oven

[tokul]

I can feel a lawsuit coming...

It is America. Lawsuits are always coming. Get used to it.

Re:Lawsuit in the oven

[L1feless]

They left to a company which was their own making chips. Apple also acquires chips. Google bought out their company. Technically their previous company would or could have been deemed a competitor. I hope Apple tries to sue and looses. rather than a lawsuit I smell an innovative product coming to market soon.

Re:Lawsuit in the oven

[dave87656]

Typically, a group like that builds a chip or architecture and learns from the experience. Sitting around having a few beers or in the cafeteria they start talking about how they could make the bestest, badest, most energy efficient or highest performance or whatever chip ever, given what they know now. Someone knows someone who works higher up at google and pretty soon they are talking.

Android

[Enderandrew]

Google wants to make phones, netbooks and tablets. They've been investing money in coding for ARM, but it makes sense to look into producing their own chips for these devices.

Re:Android

[Jeng]

With everyone saying ARM is so efficient how are we to know that Google is not investing in ARM to make more energy efficient ARM based servers?

Re:Android

[WrongSizeGlass]

Is Google branching out into too many areas? Phones, tablets, energy, fiber, etc ... and now chips? Makes me wonder if they are going to turn into a 'Jack of all trades'.

Re:Android

[Halo1]

With everyone saying ARM is so efficient how are we to know that Google is not investing in ARM to make more energy efficient ARM based servers?

It's definitely a market that ARM itself is also targeting: ftp://ftp.cordis.europa.eu/pub/fp7/ict/docs/computing/arm-emre-ozer_en.pdf (see e.g. the EuroCloud project mentioned on slide 3)

Re:Android

[jgagnon]

But master of none, to finish the phrase. ;)

Re:Android

[Anonymous Coward]

Hey, scamming companies for ineffective advertising isn't going to last forever. One of these other things better take off soon....

Re:Android

[MBGMorden]

Google's famous for being a risk taker. They try a ton of stuff and keep what seems to work. It keeps them fluid. The search engine game was good for them for a long time but they seem to get the hint that the way of the world is "innovate or die", so they're branching out. Gmail/Google Apps and the Android seem to be working out for them pretty well.

I think that this branching out is just a sign of a company doing the right thing and keeping active rather than resting on it's laurels.

Re:Android

[Anonymous Coward]

The entire reason that Google is so successful is that these ads ARE effective, and measurably so.

Re:Android

[RyuuzakiTetsuya]

That's fine. While I love my iPhone and I'll gladly shill Apple's crap for free, there's hardly a master of any of these(I'll argue that yes, Apple deservedly top of the heap for mobile devices; but if there were no iPhone, Android *would* be the sexiest thing on the block; even if there are huge, radical flaws with the platform). When it comes to maps, fiber, search, even Usenet, Google does things well.

Re:Android

[Anonymous Coward]

Nah. They're looking into cutting operating power for data centers, just like Microsoft.
The time of portable clouds is coming.

Re:Android

[Fentekreel]

Not really android is pretty much just what a tablet and touch screen interface is for. as for fiber they've been doing that for a while i'v read. I hope the are a JOAT soon :) i'd take their internet gladly.

Re:Android

[Hylandr]

This was modded funny but I think the poster is dead on. Smart for them to diversify and hang onto the trends that stick. Or at least last another 10 years before the next big thing.

- Dan.

Re:Android

[poetmatt]

apparently nobody paid attention to the fact that every android phone runs on arm, aka google is already investing in ARM.

Re:Android

[StrategicIrony]

They are.

Google has job postings all over advertising looking for infrastructure architects with ARM experience and solid state drives, with an aim of reducing power consumption.

I'd have to guess that was a server play, though maybe it's a handheld one....

Re:Android

[Enderandrew]

IBM was known as a typewriter company. Diversification and investing in future markets is a major reason IBM is still a major player today, even though the typewriter business is dead.

Re:Android

[feranick]

Very possible. Even Microsoft (of all software houses), may be thinking along the same lines in a recent job posting...

http://hardware.slashdot.org/story/10/04/20/2134213/Job-Ad-Hints-At-Microsoft-Move-To-ARM-Servers

Re:Android

[Dahamma]

Google's famous for being a risk taker.

They are?? Seems like a "risk taker" wouldn't have required months of research and usability studies to decide on a simple color change or that they should in fact increase the font size of their home page by 2 points...

