SEMI Wafers to Wallstreet – New England Forum March 12, 2015

On March 12 SEMI held a New England Forum breakfast event entitled “Wafers to Wallstreet” with four speakers. The main focus of the discussion was on the “Internet of Things” and the following are my impression from the talks in a bullet point format.

Device Scaling and Performance in the Era of IoT – Gary Rosen, Applied Materials

• New devices and technologies are required for IoT.
• Ion implant has a growing role in precision materials engineering.
• Even though IoT devices may be simple, the servers in the background that support IoT will need a lot of processing power.
• Traditional ion implant has been focused on doping, emerging ion implant applications are in precision materials engineering – using ion implant to change materials properties.
• Companies create baseline flows and then add ion implants to meet their performance targets.
• Hot implant is ion implantation at high temperatures (how high wasn’t disclosed) to maintain the crystallinity of the area being implanted and to achieve high dopant activation.
• Cold implant is ion implant at lower temperatures (once again the actual temperature was not disclosed) to increase damage for amrophization implants.
• First generation FinFETs have approximately 3 doping implants for each material implant, for generation 2 it will be approximately one to one and for generation 3 FinFETs materials implants may outnumber dopant implants.

Semiconductors Everywhere – Mark Thirsk, Linx Consulting

• Electronic systems show a high degree of correlation to worldwide GDP.
• Electronic systems sales are growing as a percentage of GDP although they are still very small on a percentage basis (~3%).
• Industrial and automotive are showing the best electronics growth at 15% and 9% CAGR respectively. Consumer, wireless and wired communications and data processing are all in the 0% to 3% growth range.
• Smart phones and tablets are growing faster than PCs and in some cases replacing them.
• The traditional markets for semiconductor are changing with the PC no longer as dominant.
• Billion of sensors and actuators will be needed mainly made in older technology.

Semiconductors and Semiconductor Equipment – C.J. Muse, Evercore ISI

• This is a demand driven semiconductor cycles, less volatility, more gradual, think it has legs to 2016 and beyond.
• Semiconductors are becoming GDP plus growth – GDP +5%, thinks semis grow 6% for 2015 and 2016.
• A lot more generalists are starting to invest in semis.
• Free cash flow is becoming important; investors are interested in sustained cash flow.
• Investors are recognizing the higher margins in analog.
• He thinks there is room for semiconductor stocks to move higher.
• Smart phones demolished GPS and cameras, think consumer has now bottomed out.
• Automotive content is rising.
• Thinks there will be more consolidation and m&a.
• Growth will be slower but less cyclical.
• Analog is moving into the top ten for market caps.
• Lines are blurring between analog and digital, IoT will need both.
• Thinks Broadcom and Qualcomm are going to need to add analog capability to service IoT applications.
• Capital intensity is rising, 40% increase for 10nm over 14nm.
• Etch and deposition is a key now, foundry and 3D NAND are growth areas.
• Fabless guys are getting squeezed, talk about Apple and Huawei making their own chips.
• Thinks China will be a player in 28nm and lagging edge technologies.

A Wall Street Perspective: How Process Complexity, IoT and China – Weston Twigg, Pacific Crest Securities

• Thinks costs are increasing 20-30% per node versus historical 5-10%.
• Seeing a lot less design starts.
• Thinks timing between nodes may slow plus slower ramps, may see less companies make the change to new nodes, more reuse.
• Value of local processors is declining, more power in the cloud.
• Processor costs were hundreds of dollars and are now evolving to tens of dollars.
• Samsung Xian is at 40k wpm of 3D NAND ramping to >= 100k wpm.
• Demand is growing fastest in lagging technologies.
• China investing heavily, may play well in lagging nodes.

Discussion

• Mark – thinks the real value of IoT is when the data gets up to servers and you have the right kind of expert systems.
• Security concerns could be a possible headwind for IoT.
• IoT can add biometrics for enhanced security.
• New fab construction slowing, a lot of repurposing of 200mm fabs, some 300mm fabs on older nodes in China.
• C.J. – could there be a shortage of 200mm equipment. Will companies make the investments given the new focus on cash flow. TI analog only at 50% capacity, other analog around 80%.
• C.J. – 28nm was a 2 year ramp, 16/14nm will be a 3 year ramp. Thinks 10nm is a key node because it will be a full shrink.
• Mark – tier one guys are going down the roadmap but guys like NXP and ST are using older technologies and their fabs are ageing and spare parts are hard to find. Will they go to 300mm and is there a market for less capable 300mm equipment.
• C.J. – think China is primarily a threat as a fast follower.
• C.J. – thinks Micron will roll up Inotera and Nanya in the next two years.
• Weston – Teradyne has said parallel testing has hit a limit and could drive more demand for test equipment.
• C.J. – DRAM, Samsung had a one year lead over Micron and 6 months over Hynix, what happens when Micron comes up on 20nm. What happens as 3D ramps (presumably NAND), will there be a drop in margin. Intel qualified into iPhone 6 and Samsung Galaxy, thinks Intel will be a real player in SOC. Thinks 10nm and below for foundry really changes their business model. TSMC has too much pricing power, if Samsung stumbles Intel could be a big player. Thinks Intel could be a player at 10nm and 7nm.
• Weston – in response to a question about the next equipment down cycle – 3D NAND and FinFET build outs are driving the current cycle, could be weakness when they end but haven’t see over building that has driven contractions in the past.