不知道有沒有 UMC Vanguard 相關


特別是 analog IC 不是須要高壓 process ..

GOOGLE
牺牲利润换市场，Intel吐血价推销Atom平板处理器

Intel公司是PC產業的老大，他們的經營風格可以說是高投入高產出，毛利潤常年維持在60%左右，以致於很多PC公司實際上都是在給微軟、Intel這樣的老大打工而已。但在X86市場之外，Intel的方式就不適合了，為了在移動處理器市場站穩腳跟，Intel幾乎是半價出售Atom處理器，平均每個處理器售價只有20美元，而Intel要想盈利，Atom的售價要達到51美元才行，Intel現在簡直是吐血大拍賣啊 英國Bernstein Research公司的分析師Stacy Rasgon發表了一份報告，根據Intel財報會議上公佈的資料研究了Intel公司平板處理器的價格策略。Intel去年營收527億美元，其中平板處理器貢獻了1.5%的營收，今年營收預計不變，而出貨量則會從去年的1000萬增長到4000萬，以此來算每個平板處理器收入只有20美元。
當然，實際算起來會更加複雜。處理器的成本不僅有製造成本（大約12美元），還有研發、測試等，還有對OEM廠商的補貼等等。根據這位元分析師的計算，Intel的Atom處理器實際上要買到51美元才有錢賺，現在20美元的平均價格算得上是虧本甩賣了。
不管怎麼算，Intel在Atom處理器上不賺錢甚至虧本都是肯定的，再舉個簡單例子，Atom四核版核心面積約為102mm^2，官方標價37美元，給OEM廠商實際上只有20美元左右，而Core i5/i7四核核心面積177mm^2左右，賣價能達到200美元左右，價格差距可窺一斑。

=> TSMC + ARM 會板倒 Intel x86 ATOM ??
Intel 推 ATOM 且 bay trail 都出來了
但是 平板電腦如看 win8 + x86平板電腦 市場雖然有
但是 Android ARM or Ios + ARM 有變多嗎??
先前 Z2760 CPU 還一堆廠商推出平板
但是到 bay trail CPU 好像變少..

Samsung Galaxy Note 3 installed with a 2.26GHz quad-cored MSM8974


Alcatel One Touch Idol X+ equipped with a 2GHz octa-cored MT6592 SoC

[YOUTUBE]usVANklyZxQ[/YOUTUBE]

MT6592 PK MSM8974

測試內容主要是執行自Google Play下載的遊戲軟體，在總合評分上，前者領先18%！

the MediaTek chip particularly stood out for its battery life — a metric that mainstream customers generally consider to be at least as important as max performance. Although the Snapdragon is a solid all-round performer, MediaTek’s power efficiency was very impressive and it led to a higher overall result… In our preliminary scores, the MediaTek chip is better at playing popular 3D games, downloaded from the Google Play store, because it can play them smoothly for longer periods.
- Sri Kannan Iyer, CEO of GameBench, a UK-based chip analysis company

For the time being, I actually don’t think much will change, due to a flaw: the MediaTek chip doesn’t support LTE, and in 2014 a lot of people are going to want LTE phones, including in China. That’s probably why the stock market has so far barely reacted to MediaTek’s latest chip or to the optimistic projections in its recent quarterly report.

Once LTE becomes a commonplace feature, attention will return to the more traditional aspects of what makes a good mobile processor — namely, price and performance — and this is exactly the sort of battle MediaTek is equipped to fight. By the time MediaTek starts delivering LTE-enabled processors in Q3, MediaTek should at least be able to mount a challenge for the mid-range tier, which cost $300-$400 and use lower-specced chips like the Snapdragon 400.

驍龍不是有人說高溫下CPU 會自己降頻嗎???

Peter兄的疑問是？

這次MTK的MT6592(8核)配上MTK optimized linux kernel大成功後, 再加上Android的 多核+64bit 支援度也會隨著Google的開發而越趨完整, 還有TSMC的先進製程(16nm FinFET)也算順利.