Re:Android

[naplam33]

They're trying to find something that works besides online advertising. It's called not putting all your eggs in one basket... but it's not an easy thing to do, most businesses do have all the eggs in one basket. But so far Google's branching out is a FAIL. I'd keep trying too, though.

Re:Android

[maxume]

Google needs to try a little harder, Google Groups is one of the biggest sources of spam for discussion groups, to the point that many people simply block it altogether.

Re:Android

[Anonymous Coward]

Google's famous for being a risk taker.

They are?? Seems like a "risk taker" wouldn't have required months of research and usability studies to decide on a simple color change or that they should in fact increase the font size of their home page by 2 points...

There is taking a pointless risk and taking a calculated one, you wouldnt find any google executives playing Russian Roulette, in the same sense that they are going to be very careful not to ruin the number one way people access google.

Re:Android

[inKubus]

Seems smart. Why am I paying for (and paying to power) this Intel floating point unit when I'm only serving web pages?

But, Google's growth has been perpetuated by use of cheap commodity hardware, ie: profiting off the fact that the rest of us drive the price down by buying lots. A switch to specialized chips would mean a new direction for the company.

I mean, they could have done this from the beginning with fancy IBM or SUN or Unisys mainframe stuff, which typically allow you to configure IO subsystems (which is the main bottleneck of web serving). Likewise if they are doing database stuff you'd want a lot of RAM and wide I/O bandwidth, 128 bits or more. All standard for a long time on IBM stuff. But it's expensive, not commodity. Even Google's 100-300K servers (or maybe it's a million now, who knows) is not going to bring the scale of the whole worldwide market for Intel chips (100M plus annually).

So I don't see how this could benefit them long-term. Sure, power savings might add up to a lot so it's a good investment. And since they want to be the entire Internet (including your desktop), it's really a matter of energy over all else. But they are definitely going to need to keep adding hardware to keep growing, so that means higher chip expenses upfront. But, if they can spin the same processor into a little home or mobile computer to connect to their services, they might be able to start leveraging this scale thing again. But it seems to be a big risk to get into the manufacturing business.

Re:Android

[Areyoukiddingme]

What's even more amusing is that some of the seed capital that founded ARM Holdings came from.... Apple corporation.

Re:Android

[tcjr2006]

I like turtles.

Re:Android

[maxume]

In soup?

Or do you prefer them smushed flat on the road?

Re:Android

[billcopc]

The reality is that Google is not a monolithic entity that does everything poorly. I don't think the search team has much to do with the Android team. They probably don't have many "jacks of all trades", but instead have many small, focused teams that are really good at what they do.

Just because they share the same brand name doesn't mean they all work out of one homogenous brain-pool.

Re:Android

[socz]

i don't hide any comments for bet better moderation!

Re:Android

[Anonymous Coward]

This is the 21st century. Simply thinking beyond next quarter's profit statement is considered "risky" these days. Businesses these days are just that myopic.

I'm no Google fanboy, but I have to admit that they are probably one of the last few technology companies out there that has an actual skunkworks policy in place. So funding all these dumb ideas knocks down their earnings 5%. Big deal. But if they ever hit on something big, it'll be worth it. The rest of the suits in all the rest of the corporate boardrooms just. don't. get. that. They'd rather have $10 today than $10000 next year.

Re:Android

[net28573]

Keep in mind. Microsoft started this company war. is there a translation of the concept of using a nuke in the corporate world, because i wouldnt be suprised if someone pulled out a big trump card.

Re:Android

[Dahamma]

Don't get me wrong, I think their "20%" concept, their "labs/beta" products, etc, are great ideas. I just don't think they are "risky".

I would define risky as taking a big enough chance on something that if it goes badly it could result in significant downside. Palm betting their future on a new phone OS... that's risky (oops!) Google allocating a higher-than-industry-average R&D budget? That's somewhat innovative, and probably good business practice, but not particularly risky.

In fact, they do some up with a lot of innovative and interesting product features, but I am having a hard time even thinking of a Google product that is not either an improved version of an existing idea or an acquisition. Not that there is really anything wrong with that, when the core business is as wildly successful as theirs...

Re:Android

[wealthychef]

He is talking about those that smear the inside of his underwear.

Re:Android

[XiaoMing]

But master of none, to finish the phrase. ;)

At the same time, that's the philosophy that's allowed Google to grow so big without having been hit by so many (successful) lawsuits like Microsoft.

