Reports of the death of the gallium arsenide industry have been greatly exaggerated with GaAs ICs just short of being a $4 billion market last year, according to analysts Strategic Analytics. The 2008 market, at $3.9 billion, was 8 per cent up on 2007.
GaAs, derisively dubbed: 'The technology of the future and always will be', is a consolidating industry - the top three players, RFMD, Triquint and Skyworks commanded 59 per cent of total worldwide GaAs revenues last year, while the top ten had 82 per cent.
The main market for GaAs is MMIC-based products for at wireless apps.
Apart from the top three, leading players were: Avago, M/A Com, Mitsubishi, NEC, Toshiba and Eudyna.
Back in 1999, Allen Podell, founder of GaAs IC developer Pacific Monolithics wrote a paper for which the abstract read:
'In 1980 the future for gallium arsenide devices and integrated circuits looked very rosy. It was clear that silicon was going to run out of steam as lithography became the limiting factor in raising the high frequency cutoff of the active devices. A change in materials from silicon to GaAs, or some other compound semiconductor, was going to yield a large increase in performance approaching ten to one, if the predicted ballistic effects could be realized at room temperature. The semi-insulating substrate with reduced parasitics was far better than lossy silicon for high frequency circuits. Nothing stood in the way of success.'
Except, and rather importantly, the ability of silicon lithography to stretch all the way to 32nm going on 16nm.
TOMORROW MORNING: THE TEN MOST DESTRUCTIVE M&As
I know I'm a bit late to reply to this, but a thought just crossed my mind: will we still be able to say the same thing about GaAs 5 years from now?
GaAs right now is all about cellular & WiFi front-ends/PAs, it seems. But both domains are under pressure: front-ends because more and more components, especially LNAs recently, are getting integrated in the RF chips.
And PAs because Axiom has a genuinely successful CMOS PA product for the GSM/GPRS market and they want to extend it to EDGE, 3G, and eventually 4G. The only concern would be power efficiency, but they claim their deficit is small and it should eventually reach parity - no idea if that's believable, but that's good enough for the entire low-cost market as far as I can tell. Other companies like Acco and Blacksand are also working on that.
I'm not saying GaAs will die completely, I honestly think it'll remain a fairly large market, but it does seem to me that industry forecasts are much too optimistic for the next couple of years. This is far from being a mid-term growth market in my book - all IMO...
That's interesting stuff, Arun thanks. It's way beyond my competence to judge but it will be most interesting to see how it pans out. Maybe we'll have some completely different material, maybe even graphene transistors, in five years.
It's certainly also way beyond my competence to judge whether CMOS PAs can ever be made 'good enough' for most of the market - it's hard to argue against a company that had already shipped 10M units more than one year ago.
And indeed, who knows what will happen five years down the road :) Hopefully though we won't have run all out of sand because of the gigantic hole the bankers dug for us! ;)
Yes, the customer's usually right. I have to say, Arun, that despite the depredations of the bankers, the semiconductor people I've talked to this week have been modestly optimistic. For instance Rich Beyer of Freescale sees the market 'stabilising', and Malcom Penn, the analyst, thought the semi recovery, when it comes, will be quite steep.
That's good to know - I definitely suspect inventory has gotten too low in some areas (i.e. the PC channel for example; PC semiconductor vendors, however, nearly have too much) and eventually things have to stabilize and even improve quite a bit. Once that happens, however, it all depends on end-customer demand, which will probably take longer to recover... Here's hoping :)