Jeff, Gottfried, and all the other interested AMAT followers: Greetings!
I attended the SEMI-sponsored session this morning entitled 'Polishing the Crystal Ball: The Semiconductor Manufacturing Outlook.' As requested, I'll post some of the tidbits I gleaned. (Usual disclaimer applies: If you see anything here that looks like investment advice, you?re on your own!)
This will be a rather long post, and I'm sure we will have ample opportunity to discuss further in the coming days. INFRASTUCTURE was the firm that presented the material; they will have the slides available on their website (http://www.infras.com) later today or tomorrow.
Their presentation, while discussing lots of technical elements, was basically from a financial perspective. After an Introduction section, they presented information on the Business Cycle, Bookings and Shipments, Outsourcing and Supply Chains, and Market Drivers. The first data slide showed growth trends over the last 25 years for Electronics (4% CAGR), Semiconductors (17%), and Semiconductor Equipment (as much as 40% CAGR). Naturally, the variation in electronics was small, while the equipment subsector showed wild swings, including several ?negative growth? periods.
As many of you are painfully aware, we are on our way out of a serious financial 'double-dip' that began in 1998. One portion of that downturn was quite predictable, due to overcapacity in semiconductor manufacture. The anomaly that made the downturn so painful and even frightening was the Asian financial meltdown, often referred to as 'the Asian 'flu.' Interestingly, general electronics demand was constant throughout the downturn period, and isn?'t really showing any unusual growth. Indeed, electronics overall, a $1.2 trillion industry, is expected to move smoothly along to about $1.4 trillion globally in the next few years.
Silicon content, however, is quite another matter. Units are growing at about 10% per year, with revenues therefrom growing at about 15% CAGR. Wild swings are possible, from +40% YTY (year-to-year) or larger, to ?20% YTY. And Semiconductor equipment will have larger swings. More on that a bit later.
Here?s some growth forecast numbers for equipment, from Dataquest and VLSI Research:
99 00 01 02 03 04
Dataquest 13% 47% 36% 12 -8 0
VLSI Res 18 25 25 -- -- --
(One attendee pointed out that VLSI updated their numbers in the past two days, revising this year and next year upward. www.infras.com will show the revised numbers if/when they get them.)
Everybody expects the sector to take a 'small breather' this summer; a 'flat spot' if you will. As one speaker quipped, 'Don't you just want to lie down for a bit after a huge meal?' Nobody is predicting a downturn, although I suspect there will be analysts willing to sound the Chicken Little alarm when their growth curves flatten.
More data, in line with recent trends discussed on this message board, include YTY revenue (not unit production) growth numbers for 2000 by sector: MPU's, 14%; MCU's, 32%; Flash memory, 66% (was 80% in 1999); DRAM 50%; SRAM 54%; Logic, 49%. Some other sectors, such as DSP and Analog, were shared, but I wasn?t quick enough with the pen. <sigh> They all looked good, to me.
Analog devices may grow at 19-20% revenue CAGR over the next 4 years. MPU's, which are becoming commodities according to the speakers, will see about 22% RCAGR. The various RAM chips will see about 35% RCGAR in the aggregate, with some subsectors ahead (Double Data Rate, graphics RAM). In spite of the fact that 64M spot prices dropped to $5 again this week, there is still a projected shortage for SDRAM this summer. A lot is happening technically in this sector of semiconductors, and we can expect to see solid revenue growth as newer products become available this year.
Interesting factoid: More wafer surface is no committed to Flash RAM than to SDRAM! Mainly because the Flash geometries are still rather large. Another memory-related factoid: All memory products, from the 1Kit to the 16Mbit versions, have dropped to between $3.10 and $3.20 per unit and gone flat there, before being replaced by the next generation. If that holds, then there?s not much revenue left in 64Mbit units!
Hottest Chip Markets: Flash RAM (phones, cameras, game systems, telecomm gear), maybe 25-27% CAGR for next 4 years; Analog, about 19%; DSP's (Telecomm, autos, robotics), which will grow from $4 billion in 1998 to about $13 billion in value in 2002; DRAM/SDRAM, mainly because demand and capacity will balance better in the near term than recently, and new products. [Aside: The speakers noted, however, that only the high-tech-uninformed Wall Street analysts expect to get rich from memory products. This has been true for each generation of memory product, apparently.]
