Samsung及Globalfoundries結盟宣告其14 FinFET製程首度採用3D電晶體架構，相較平面電晶體架構的20nm製程，速度可以提高20％，功耗則可降至35％，晶片尺寸也可減少約15％。早先言明，其採用20SoC製程的產品不擬在2014年上市的AMD率先呼應，成為首家採用14FF製程投片的業者，應用在內建CPU及GPU的行動裝置AP上。



This offers a huge competitive threat to TSMC and provides an alternative to those companies who are looking to use leading-edge production. It’s a huge deal.
- Jim McGregor at Tirias Research

If Samsung and Globalfoundries bring the technology to the market in the time frame they are promising, it could put them at least six months ahead of the competition and tempt multiple chipmakers to give them orders. The partnership would also reduce the cost for chipmakers of alternating between the two suppliers, because products wouldn’t need to be redesigned, as they do now when companies switch. Now you have two well-respected organizations combining forces. For a customer, if you can truly have multiple sources, it changes everything.
- Len Jelinek, an analyst at market-research firm IHS Inc.

Samsung’s announcement […] is consistent with our Moore Stress thesis, and a big step forward to disrupt TSMC’s leading-edge business model by enabling a superior foundry process node with approximately a one-year lead over incumbent TSMC. Beyond enabling a stronger competitive position for Samsung in mobile devices, we continue to view TSMC disruption as a positive for MicronU, Sandisk, and Intel. Longer term we believe Samsung would be well positioned to take >50% share of the mobile device market, enabled in part by disruption of TSMC’s leading-edge foundry model this year.
- Sundeep Bajikar with Jefferies & Co.

We see the announcement as no surprise, given our having previously seen “2 new 193 nanometer scanners from ASML” installed at Global’s “Fab 8.” We believe this agreement is a landmark event that has the potential to massively disrupt the current state of the foundry industry, where TSMC has a near monopoly on the leading edge process nodes. The joint program offers a single process design kit (PDK) and volume manufacturing at four different fabs with identical processes: SAS (Austin), GF Fab 8 (Malta), S-1 (Korea) and S-3 (Korea). Multi-sourcing has become a huge issue with the foundries, particularly at the advanced process nodes. In 2012, Qualcomm suffered from severe shortages of 28nm Snapdragon devices that impacted their revenue when TSMC struggled with surging demand. In addition to the multi-sourcing benefit, we believe the Samsung/GF alliance will capture several major customers because of their technology lead with 14nm FinFET. Industry sources point to Samsung’s investment for over 3 years developing FinFET, giving them a sizeable lead on TSMC. The Samsung/GF FinFET is up to 15% smaller than TSMC’s process due to aggressive gate pitch and superior layout schemes. Additionally, the 14nm FinFETs offer a 20% improvement in performance and a 35% reduction in power compared to existing 20nm technology. We believe QCOM is also actively engaged with Samsung/GF at 14nm for next generation designs. While this may be surprising given the company’s long standing relationship with TSMC, we think there are several factors that swing in the Samsung/GF direction. The multi-sourcing arrangement alleviates QCOM’s concerns about having another severe product shortage like they did when TSMC ramped the 28nm node. The Samsung/GF FinFET technology lead and smaller transistors gives QCOM a time to market lead and superior integration capability versus their competitors.
- Steve Mullane and Paul Peterson of BlueFin Research Partners

Although Samsung, GF, IBM and STMicro have all been closely sharing R&D through the umbrella of Global Alliance, the Apr-17th joint press release takes this collaboration one step further and involves mass production of multiple versions of 14nm FinFET technology. This way, customers ]no longer would need to have different design rules nor mask sets and can send the same design for mass production[/COLOR […] We find the above news as negative for TSMC because:
(1) one way or another, the above announcement is aimed at taking market share from TSMC, and [COLOR=Yellow]lower ASPs is the most effective way of doing it.
(2) There are only a few customers that would be able to afford to migrate to 14nm; Apple/Qualcom (in AP), Nvidia/AMD (graphics), Altera/Xilinx (FPGA). Any market share loss at TSMC would lead to significant margin erosion since there may not be another customer with sufficient volume to fill the void left if any of these customers were to defect to Samsung/GF.
(3) Samsung/GF can offer mass production of chips in US, closer to the above customers and also a positive factor for the likes of Apple/Qualcom as they can also claim enabling job creation at home. All TSMC’s fabs are in Taiwan.
(4) TSMC’s 16nm FinFET Plus, which is closer to Samsung’s 14nm FinFET, won’t be available until 2H15 vs. Samsung/GF 14nm FinFET availability by YE14.
- Mehdi Hosseini at Susquehanna Financial Group

While TSMC will get the boost at 20-nanometer, it is likely to split those volumes across Samsung, Global Foundries, and TSMC at 14/16-nanometer and beyond, which will prove a negative for TSMC by the late 2015-to-early 2016 timeframe. TSMC's top- and bottom-line growth could come under pretty significant pressure as a result of a margin squeeze for leading-edge foundry capacity as well as share loss.
- Ashraf Eassa, an independent research contrarian.


蘋果於2015年據聞已同時對台積的20奈米製程、以及16奈米FinFET+製程下單來看，台積於16奈米製程已有了大客戶的背書，取得了好的開始。關於三星與格羅方德結盟對台積可能形成的衝擊，市場似乎也無需太早、太過擔憂。
- moneydj@Apr 18'14


往昔TSMC以40/28nm領先策略在45/32nm節點勝出的優勢，或將在Samsung/Globalfoundries的技術與市場互補聯盟正事成軍後，產業態勢呈現可能的轉捩點，後者的14nm製程以領先競敵16nm一季度之姿投產，扭轉落後窘境。由於IDM霸主Intel已昭告，其跨足foundry領域不在爭霸，而是挹注未來先進製程研發經費，是故TSMC真正的對手明顯已是星格聯盟。雖然TSMC過半的市占優勢形同護城雷池，短期內依舊強不可撼，但星格聯盟在14nm製程量產上趕超TSMC，不啻一針提振士氣的強心劑。