<strong>Semiconductor</strong> <strong>Equipment</strong> Technology Hardware but to keep spending to maintain their leadership in the consumer devices market. Where are we in the cycle? We believe we are halfway into the current recovery cycle, and will enter a more muted growth phase by the end of 2013 or early 2014: ASML, AMAT, TEL, ASMI, KLA and LAM have all shown improved order intake/shipment levels from the beginning of 2013, compared to end-2012; hence, we believe we have entered a recovery cycle from the 2012 trough. We believe we may exit the recovery phase and enter a more muted growth phase of the cycle by end-2013/early 2014 for the following reasons. 1) Samsung and Intel’s capex is likely to be spent in H2 and will drive order growth in H2 2013. 2) TSMC, Samsung, Intel, Globalfoundries and other logic/foundry players have all budgeted higher/similar capex compared to 2012, and total logic and foundry capex is currently at a historical high (2013E: $37bn). After 2013, we expect foundry/logic capex to remain at flattish until 2016. In addition, TSMC has indicated that its 2014 capex will be similar to the 2013 level. We expect the solid order intake level to be maintained after 2013/early 2014, but growth is likely to be muted. 3) Memory spending has already picked up from the 2012 trough level since mid-2013. ASML, KLA and LAM have all indicated that memory spending is recovering. We estimate memory capex to grow at 11-15% in the next four years. However, as it only accounts for 29% of total capex, which is too small to drive significant capex hikes. Historically, memory spending used to account for up to 57% of total capex and hence had a greater impact on overall capex. Will the Apple/Samsung/TSMC/Globalfoundries shift be a zero sum game? In our view, any shift is a small positive for equipment vendors, depending on which of them Apple is shifting to: Since last year, there has been much speculation about whether Apple will shift its foundry orders from Samsung to TSMC/Intel, and whether Apple will buy fabrication plants (fabs) from United Microelectronics Corporation (UMC) or Globalfoundries. In our opinion, these shifts can be a small positive on a net basis for some equipment vendors, instead of simply a zero sum game. The new foundry is likely to build capacity which was not required previously; at the same time, the old foundry may cut its original spending level, but it may have already committed some capex as fab-building projects are multi-year projects. If Apple moves to TSMC, we believe ASML, AMAT, ASMI, and LAM will all benefit, as: 1) TSMC’s 20nm is lithography-intensive, and it only buys from ASML; 2) AMAT has a strong relationship with TSMC (although the decrease in the Samsung order may have a negative impact on AMAT); and 3) TSMC 20nm is high K metal gate (HKMG)-based, which requires the ALD tool from ASMI. If Apple chooses Intel for its foundry orders, we believe TEL may benefit, with ASML also benefiting, but to a lesser extent, as 1) TEL has a close bond with Intel in relation to its etching tools, 2) AMAT does not have a strong relationship with Intel, 3) ASML may receive new orders, but it has to share Intel orders with Nikon, and 4) Intel’s capacity is 100% HKMG-based, which only leaves limited upside for ASMI. 8
<strong>Semiconductor</strong> <strong>Equipment</strong> Technology Hardware If Apple buys Globalfoundries’ fabs, we believe ASML, TEL, ASMI, AMAT and LAM may all benefit, as 1) leading-edge nodes are lithography-intensive, providing upside for ASML, 2) TEL is strong at Globalfoundries in etching, 3) LAM has close relationships with foundries in etching, 3) AMAT has strong links in deposition with foundries, 4) Globalfoundries’ HKMG adoption rate is currently quite low, so it may require more ALD tools from ASMI to ramp up capacity as required. If Apple buys UMC’s fabs, we believe ASML, ASMI, AMAT and LAM could benefit as 1) leading-edge nodes are lithography-intensive, and hence are a strong suit for ASML, 2) LAM benefits from good relationships with foundries in etching, 3) AMAT is strong in deposition with foundries, and 4) UMC may need to ramp up its HKMG processes, which would benefit ASMI. 9