Currently, 64M memory units are dropping in price faster than Moore?s Law would predict; that won't stay for long, and a shortage this summer could ?overcorrect' and perhaps even drive prices sharply up for a while.
The semiconductor equipment business cycle is obviously in a strong upward period. Bookings have generally risen 4x in the last 12 months, for pretty much all equipment suppliers. Shipments lag this, because of manufacturing cycle times; that explains book-to-bill (BB) numbers greater than 1, and for some smaller firms, over 2. Back-end equipment is also up, about 4x-5x, on bookings.
The 300mm wafer 'debate' was discussed. Although much has been said recently about 300mm tools, little is actually happening on the production front (when compared to 200mm equipment sales). The speakers look for about 10% of equipment sales in 2001 to be for 300mm tools, with that number growing to 15% by 2003. The 'takeoff' point won't be reached until 2004, when about 30% of IC equipment will be sized for 300mm. This arena is often referred to as 'the playground of the Giants,' meaning only firms like Intel, IBM, the foundries, and so on, will be able to afford the $3 billion fab price tag. Further complicating matters, it's hard to justify capital expenditures now on tools that will only begin to show returns in 2003 and beyond. This industry isn?t used to that sort of payoff lag. Another damaging point, many smaller equipment manufacturers were absolutely killed by the last downturn and its subsequent delay in developing and deploying 300mm toolsets.
Side comment, made by one speaker: 'Those cassettes are huge! And dropping a cassette full of 300mm wafers may be a career limiting move.' This leads to considerations about manufacturing IC's using 300mm wafers: More automation (ASYST?), less people contact, maybe smaller lots/cassettes. These changes all make opportunities for investing!
How does INFRASTRUCTURE suggest an investor pick a 'winner' in a semiconductor equipment stock? The following items were presented:
 Company must have leading-edge technology that fits on the SIA technology roadmap
 Company must have or produce productivity enhancements
 New product enablers are key
 The management team must be responsive (agile was one descriptor)
 Company must have solid financials
The speakers had the following forecast for the summer: Expect high volatility in the semiconductor equipment sector! Bookings will go flat, due to several possible contributors: There's only so many capital $ in the industry, and many have already been committed for the year; some double-bookings will be resolved; lead times are becoming so long for many factories that commitments are hard to make, and harder to defend. So once the BB numbers go back towards 1, and maybe a bit below for any one month, an investor has to ask: ?Will it go back up? Or not? And when?? The way individual investors respond to what they think will happen will cause even more volatility in these stocks.
When asked about AMAT investing strategies, the speakers said, ?Now is not a great time to buy AMAT or any equipment stock.? They defended this by saying, 'Valuations are driven by bookings, which are high; the best time to by an equipment stock is when they're not shipping anything!' True, but they didn?t say what they thought a reasonable P/E on AMAT or any equipment stock might be today. It is interesting that they have recently sold off a lot of their equipment stocks, holding the cash; they moved from 10% cash to 30% cash position early this year. Mostly this was a protective strategy, after having such a good year last year. They plan to go back to 10% cash over time, picking and choosing what they thing the next great deals are (they didn?t specify, even when asked).
Although I would say these speakers were 'momentum trader' types, they pointed out some interesting, perhaps even disturbing, elements about our industry. First off, the weakness of today's cycle time reduction programs is that they make forecasting more difficult and risky. Engineers are familiar with this concept: Fast response systems are inherently less stable, with higher variations. Indeed, our industry has basically two gears: 1) Stop; 2) Go flat out, max acceleration, while strapping on afterburners. It?s not going to get better as long as this factor applies. (One speaker even opined that 'flattening the business cycle' would be a bad thing, because it would require us to make the industry much less responsive. You can't have great agility and no cycles; ask anybody who's worked on highly responsive mechanical or electrical systems and tried to damp oscillations.)
In closing, the speakers pointed to the following for the industry to continue to grow (now and in future upturns):
Companies must align to the ISA technology roadmap
Companies must be ready to integrated processes (technical AND business)
The three A?s apply: Access to business information across the supply chain; Agility; Alignment with business partners.
I apologize for the length of this message, but I actually chopped quite a lot of stuff. It was a Very Busy session!
Mitch |
