IBM 和 Google 等公司組建的 OpenPower 聯盟，會有何影響？

01-21

《華爾街日報》網路版今天發表題為《IBM組建聯盟挑戰因特爾(IBM Gets Allies to Chip Away at Intel)》的文章稱，IBM 已與谷歌 (Google) 和英偉達等公司組建OpenPower聯盟，共同推進Power架構技術的發展，在伺服器、網路和存儲設備領域向英特爾發起挑戰。 via IBM組建聯盟挑戰英特爾：用Power對抗x86|IBM|英特爾

這個問題我問了下自己的導師Dr.Walker. 其回復如下:

IBM is trying to find a market for the POWER architecture. Since Apple dropped the PowerPC, and IBM lost the game console market, the primary uses of the POWER architecture have been in IBM servers (pSeries), and automotive engine controllers (from Freescale). They formed a consortium power.org many years ago. What is new is trying to get Google and Nvidia on board with a focus on data center processors. I am not sure if anything wil come from this collaboration. Nvidia is trying to sell ARM-based Tegra processors, but was kind of late to that market, so has had only modest success. Google builds their own boards, and doesn』t really care what processor they use – they just want cheap, energy efficient, fast. They have hired chip designers, so maybe they will make their own processors, but I am fairly dubious about that. They own a lot of processors, but I don』t think they have the internal volume in processors for that to make sense. Facebook is trying to get a consortium around a standard data center processor module instead.

So essentially IBM is trying to push POWER into warehouse scale data centers, since an increasing fraction of that market (e.g. Facebook, Google, Amazon) build their own systems, rather than buying from IBM, HP, etc. So IBM would really be in competition with ARM (and companies using it, such as AMD and Nvidia) and Intel.

I don』t see IBM hiring in their chip design business. They have been trying to sell their fabs for several years to Global Foundries (Global wants IBM to give them $2B to take the old fabs), and their hardware business has shrunk by 30% this past year. IBM needs POWER and zSeries processors to sustain its legacy systems, but it will be increasingly difficult to afford the CPU design effort with a shrinking volume of chips. So I expect that they will start to use smaller design teams depending on more automation and less optimization in future chip generations, with more sharing between POWER and zSeries (they share already).

IBM has a track record of selling off underperforming hardware businesses. They sold laptops to Lenovo. Then Intel-based small and midrange servers. Disks were sold to Hitachi, and printers spun off as Lexmark. The only hardware business they still retain is high-end servers, including supercomputers (e.g. BlueGene) and a lot of that is legacy business – corporate mainframes and database servers. They have built special data analytics servers, but that is a new business.

垂死掙扎

謝邀。

IBM的power處理器基於SOI技術，目前主要用在IBM的小型機和大型機上。Intel的X86處理器基於最新的FINFET技術主要是PC市場，這個格局很難輕易被打破。不是技術問題，至少主要原因不是技術。Intel的移動處理器技術也不差，但是市場基本都是ARM架構的，Intel很難搶佔市場。IBM肯定不願看著PC市場長期被Intel佔據，有所行動是應該的，但是想要翻盤，難難難~

intel應該是直接被忽略的

因為這個動作不是在挑戰intel，而是改變當前整個產業鏈的格局和生態

intel所處的產業鏈和生態是wintel及其軟體鏈，目前這一局勢已經在逐漸變化

首先，移動端就沒有intel和微軟什麼事

然後，web服務的服務端intel和微軟占的市場都不是很大

重點是一線互聯網公司和大的組織機構的伺服器怕沒有一家用的是wintel組合

而傳統pc市場正在衰退，五花八門的其它設備正在興起，用的都不是wintel

IBM做硬體的實力是非常強悍的，80年代時更是遠勝於intel，所以IBM的基因是有的

英偉達本來是後起之秀，但居然在把顯卡做高端的同時搞出了GPU，搞得英特卡惶惶不安，一直想吃掉它而不得。

google的格局擺在那裡：web頭號交椅，其天文量級的大數據在互聯網能統管一切；移動操作系統頭號交椅；居然還搗鼓點其它五花八門的硬體；而這三者是完全關聯的，這個就是統治啊。

（其實，對用戶來說，"統治"一般都不是好事，新加坡算特例）

這三家加在一起，wintel再不上路就是作死的節奏了。

所以，完全不是挑戰，真的是忽略啊。

瀉藥。直覺認為這是個偽命題！power再open也無法達到x86的市場寬度，不具備挑戰的能力！power家族最近幾年狀況不好，銷售萎靡，牛人離去，靠著企業級市場的傳統積累，日子還算過得去.ARM的授權模式啟發了IBM：移動端不是power的菜，不如拉人一起玩企業級，擠佔一些x86市場。Google的能力在於數據管理，NV的能力在於超級計算（圖形工作站市場NV已經第一了），IBM的能力在於穩定的市場穩定的支持穩定的供貨穩定的管理等等一系列可以用「穩定」來形容。

結論：IBM要保護自己的企業級服務市場，自己玩太累了，所以。。。。。（）

說實話，如果power 真的很牛逼，利潤大大的，銷量剛剛的，IBM會 open？

intel + windows

intel + mac

個人計算機市場

intel + linux

intel + windows

網路伺服器 + 超算集群 + 圖形工作站

power + aix

大/小型機市場（金融，銀行，票務等要求高原子性的，可擴展性低，價格昂貴）

但是 mainframe 這塊，當年有intel的安騰，但是安騰不敵至強啊，而且intel的伺服器，可擴展性好，價格便宜，linux 發展的很好，intel 每年都投入大量資金資助linux，IBM節節敗退啊

ibm遲早要賣掉power伺服器業務，預計不會超過5年！

intel野心很大，儘管現在是pc市場為主，但是足夠讓IBM未雨綢繆拉人入伙了……畢竟兩家的實力差的不是一點,如果intel砸錢出新技術IBM一家是撐不住的

應該說，聯盟與intel不完全在同一個領域，他們有重合的部分比如intel一直想要進入的行動晶元領域，一方面他們又有自己獨當一面的優勢領域，比如n的顯卡領域，ibm大型伺服器領域，intel桌面PC領域。這樣的聯盟由於他們都是fabless，聯盟可以統一設計資源，開發資源，節約人力成本，建立壁壘，同時與foundry結合，構成一個整體，抵禦IDM垂直與平行的蠶食~~

簡單來說 @gao dexter 引用的英文說的是 ibm想做Intel做的風生水起的數據中心晶元業務，但是很可能無果，因為背後的Google需求不強

其實說反了，伺服器領悟是英特爾向IBM發起挑戰。

聯盟的作用只是拉盟友