July 26, 2011 - Based on meetings and observations from this year's SEMICON West, analysts Satya Kumar (Credit Suisse), Tim Arcuri (Citi), and CJ Muse (Barclays) -- all panelists from the Thursday morning Bulls/Bears session -- have some top-takeaways list for the industry: why WFE spending is slow, why it's only a short pause, and which will comes first, EUV or 450mm.

The correction's worse than thought... The consensus is that order pushouts are significant, causing caution among suppliers and weakness in 2Q11 and especially 3Q11. Kumar now projects 3Q11 orders will be below $30B-$34B (normalized) to $23B (annualized). Citi's Tim Arcuri sees the "run rate" at around $20B, well below the $29B he deems "normalized." Wafer fab orders are down nearly -50% from peak levels in 4Q10, similar to the 2004 correction, he notes.

Kumar is lowering his full-year semi capex outlook again, just weeks after dropping his view to 7% growth (~$31B) from 15%-20%; now he's thinking more like flat to up 10% to $29B-$32B. Likewise, Barclays' CJ Muse is lowering its 2011 wafer-fab equipment outlook to $31B (7% growth), down from $33B. He's maintaining his 2012 outlook of a "muted decline" of -3% to $30B.

Who's to blame for the softness? Foundry spending is slowing and even pushing out orders, everyone seems to agree (analysts and suppliers). Arcuri points specifically to 28nm yield problems encountered by key foundry customers (QCOM, XLNX, NVDA) and delay of TSMC's Fab 15, though he thinks "most of these issues could be resolved quickly, and we could see fab projects resuming soon" if end demand keeps up. Muse calls out TSMC and UMC for 2Q11 pushouts, and GlobalFoundries "and others" in 3Q11; GloFo's problems extend beyond yield issues to its management uncertainty, he adds, and there's confusion about investing in both Dresden and Malta operations. And there's "no sign of life yet in DRAM spending" thanks to sluggish PC unit growth and ratcheted-down bit growth estimates, notes Muse: "any hope for DRAM recovery in 2H11 (as hoped for by AMAT, LRCX, etc. as of a few months ago) has evaporated, we believe, driving a more subdued 2H11 outlook."

Kumar, meanwhile, says weakness "is across the board," including not just foundry and DRAM but even NAND, all with softening orders over a six-month period (March-Sept.), and with all top equipment purchasers "showing signs of caution."

...But it's just a pause. While all signs point to a soft 3Q for suppliers, this could only be a short-term correction. Total fab capacity is still -6% lower than 2008's peak, while IC units are 33% higher, and only ~25% of announced new fab capacity has been equipped so far, Arcuri points out. That aforementioned run-rate gap (~$20B vs. $29B normalized) "is clearly unsustainable," he says, so he predicts a bottoming-out in 4Q11.

Muse too believes in a short-term "pause" in semicap, though his window of inflection for an orders rebound is a little wider, possibly leaking into 1Q12. The $40B WFE order runrate in 4Q10 was clearly a peak; today's ~$22B runrate (by his calcs) is similarly too low, so there'll be a ramp-up to get to $30B WFE in 2012.

Everyone's seeing softness. ASML's 2Q11 orders were €840M; (flat Q/Q and below €900M-€1B guidance), and the company says 3Q11 will plunge -40% to below €500M. Early in the week NVLS said 2Q11 bookings sunk -25% (2Q sales were ok though), and will be down another -10% in 3Q11. TEL told investors that its 2Q11 semicap orders will be down significantly (-25% in USD, -28% in ¥), and will be "directionally down" in 3Q11, suggesting similarities to NVLS' expectations. For ASML, see above. AMAT said its 3Q11 sales are "tracking to guidance" (which in May was at -3% to -10%), but did not update 3Q11 orders or 4Q11 sales. And Lam Research could see a -20% to -30% dropoff in 3Q11 orders, Kumar, based on execs' comments that SAM is "flat to down -10%," and assumptions of -5%-10% shipments (due to share gains) and better 4Q11 shipments sequentially. KLAC, meanwhile, which was "an outlier" when it reported flat 2Q11 orders, guided slightly better for 2H11 (-10% to -15%, vs. -20% fears and -25% from aforementioned suppliers), probably due to strong demand for its reticle product and bare wafer inspection and increased business from INTC. (The company wins more points by upping its dividend, he added.)

EUV vs. 450mm: Who blinks first? Most industry companies and watchers expect EUV lithography to get pushed out yet again, to ~2015 for a volume ramp, and then only for DRAM to start. (Note the two most tightly-tied-to-EUV suppliers, ASML and CYMI, are still bullish on EUV's sooner adoption; ASML is still saying 40wsph by end of this year and 60wsph by mid-2012. And for his part, Barclays' Muse says that while EUV progress has been slow, "adoption at least for certain layers is inevitable.") Such a pushout obviously dents ASML, but it would help others e.g. KLAC and probably AMAT and NVLS, Kumar suggests -- and ASML gets a benefit from double/multiple patterning, the presumed short-term alternative to EUV, which will require more litho tool purchases.

Meanwhile, around the same timeframe (maybe) is when people are talking about the possibility of a 450mm wafer-size transition. Several firms at SEMICON West were talking about tools that were mainly 450mm "capable" -- though for some that's a lot simpler to do, e.g. just a bigger clamp or chuck, while for others it's a lot more work to improve patterning, processing, inspection, and manipulation of bigger substrates.

Judging by the numerous presentations and forums on 450mm, and our own chats with attendees and analysts, there are two camps: those who think 450mm is increasingly inevitable, and those who still doubt that what is purely an economic decision for chipmakers will take any precedent over the numerous economic/technical challenges involved with upcoming nodes and chip architectures. No less than Applied Materials has pledged $100M to develop 450mm tools, with tens of millions pledged by other suppliers. It sounds like a lot of progress has been made in tooling and 450mm-supporting technology -- as well as in swaying suppliers' mindsets -- but the cost issues surrounding 450mm for both suppliers and chipmakers still needs to be hammered out.

More M&A, stock maneuvers. With the two biggest M&A deals in the industry largely in the rear-view mirror (AMAT+Varian Semi, Advantest + Verigy), there may be more coming. TEL said it has an "open mind about inorganic growth." LRCX had been M&A-chatty earlier this spring when its stock was in the $50+ range earlier this spring (since then it's dropped to the low-$40s and such chatter is muted). Kumar notes that those who are serious about M&A probably are careful not to broadcast their intent too much. He thinks both LRCX and TEL "are potential suitors for NVLS," with valuation a sticking point. Teradyne also has been in the M&A game but probably will prioritize stock repurchases over buybacks.

Barclays' Muse agrees that a big catalyst approaching the semicap's market trough is about companies' cash strategies, citing KLAC's increased dividend and stock buybacks for LRCX and NVLS -- and LRCX's apparent about-face regarding potential dividends of its own.