November 06, 2009

􀂃 Based on our checks with foundries and NVDA's comments overnight, we continue to see upside to capex estimates fueled by demand for leading edge capacity, a heightened sense of competetion in the foundry space, and constrained spend from memory makers that will likely enable a multi-year upcycle ahead. While some may point to near-term peak in Samsung/TSM capex in 4Q09, we point to a wider breadth of spending from the likes of Hynix, Chartered, UMC, Toshiba, Global Foundries etc. that should drive Q/Q growth for semi equipment orders in 1Q10 and into 2010 and 2011. Our favorite names in front-end continue to be VSEA and LRCX
followed by CYMI and KLAC. We would also point to VRGY (as primary test supplier to NVDA) as positive lateral to NVDA print.

􀂃 Insufficient 40nm Capacity: Overnight, NVDA discussed 40nm capacity constraints (not new news) at its foundry supplier TSMC. Mgmt discussed yields improving but that allocations still remain inadequate. Though yields were improving, they were important enough for TSMC to mention a chamber matching problem on its
conference call earlier this earnings cycle. One way to improve production quota is by improving yields (technology and process control buys), the other is by increasing capacity (capacity buys), both positive for
equipment. Finally, what we found interesting was NVDA's comments that they have visibility throughout all of 2010 as a result of 40nm capacity constraints. This clearly, in our minds, suggests sustainability of spending
from the foundries.

􀂃 Our Checks on Capex Suggest TSM/UMC Spending will Grow 70+% in 2010: Our most recent checks this week suggest TSM is planning to spend ~$3.5-4.0B in capex in 2010 (vs. cons ~$2.7B) and UMC is planning to spend ~$1.0-1.5B (vs. cons $750M). Add it up, and these 2 players alone will grow capex by ~70+% in CY10.
This may prompt GlobalFoundries to scramble to keep up with the foundry investments. All in all, we continue to see a heightened sense of competition in the foundry space developing in the 2010-2011 timeframe and that we will reach peak foundry spending in 2010 following underspending over the last 2-3 years.
􀂃 Lastly, with changes to our TSM and UMC budget estimates, we now formally model 50% growth in capex in 2010 (old 45%) to $33.4B. We would note that we continue to see upside to our estimates for Samsung, Hynix, Elpida, Toshiba which could lead actual capex to equal +65%-75% when all is said and done. And then
importantly, we see continued spend from memory to support continued growth in capex into 2011 and potentially beyond.

􀂃 With front-end shares trading an average of 30% discount to our price targets driven by our 2011 normalized earnings estimates (based off modest WFE estimate of $24B), we continue to like front-end shares. Most potential upside, per our estimates, is with VSEA and LRCX followed by CYMI and KLAC