基於以上這些點, ARM陣營在mobile market還是一片光明呀, 效能+功耗 持續進步中

INTEL應該不會太短命，氣應該會很長
因為上層是JAVA，chip/driver只要能持續拉近跟ARM陣營的差距
它就有理由繼續在這一行打拼
雖然一開始拉近差距是最容易的~.~

實際上，應更像是『Intel x86 ATOM會板倒TSMC + ARM ??』


在smart device領域，ARMdroid無疑是壓倒性超越Wintel的。去年第四季前，曾有人看好Wintel附贈Office的Bay Trail平板大放送會在聖誕旺季爆量；現在看來不是那麼回事，因為Intel行動晶片部門當季的虧損未曾顯著減少。




究其因，平板的消費行為目前仍屬休閒娛樂模式的media tablet，而非生產力導向的PC tablet，縱使Wintel自認為大幅讓利，再便宜也不盡然能比得上最低價的Android平板。同時，最有立場推廣Wintel平板的傳統PC大廠，在PC銷量衰退之際，就算得了Wintel這次的便宜，project投入所耗費的自身資源仍無可迴避，不是完全佔盡上風；檯面上情義相挺的，其實算酬庸性質，給個面子，要不就直接推Android平板，連中國產SoC也肯用。


Intel 2013 holiday offerings：


Lenovo Miix2
• Windows 8.1
• 8" 1,280×800 IPS screen
• Bay Trail-T 1.5GHz Atom Z3770 or a 1.3GHz Atom Z3740 Intel processor
• 2GB RAM，and 32/64GB ROM
• starts at $299.99/349.99


Acer Iconia W4
• Windows 8.1
• 8" 1,280×800 IPS screen
• Bay Trail-T 1.8GHz quad-core Intel processor
• 2GB RAM，and 32/64GB ROM
• starts at $329.99/379.99


Dell Venue 8 Pro
• Windows 8.1
• 8" 1,280×800 IPS screen
• Bay Trail-T 2GHz quad-core Intel Z3470D processor
• 2GB RAM，and 32GB ROM
• starts at $299.99 and can be paired with the optional Dell Active Stylus for $34.99.


Dell Venue 7 and Venue 8, two affordable, Android-based tablets
• Android 4.2.2
• 7"/8" 1,280×800 IPS screen
• Clover Trail+ 2GHz dual-core Intel Z2560/Z2580 processor
• 2GB RAM，and 16GB ROM for Venue 7 and 16/32GB for Venue 8
• Venue 7 starts at $149.99, while the Venue 8 starts at $179.99

面對現實的支拙窘境，迫使Intel在戰術上做出大動作，不僅與Microsoft分道揚鑣，轉投Android，甚至在銷售夥伴上放棄獨賴品牌PC大廠，靠攏深圳山寨白牌，等同宣告：『不論黑白貓，能抓老鼠就是好貓！』的無奈現實。

在Android上，Intel也不奢望Goolge對32位元X86的優化遠景，直接搭64位元話題，釜底抽薪地自行開發64-bit X86核心，讓其低價Android平板年中上市時便有X86 Android可用。


就算Bay TraiI是面向高階平板的產品，Intel迄今已心知肚明，在Apple跟Samsung的銷售強勢下絕無空間，轉進中低階區塊是唯一可行之路，而未來平板低價化也的確是莫之能禦的潮流。

The current versions of Intel’s chips, led by “Bay Trail,” are really aimed at high-end tablet computers, and Intel is unlikely to get much traction in that part of the market, which is dominated by Apple and by Samsung.

Those tablets, made by challengers such as Asustek and Lenovo, are increasingly cheap models, mostly sub-$199, whose sales were stronger in Q4.

不諱言，投資人對Intel在smart device的後知後覺，固難以期望有後發先至的奇蹟，現在反而更疑慮Intel對低價平板的補貼政策無法翻盤，且將侵蝕其PC本業獲利。這些補貼金，Intel稱之為『contra revenue』的支出，實質上抵銷了相對營收。

Intel is subsidizing its relatively high-priced “Atom” chips, such as Bay Trail, with rebates, which Intel refers to as “contra revenue.”

Bernstein Research的Stacy Rasgon本身便是唱衰Intel，抓到Intel『4000萬台的平板補貼將令其毛利率折損1.5%』的話柄，深究出Intel口稱20鎂的每台補貼（損失）不足彌補Bay Trail處理器成本，宣告Intel在平板這塊的實質虧損要更大。

And after rebates, the effective economics Intel is enjoying on tablets this year likely involves revenues close to zero, and profits that are negative (of course)….



這廂Intel聞訊也不甘坐以待斃，財務長Stacy Smith立馬反擊。

Anyone who thinks we have a $50 cost for these chips is wrong. The special costs Intel is incurring that are pushing down gross margin for the company this year are not the cost of the chip itself. Rather, “contra revenue” and “NRE” are meant to reduce the “total bill of materials.” The chips themselves will maintain “competitive prices” compared to offerings from other vendors without Intel having to sacrifice gross margin, even for tablets whose retail price is $199 or lower.

Intel is able to cost-reduce to match the lower prices of such products. The cost to Intel to produce an Atom chip is one-fourth the cost of its top-of-the-line “Core i7” parts. New versions of Atom chips next year, “Broxton” and “Sofia,” will bring down Atom's costs further. This is going to be even more so the case as we sell Quark, some in the single-dollar price range.