They blatantly hoard more information (your personal info, books, voice, etc) and often offer poorly created/executed (anyone feel that Buzz?) products to either get more of your information or sell more ads, but do so at such a level that for the common person it's not worth caring about.

If anything, I think they've mastered being an unintrusive jack-of-all-trades.

Re:Android

[pspahn]

Power savings fall neatly into both categories. They're not making video cards yet, so it's safe to assume that power savings is one of their most legitimate concerns.

Re:Android

[Anonymous Coward]

Custom chips make sense from the start if you're building a mobile device that you want small, light, and power-thrifty, as others have said above. Especially if the current leader Apple has theirs locked down, if you want to innovate in this area you'll have to create down at the hardware too.

Re:Android

[Anonymous Coward]

not much on the .com website, http://www.agnilux.co.uk looks like they missed a trick

Advantage Google...

[kclittle]

... Apple to serve.

Re:Advantage Google...

[toastar]

Google definitely wins todays headlines

Re:Advantage Google...

[Anonymous Coward]

Apple to win tomorrow?

http://www.thisislondon.co.uk/standard-business/article-23826703-city-aflame-with-takeover-talk-of-arm-and-xstrata.do

Did Google get the goods?

[StCredZero]

I wonder if Google just absconded with the real value in the chip company?

Re:Did Google get the goods?

[frank_adrian314159]

I don't actually believe that Apple lost a lot of momentum in this defection. In general, it's the folks who had alternative ideas for an architecture that didn't win out that tend to leave. I bet Google eventually gets stuck with something like a chip that has an insane pipeline structure, makes a different power/speed tradeoff (and probably for the worse), or has some other weird bag o'crap bolted to the side.

In addition, there are a *lot* of chip designers still left over from the chip manufacturer layoffs after the crashes of .com and bank that Apple can pick from. When you get right down to it, once the architecture is developed (and, by this, I don't only mean ISA, I also include things like clock/signal distribution, internal bus structure, pipelining strategy, cache structure, etc.) which only takes a small team - in fact, it works better with a small team - it's just moving little mask rectangles around with automated logic generation and, after that, a buttload of QA.

So, even if Apple sues, it's just because (a) Apple like using its lawyers and (b) it kicks Google in the nuts - it really doesn't mean much for Apple's future chip development.

Re:Did Google get the goods?

[KonoWatakushi]

Some of the key talent, yes. However, that talent would have been wasted at Apple, just as their fine processor was. I'd be surprised if any of top people stayed long, after Apple tossed out their years of effort, only to have them doing ASIC monkey work.

It disgusts me to see how Apple chews up brilliant companies only to scavenge a few bits and pieces.

Re:Did Google get the goods?

[CAIMLAS]

As someone else said, ARM may not be the best platform for Google's servers.

If what you say is true - they're the divergent group of engineers - then it's quite possible that Google hired them for that very reason. Maybe they wanted to build a PPC architecture for Apple's iPad, or something else entirely. We do know that these engineers were focused on power efficiency, and that is a very big concern for Google (it's a large part of their operations costs, after all). There are a lot of ways to save power with a processor; switching architectures may not be the intended result. Maybe they're going to try and improve on the x86 processor (and then license the tech to Intel to make the CPUs for them).

That said, everyone and their mother seems to be going ARM these days. It wouldn't be surprising (to me) if Google has that in mind, too.

Maybe Google feels theatened by Apple

[Orga]

Apple has a good handle on their vertical, from hardware to content. Google is just beginning its jump into the hardware portion. I imagine this is just another rung in the ladder from the bottom to the top, control all the way.

Re:Maybe Google feels theatened by Apple

[Nadaka]

Google has had their hands in some aspects of hardware for a while now. They have their own custom designed server motherboards.

Re:Maybe Google feels theatened by Apple

[Meshach]

Google has had their hands in some aspects of hardware for a while now. They have their own custom designed server motherboards.

But this is their first into the whole "digital music/iphone/ipad" world. Apple has dominated that world for quite some time so it may be good to get some competition.

Re:Maybe Google feels theatened by Apple

[MBGMorden]

Um, the G1 - Google's first phone - came out 2 years ago. 1 year after the iPhone. They aren't exactly new to that space.

The "iPad" world? It was released 18 days ago. Apple is as much of a newbie on that scene as everyone else.

Re:Maybe Google feels theatened by Apple

[ColdWetDog]

Google has had their hands in some aspects of hardware for a while now. They have their own custom designed server motherboards.

Next - Google Chips,
Then - Google Sand!

Re:Maybe Google feels theatened by Apple

[Amouth]

Custom designed server boards aren't as hard as you think - based on what CPU and Chip set combo you plan on using there is a starting working reference design for you to customize (only exception was when VIA reverse engineered their own P4 chip set using no reference)

but the point being - it is a lot easier to do something custom if you do it by modifying something that you already know works..

Re:Maybe Google feels theatened by Apple

[Andy Dodd]

Um, Google didn't design that. HTC did. It's not very different at all from HTC's Windows Mobile products.

Even the Nexus One isn't a Google design - it's an HTC phone carrying Google branding. (Which is very common, HTC has ALWAYS been very rebranding-friendly, it is only very recently that you started seeing the HTC brand in the United States even though HTC phones have been in the USA for quite a while.)

Re:Maybe Google feels theatened by Apple

[MBGMorden]

Um, Google didn't design that. HTC did. It's not very different at all from HTC's Windows Mobile products.

Google laid out the platform, which is what people care about. Hell when the thing first came out it was just called the "Google Phone" by most people. Google providing the firmware while other companies deal with the hardware seems to be how they're approaching this market, and they seem to be doing well with it.

Re:Maybe Google feels theatened by Apple

[raddan]

For internal consumption only, though. This may be a move into producing hardware for consumers who are less, shall we say, tolerant of addenda to manuals like "Oh yeah, that rail is +5.5V and not +5V. Sorry."

Re:Maybe Google feels theatened by Apple

[Anonymous Coward]

>>> Apple has a good handle on their vertical, from hardware to content.

That's what this entire subthread is about. That Google is, for the first time, going to be able to control the entire stack - something that Apple right now is dominating.

Re:Maybe Google feels theatened by Apple

[JamesP]

Then Google Planet Formation, Google Accretion Disk, all the way down to the Google BigBang I assume

Re:Maybe Google feels theatened by Apple

[Daengbo]

It's not. Google's CEO Eric Schmidt stated very clearly that Google isn't interested in being in the hardware business at the Atmosphere event.

Re:Maybe Google feels theatened by Apple

[Anonymous Coward]

And even then, its easier still, since you just have to partner with another company to make your boards for you. Heck, I know half a dozen companies locally who get their systems made by another manufacturer, they even get all their branding all over them.

Re:Maybe Google feels theatened by Apple

[mjwx]

Apple has a good handle on their vertical, from hardware to content. Google is just beginning its jump into the hardware portion. I imagine this is just another rung in the ladder from the bottom to the top, control all the way.

You seem a bit confused.

Apple designed the vertical integration from the word go, Google designed an open platform from the word go. What you state is the antithesis of an open platform

If Google is entering the consumer hardware business which is pure speculation at this point, but not without merit, the Nexus Two will be the Motorola Shadow but who is to say what the Nexus Three will be. If Google is entering the consumer hardware market they will be doing it as an Android competitor, not an Android controller. The introduction of the Google Nexus One had not impact on the introduction of the HTC Desire, which uses the same hardware as the N1 but with HTC's "Sense" interface.

Google has no interest in becoming a vertical monopoly.

Re:Maybe Google feels theatened by Apple

[Jedi Alec]

Any chance we can squeeze in a Google Dyson Sphere somewhere along the way? Personalized ads wherever you are on its surface?

"architected"?

[Anonymous Coward]

I did not know "architect" could be used as a verb. Let me try: Frank Gehry architects buildings for a living. Emperor Palpatine tried to architect the downfall of the rebels.

Re:"architected"?

[wonkavader]

How about...

"The firm's enormous security guard reacted quickly to the arrival of Carlos the Jackal. Reaching for the closest blunt instrument at hand, the guard picked up I. M. Pei and architected the terrorist to death."

Re:"architected"?

[dfsmith]

That word has been Englished. You've got to love type-loose languages: there are so many opportunities for verbing nouns. You should try it---I think you will have an empower.

Re:"architected"?

[StrategicIrony]

I just looked and the Oxford dictionary recognizes it, and points to archived letters from Keats in 1813 using the word as a verb.

It was considered an "overly formal" usage for awhile, but I think the use in computer-speak has brought it back toward mainstream.