憑藉其『外星人科技』，Atom晶片不過是Core i7的四分之一成本，次代Broxton與SoFIA晶片還將持續探底，連針對穿戴式裝置的Quark也能1鎂造就。（委外的SoFIA也受惠於Intel製程功力？）
PS：Intel晶片相關die size
102mm²- 22nm四核Bay Trail-T
160mm²- 22nm四核Ivy Bridge
174.4mm²- 22nm四核Haswell（mobile）
177.0mm²- 22nm四核Haswell（desktop）

We were originally targeting the $500 tablet market with our chips, because we have been bringing really high performance into that market. But the bill of materials for Atom chips at this point has very little to do with the SoC itself. The memory that works with our SoC is high-end, more like what you'd find in a PC. And so we are doing the value engineering to bring down the cost of that to our partners. That takes a little time to do. But by the end of this year, the bill of materials delta, if you will, gets pretty small.

打造500鎂高階平板的Atom晶片價位並非造成終端成本高昂的主因，匹配Atom高效能所需PC等級的周邊用料方是元兇，但在Intel帶頭降價下，年底前將消弭整體BOM的差距。

We don't fear lower price points; it is in the nature of technology for prices to come down and for things to become more and more useful, so that people adopt the technology.

科技平價化是必然，低價始終來自人性。

People made the same kinds of arguments about us, and how we faced lower prices then. But people always underestimate Moore's Law and our ability to bring down the price of manufacturing. Intel will be able to maintain gross margin even as it sells chips at lower prices, holding to the company's long-term goal of corporate gross margin of 55% to 65%.

今日外界對掌控造晶成本的懷疑，Intel並不陌生。90年代末期segment-zero千元PC浪潮來襲時，當年一樣有類似疑慮，擔憂Intel的獲利前景。



在過去八年間，僅管X86架構持續擴張，Intel仍不斷讓造晶成本下滑（尚未納入Atom產品線的數據）。

首輪政見發表終結，具有雪亮眼力的選民們，票往何方？

PS：以財務穩健著稱多年的Intel，其CFO一職向來是鐵飯碗，現任董事長Andy Bryant在此崗位歷練達13年，Stacy本人也已上任7年，說不準也是未來的董事長。其「錢財不是問題，前景不會低迷」的發言跳脫守財奴的保守本位，足已代表該司核心立場。

Now figuring a Baytrail might average $20, the 40 million "samples" is worth about $800 million. The cost of these parts would be another $400 million for a total loss of $1.2 billion. This approximates the gift that Intel will be giving to tablet manufacturers.
- Russ Fischer@Nov 23 2013

They indicated that Bay Trail is still too costly to reach the mainstream and value price points needed for volume, so they will be investing (covering NRE and contra-revenues (rebates) that will impact gross margins by ~150bps (or ~$800M in investments). Coupled with increased targeted investments, they expect the Other IA revenues to be roughly flat, but with an operating loss even higher than 2013′s staggering $2.5B.
- Stacy Rasgon of Bernstein Research@Nov 22, 2013

Half of the mobile business is sitting at Samsung, one third of it going to Apple and two thirds of it captive to Samsung products. What is left is about 500 million units at $20 (optimistic) or $10 billion total business. About half of the investor meeting was dedicated to convincing us that Intel will win a significant piece of the mobile business, and I'm sure it will. If Intel attacked it like a junk yard dog, it might come away with about $5 billion.
- Russ Fischer@Nov 23 2013


In the past, Intel has entered new markets (e.g., WiMax) with subsidies that offset less attractive solutions and, once again, that is the plan for the tablet market. Our issue is that the ROI has not been supported by the opportunity. The company is paying ~$800mil to enter the mobile market early via NRE and contra-revenues, which is slated to impact gross margins by ~150bps. This spending will result in only a 3% CAGR in total mobile growth, when including Atom plus notebook. So what value does the $800mil add, and what is the ROI? These are the main questions that were not answered to our satisfaction at the analyst day. We believe the implications of the spend could prove to have a greater impact in the near term vs. long term. In the near term, we believe that $800mil could buy potentially ~20mil in incremental units. This would suggest ~$400mil in incremental revenues, and assuming 50% operating margins, this would imply ~200mil to operating income and represent an ROI of ~20%. In the longer term, the impact of the $800mil spend is less meaningful, as the market typically rewards winners as the market matures […] Our analysis suggests that Intel’s combined notebook (performance processors) and smartphone/tablet (Atom-based) processor revenues should grow from ~$17.6bil in 2013 to ~$19.2bil in 2016, or a CAGR of 3% from 2013 through 2016. While we see the total Atom SoC mobile opportunity growing from $380mil in 2013 to $3.7bil in 2016, or a CAGR of 114% from 2013 through 2016, we expect PC-related revenues to decline from $17.3bil in 2013 to $15.4bil in 2016, or a CAGR of -4%.
- Doug Freedman with RBC Capital@Nov 27, 2013