Re:"architected"?

[WrongSizeGlass]

I did not know "architect" could be used as a verb.

Heck, it worked for "Googled" ;-)

Re:"architected"?

[chill]

A +4 Insightful the hard way, congratulations.

Re:"architected"?

[Jeng]

Even better is nouning the verbs.

What the fuck does that mean?

Re:"architected"?

[StrategicIrony]

uhm...

I think it would be "empowerment"

I'm can't think of any verbs that can't be nounded, but maybe my imagination is failing me. LOL

There are plenty of nous that can't be verbed easily. (such as noun and verb and door, etc)

Re:"architected"?

[diamondsw]

Grammatical or not, it's been standard parlance for at least a decade in datacenter (and probably other) IT. Everything gets verbed eventually. See?

Re:"architected"?

[Anonymous Coward]

You could verb door and have it make reasonable sense:

"My boss doored me after I skipped work for a week."

Re:"architected"?

[poopdeville]

You can give any phrase sense. There is no such thing as a private language.

Re:"architected"?

[tzot]

It's called NewSpeak. Don't worry. Your head will clear and your worries will be gone as soon as the war ends.

Re:"architected"?

[tom17]

Here, have some eat.

Re:"architected"?

[grcumb]

"The firm's enormous security guard reacted quickly to the arrival of Carlos the Jackal. Reaching for the closest blunt instrument at hand, the guard picked up I. M. Pei and architected the terrorist to death."

Who said crime doesn't Pei?

Re:"architected"?

[Anonymous Coward]

Why Insightful? Architect is a perfectly cromulent verb.

Agnilux conspiracy

[Anonymous Coward]

Agnilux = Ga! Linux

One Word

[codepunk]

GPad

Re:One Word

[Anonymous Coward]

We can hook up our trendy new GPads using our WiFi GSpot.

Re:One Word

[CharlyFoxtrot]

Mountain View, start your photocopiers.

Re:One Word

[StrategicIrony]

Sounds like an interesting sex toy!

Re:One Word

[Lemmy Caution]

Connected, of course, by GStrings.

Re:One Word

[Anonymous Coward]

This is a marriage with Google from people once in Apple.

Thus: NeePad.

Re:One Word

[Anonymous Coward]

Just make sure someone isnt using on the iPad first.

Re:One Word

[Monkeedude1212]

No, they sniped the engineers, not the marketting team.

Google isn't THAT dumb.

Re:One Word

[Anonymous Coward]

Well... there is a first for everything... at work...

I'm going to the bathroom.

Re:One Word

[Anonymous Coward]

Why pass up the opportunity?

GSpot by Google.

Re:One Word

[Hurricane78]

And then, about 5 years later, the WinPad.
Right with a “early experimental hacked together” OpenPad, which is only actually usable 5 years later,
and looks and behaves exactly like the WinPad with some missing smoothness, which itself behaves exactly like the GPad and iPad only a lot more annoying and with shitloads of security holes. ;)

Meanwhile in useful innovation land... nobody has ever heard of *Pads. ^^

Re:One Word

[Anonymous Coward]

It'll probably be marketed as a "iPad killer".

Re:One Word

[StrategicIrony]

I guess you would get a "giPad" which sounds like something much more smelly...

Re:One Word

[drewhk]

Or rather:

uPwnd

Re:One Word

[Anonymous Coward]

And then, about 5 years later, the WinPad.

Also, paradoxically, called FailPad or BSODPad.

Re:One Word

[bigngamer92]

Agreed. It'll probably have some geekier name from Star Trek or something.

Re:One Word

[Anonymous Coward]

Meanwhile in useful innovation land... nobody has ever heard of *Pads

Of course they have, they're the things some women use when they have their period, right?

Servers

[imgod2u]

Google has been spending a lot of effort -- from custom power structures inside their buildings to buying that magic box that generates power form minerals to custom-making their own server blades -- to reduce power and make energy efficient servers; they have so many of them after all.

These guys, while formerly PA Semi, focused their new business on energy-efficient server CPU's. So I wouldn't so much expect a gPad. It's likely the consumer will never see the chips that are being produced here.

Re:Servers

[MBGMorden]

It's also unlikely that Google would ever see a ROI from purchasing an entire semi-conductor company just to save a bit of money on the chips in their own servers. They're planning on marketing these to SOMEONE. They might be part of servers, or they might be part of a pad (I think Android-based pads are an inevitability myself - it's just a question of if these will power them or not). Heck they might be part of a set-top box or something else, but they are definitely planning on selling these chips as part of commercially marketable products.

Re:Servers

[imgod2u]

We're talking millions per year worth of money for powering their server farms. The custom blade design alone saved them a few million annually. We're not talking small money, especially as they keep expanding to meet the needs of the tubes.

Re:Servers

[MBGMorden]

I know we're talking millions per year powering their server farms, but keep things in perspective. NOTHING will eliminate that power bill. All something can do is REDUCE it. So you have to look at the difference in power consumption as the return. Total power consumption is moot. Now consider: how much did they pay for this company? We don't know, but semi-conductor firms tend to run in the HUNDREDS of million of dollars. You also have to take away from that the cost associated with RUNNING this company. Salaries, power bills associated with the extra facilities, equipment, etc.

Lets assume that they paid $100 million, which is a lowball estimate. Lets assume that merely switching architectures is going to save them $5 million per year in power costs, which is being VERY generous there. Now lets assume that this semi-firm can run with only $1 million in annual funding - lowball again on that. Even being INCREDIBLY generous with all assumptions there they're looking at breaking even in 25 years.

It makes no sense. Even at Google's scale, it's almost never a good idea to own a whole company to produce products that only you yourself use, unless they're INCREDIBLY simple devices (which computer chips are not).

Re:Servers

[Areyoukiddingme]

It should be fairly obvious that your estimate of the acquisition price is substantially wrong. Yes, semiconductor manufacturing firms are worth hundreds of millions of dollars. Intel and Motorola are worth billions.

But that's not what we're talking about here. We're talking about a startup that doesn't even have a webpage. They have a handful of employees and very likely they have little or no revenue at all. Google was buying a convenient blob of talent, nothing more. I'd be astounded if they paid more than $10 million. I'd be fairly surprised if they paid even that much.

On the other hand, your estimate of how much that talent will cost on an annual basis is probably substantially low. 20 people making senior engineer salaries with benefits could run $3 million per year. Given Google's previous successes at reducing their annual power bill, it's not hard to imagine that team paying for itself and then some, especially considering Google has long since gone global, with data centers all over the world. They're not paying US hydro power prices everywhere.

Re:Servers

[Anonymous Coward]

I know some of the people involved. $100 million as lowball? Damn, they would have creamed their pants if you were on the google side of negotiating that deal. Look, Google paid for the talent not the company. They got a nice little hiring bonus and some good stock options (which is why they didn't stay at Apple).

$100 million (shakes his head in laughter and pity)

Re:Servers

[bill_mcgonigle]

to reduce power and make energy efficient servers; they have so many of them after all.

These guys, while formerly PA Semi, focused their new business on energy-efficient server CPU's.

Yeah, Google has to realize their trends and that they can't cover the entire Earth's landmass with servers or they'll have nowhere for their customers to live.

There was some talk about a packet-processing oriented CPU that would be less general-purpose, but better suited to Internet-style workloads. Maybe get a 30:1 improvement and not have to keep building data centers for a few years.

I worry about robots

[jwhitener]

Now that Google is getting into hardware, it's only a matter of before we see:

    The Google Search (and destroy) Robot. ;)

Re:I worry about robots

[Nadaka]

Its ok, all you have to do is tattoo the following on your forehead and they won't bother you:

User-agent: *
Disallow: /

Re:I worry about robots

[diamondsw]

Thankfully we've been defending against this for years.

Re:I worry about robots

[w0mprat]

Now that Google is getting into hardware, it's only a matter of before we see:

The Google Search (and destroy) Robot. ;)

In that case, any future Robot invasion could be thwarted with robots.txt

Re:I worry about robots

[CAIMLAS]

The mark of the beast is an htaccess file?

Who'd have thunk.

Re:I worry about robots

[thisisntme]

That would be robots.txt file.

Let's ask the Chinese!

[18_Rabbit]

Bet they know what the engineers are doing...

I don't really think it does

[wonkavader]

I mean, there are plenty of companies making ARM chips for phones. Google will want to use commodity stuff for that -- it means that the cost of innovating around the phone platform (hardware side) is someone else's problem, and that's already happening.

On the other hand, they have enormous power bills and would gain personally from computers which do the same amount of work as what they currently have for 1/10th the power.