Even thought this market is likely to be in the hundreds of millions of units annually, how much revenue opportunity is Intel likely to enjoy? On those numbers, maybe a couple of billion dollars worth even if they get it all (which they likely wouldn’t). And even if they could somehow capture that much of the market, it wouldn’t go very far in moving the needle on a $50B+ company.
- Stacy Rasgon of Bernstein Research

雖然身家夠本，砸大錢所得的市場回報充其量不過數十億鎂，相對於總營收規模500餘億鎂的Intel，意義不大。言下之意，Intel應該退出這個紅海，另謀高就。

In the near term, the volume in mobile will not significantly change the revenue growth story for Intel given that the ASP for both tablet and smartphone processors is significantly lower than PC and servers. However, We believe that for investment sentiment on Intel stock, the success or failure in mobile is very significant.
-Daniel Amir at Lazard Capital Markets@Nov 22, 2013

In terms of units, the smartphone market is nearly 4x the size of the PC market, and more than 100x the size of the server market […] We believe this strategy is worth taking a risk given the fact that the smartphone market is 3 – 4x the size of the PC market, and growing at a 20%+ rate, while the PC market is shrinking.
-CLSA Asia-Pacific Markets’s Srini Pajjuri@November 18, 2013

即便大餅當前，沒有優勢，緣何Intel前仆後繼，執迷於mobile市場？面子問題使然抑或挽救股東信心？還是真有鴻鵠將至，不容錯失？

No one thing moves the needle for a $50 billion company. But when you add up data center, and our software business, the NAND [flash memory chip] business, the Internet of things business, growing 15% to 20% per year, it adds up. We want to have a nicely growing PC market, and maintain position there, and then have these new markets where, if it connects, it does it best with Intel.
- Stacy Smith

即便22nm的Merrifield挾其效能與功耗或能提振戰績，但仍不足以震撼ARM陣營。Intel也有自知之明，直到其LTE modem與Atom的整合方案問世前，都可能只是個智機市場的見習生。藉由合併Infenion，Intel基帶能力已不差，後又併購Fujitsu在美國的射頻芯片部門，增加了LTE-A RF收發器的設計團隊。預計其生聚的戰力將在2015年浮現。

There are currently more participants in the baseband market than there are profitable tier one customers: Qualcomm, MediaTek, Intel, ST Micro, Broadcom, Marvell, and NVIDIA, among others. Collectively competitors ex-QCOM and MediaTek are spending over $3bn on R&D for less than a $3bn 2014 revenue opportunity in the market for LTE baseband. LTE may ultimately be a catalyst for new entrants but in 2014, the move to LTE likely makes things tougher, not easier, as multimode capabilities continue to be key. Qualcomm and MediaTek are only getting stronger, especially in serving high end phones from Apple and Samsung, which is where the revenue and profit opportunities lie. By year end 2014, we’d expect to see significant reductions in R&D from at least some of the participants.
- Joseph Moore, semiconductor analyst with Morgan Stanley

不過大摩沒那麼樂觀，在高通跟發哥手底下，就算是Intel也沒啥機會的。

There will be just two suppliers at the end of the day in mobile. You can decide.

史密斯先生看法不同，更語出驚人：最後只會有兩家在行動晶片市場存活。哪兩家？不告訴你。

總的來看，2013跟2014年，不期望Intel對ARM陣營造成顯著的衝擊；而ARM架構跟Foundry技術也將在兩年內遭逢效益瓶頸，屆時，或是Intel最後一擊的tipping point，再錯過，可不只難過了。

Intel looks forward to providing its services to many companies that “want to use the Intel architecture.” There will be a number of partners and customers who will say “we want to use these leading-edge transistors in segments of our business.”

關於未來Intel的foundry事業，Stacy的回答與市場認知再度扞格，讓人迷糊了。意指一堆人想用X86架構讓Intel代工。


Wearable devices as well as 'Internet of Things' (IoT) devices are the next growth drivers for hardware vendors that can offer the required energy saving technologies and energy harvesting technologies to realise these applications.
- Masatsune Yamaji, principal research analyst at Gartner

在高度擁擠的智機紅海之外，穿戴式裝置與物聯網似乎獨立其外，儼然下一個藍海。


Morgan Stanley table of Internet of Things suppliers

以Intel領頭羊的製程功力，在NTV跟STV這方面另闢戰場，轉換起跑點，更有看頭。