Google's avoided making their own servers (using a commodity board) because other people were doing that already. There's not much available for ARM for servers, so here it makes sense for them to pick up the ball and make a bunch of machines for themselves.

Google's probably paying very roughly $100 per server every year for the electricity to power the board and then to cool the air it warms. They have something like a million servers (http://www.pandia.com/sew/481-gartner.html), so ARM boards for their data centers could be a huge financial win for them.

Re:I don't really think it does

[Enderandrew]

I know ARM is very efficient for low-power, small devices. But ARM has had Linux support for years. If ARM scaled up well, wouldn't we have seen ARM Linux servers by now?

I'm assuming it doesn't scale up the way Google would need it. If I'm mistaken, please enlighten me.

Re:I don't really think it does

[TheRaven64]

If ARM scaled up well, wouldn't we have seen ARM Linux servers by now?

I suspect the answer is that none of the off-the-shelf ARM SoCs come with the kind of network and disk controllers that servers need. This means that you can't just use off-the-shelf ARM parts in a server, you need to do a fair bit of custom work. A quad-core Cortex A9 with a couple of SATA and GigE controllers on die would be a pretty nice server chip for a lot of workloads.

Re:I don't really think it does

[Enderandrew]

Going back to my question, if this was feasible, why hasn't it been done before?

Re:I don't really think it does

[TheRaven64]

Because you need custom silicon. The people who make ARM SoCs are not in the server market. They'd need to design a new SoC for that market, and they'd then need to persuade system builders to use it. It's a lot more effort than just buying an off-the-shelf x86 chip and using that and it's a lot higher risk because you are assuming that there is a market for ARM-based servers. It makes sense for a company like Google, because they know there is a market for whatever hardware they produce; themselves. They can buy an A9 license from ARM, get some custom bits added to a SoC, and get a company that owns some fabs to run off a few hundred thousand of them, and it's worthwhile.

Re:I don't really think it does

[Anonymous Coward]

It's not that it scales well; it does. It's that nobody has, until recently, fielded silicon that was scaled up to those levels.

Re:I don't really think it does

[pspahn]

How can there NOT be a market for lower power server chips? It seemed there was a market for netbooks. Apparently some people care enough about having a battery die after two hours to want something more efficient. Are you saying that data centers don't want cheaper bills?

Different Take

[JackSpratts]

the times has a different spin. it's not chips so much as low-power hardware/software integration google's paying for. http://bits.blogs.nytimes.com/2010/04/21/google-acquires-another-piece-of-the-tablet-puzzle/?hpw

Doesnt have to be a tablet

[nurb432]

It could be for their back end, to save power over intel.

They might save enough $ in power savings alone to make it worth buying the chip company.. AND not be reliant on anyone else.

Yes but,

[hellop2]

does it run nilux?

Architected?

[cyber-vandal]

Speak fucking English dammit.

Re:Architected?

[Anonymous Coward]

http://www.askoxford.com/concise_oed/architect?view=uk US

Which fucking flavor, dammit? Verb, past-tense. Not even of the oddly-ending varietal. Architoke? Architaken? Architoken?

Re:Architected?

[Anonymous Coward]

Bullshit. 'Architected' is a buzzword used by the article to obfuscate and dazzle; the correct term is 'designed'.

Re:Architected?

[Pulzar]

In ASIC world, to "design" means to write RTL code (and all that follows it) that matches the desired architecture. To "architect", means to write the high-level spec of what the design should look like.

It's not bullshit, it's the proper terminology for the topic at hand. You could argue that the terminology is lame or whatever, but it is what it is.

Web is capitalized?

[fortapocalypse]

Why would Google acquire anyone with a website ( http://agnilux.com/ ) that currently states "This Web site is coming soon." Are we supposed to be capitalizing the word Web in website, and why can't it be one word? This acquisition is causing us to reassess fundamental assumptions about spelling and capitalization of common terms! Aha- now I realize why they acquired them.

Re:Web is capitalized?

[billsayswow]

Well, these guys did work on the iPad. I'm sure after working for Apple long enough, one really starts to lose track of capitalization rules, product names or otherwise. iMac, iPad, iPod, iPhone, etc.

Is it wrong...

[billsayswow]

...that when I saw this headline, for some reason my first thought was "Oh, Frito-Lay is gonna be pissed, Google is moving in on them now..."

A4, PA Semi, what?

[Yvan256]

Are you telling me that Apple bought PA Semi in April 2008, the formerly PA Semi team designed the A4 and then they left Apple to start their own company which Google bought right away?

Probably something new, not ARM...

[KonoWatakushi]

A 32-bit architecture like ARM really has no place in Google's servers, and it is hard to imagine that those who jumped ship from PASemi/Apple would want to do the same sort of ARM integration monkey work at Google.

It is a shame that Google didn't pick up PASemi before Apple wasted their processor and years of effort; the PA6T would have served Google very well. I expect that Google is thinking long-term here, and we may even see a brand new 64-bit ISA, something that scales well from phones to low-power servers. (Okay, that may be a little hopeful, but I expect something new and interesting in any case.)

Re:Probably something new, not ARM...

[JamesP]

A 32-bit architecture like ARM really has no place in Google's servers,

Why are you so sure about that??

I'm thinking that, for what Google does, it fits like a charm.

Think 4 core ARM chip (maybe) with 4GB RAM and an SSD running using 10W. Compare that with the, I dunno, 50W maybe 80W current x86 + HD server

Especially since Google doesn't care (much) about server performance rather than shoving as many servers as they can onto their datacenters.

Re:Probably something new, not ARM...

[KonoWatakushi]

That may seem fine now, but core and memory density will continue to grow. It just doesn't scale, and in a few years, the efficiency advantages will be lost with increasingly integration. For example, a single 32-core/32GB machine vs. eight 4-core/4GB machines, 8 network connections, etc. Also, 32-bits of address space is already insufficient to properly utilize 4GB of physical RAM.

The fact is, it is a dead end, and I don't think Google is short-sighted or arrogant enough to employ some of the most talented people in the industry to glue together some custom ARM design. If they were happy doing that, they would still be at Apple.

Up until now

[mjwx]

Up until now I've been saying this Google-Apple war is very one sided with Apple doing a lot if ineffectual attacking with Google (removing the word "google" from the Iphone, various comments including the two "porn store" comments). But this changes that.

Google I would say have just pulled off their own Doolittle raid, whilst completely ineffectual from a business standpoint it does send an important message to Apple, don't forget we can strike anywhere, even in the very heart of your business.

On the plus side, Google diversifies and gets insight and input into ARM development.

"architected"???

[nuke99]

"architected"??? I suppose they "partnered" & "networked" at work, and in their spare time I bet they "imaged" and maybe even "gifted". ugggggh!!! Last time I checked, the English language had lots and lots of verbs. Using nouns as verbs is not only unnecessary but really annoying.

Rockefeller

[Anonymous Coward]

Hopefully this doesn't give any substance to those already crying Anti-trust foul on Google. Rockefeller had the same idea of controlling all aspects of his industry by incorporating everything he could into his company and in turn was forced to break up for being a "monopoly".

Re:Rockefeller

[sowth]

If Microsoft wasn't broken up and its leaders put in prison for its obscene anti-trust violations and "questionable practices", I don't see how the government could possibly prosecute Google with a straight face.

Here's a nice start

[symbolset]

I'm sure someone is working on this. Creative has the Zii that looks promising. In addition to a Cortex A8 it has a stemcell 64-unit floating point processing element array and all the System on a Chip stuff. All that and 512MB RAM fits on a SODIMM.

Up to now while the performance per watt for ARM was fantastic, it just wouldn't clock up enough to make it worthwhile to gang them on that scale. Now that they have ARM running at 1GHz they're going to give it a go. I imagine you could fit hundreds of those little rascals in a 2U chassis. At Google's scale they're probably cheap too. They'd probably save so much money on power and cooling the things would pay for themselves.

ARM up for grabs?

[symbolset]

I saw some speculation today that Apple might buy ARM. Although that could happen, Google's the one that needs the exascale computing (pdf).

Since Intel seems determined to go down with the WinTel ship, somebody oughtta.

The name Agnilux

[slack_prad]

Agni is Sanskrit for fire and Lux is the Latin for light.

I wasn't surprised to see that founder is an Indian

Re:The name Agnilux

[thalassinos]

Agni is also Greek for ''Pure''.

These guys had nothing to do with iPad

[gig]

These are people who left PA Semi as soon as it was bought by Apple. They had nothing to do with iPad.

Apparently, ex-Intrinsity people had more to do with the A4 than ex-PA Semi people also.

The rumor now is that Apple will buy ARM, which they co-founded to make the Newton.