Chip manufacturers continue to play a canny game, holding back on capex while enjoying the higher prices that follow shortages.
Shortages are particularly severe at TI which has, according to a report from FBR, increased some prices by 40%.
Some DRAM prices have tripled in the last year while Semico Research predicts another 60-70% rise in 2010.
Days of Inventory (DOI), as measured by iSuppli, are close to historical lows at nine of the 10 supply chain nodes tracked.
FBR sets out the competing possibilities: ''Currently, visibility into 2H component shipment trends remains limited, though OEMs are largely constructive on end demand. Subject to much debate is whether contracting lead times will cause a soft landing in 3Q or 4Q with muted 2H seasonality, or whether contracting lead times will cause a hard landing with component shipment shrinkage in 2H."
FBR reports shortages at ST, ON, Broadcom and Fairchild, and says that multi-layer ceramic capacitors are generally in short supply.
UMC and TSMC have production queues, though companies can jump the queue by paying more for their wafers.
Yet industry capex remains constrained, and iSuppli puts its finger on why: "One reason for the lack of interest in expanding production is that fourth-quarter earnings for component suppliers were exceptionally high."
In other words: when you've got your customer by the throat - squeeze.
TOMORROW MORNING: The Ten Best Semiconductor Scientists
My my ... and I thought we were now living in an enlightent chip world where we all loved each other and foundries were going to save the world by supplying ever cheaper wafers at ever decreasing geometries to an ever growing market of fabless and fablite firms? Shame the thought of paying a premium for faster delivery and worst still squeezing the customer in times of shortage. Hang on though ... if I'm a fablite company ... I am the customer! Surely the foundries won't squeeze me; I've just sold my soul to them with no turning back ... worried from Eindhoven, Munich, Dallas, Austin, etc etc etc.
Rumor has it that home business unit of a once great European company, is really getting their **** squeezed. They bid low (sub 25% Gross margin)to win some big name sockets and now their customer wants them to deliver but the chip supplier is raising the prices. Ouch...
Why is it that these guys cant ever seem catch a break?
And to think if the EC or German government had provided a relatively small loan to Qimonda (absolutely minor by banking numbers) it might have been Hynix that went bust and Europe would have a highly profitable DRAM business again.
But of course we were assured by the panel of 'experts' at the European Nanoelectronics Forum last November that Qimonda was a dead duck and not worthy of such support as the future of European semi companies was 'fab-lite' (sic).
What foundries are we talking about anyway? There is one player, TSMC, which has 50% market share and probably 70% or more of the money that is to be made. UMC is hanging in there at 15% market share without any hope of catching up as TSMC will outspend them. The others don't really matter.
In the end TSMC can or will dictate how much your fabless or fablite company will grow and how much profit you are allowed make.
The problem is that I do not see a way out. It will help if Globalfoundries becomes successful. But that will not be short term and I fear it will not be enough.
I couldn't agree more, Ozzie, and Globalfoundries isn't spending enough to become significant.
Ozzie, What I find more disturbing is the growing trend for Fabs to be happy with their fabless partners for a given market segment.
So if you turn-up with a great design solution (small area) that will enable you to have half the cost of their existing customer (who is dominating the market) than don't be surprised when the wafer prices offered are a little disappointing!
This is a real problem for mixed signal designs because TSMC typically has much tighter parameter spreads than all their competitors, so yield at the competitors is often problematic.
Unfortunately this means that smart fabless companies need to know who their fab partners are already in bed with, and for which sockets!
There are 2 sides to the coin. And I think it is pertinent to look at both of them.
If you stop cooking at home and complain that the only restaurant in the neighborhood is too crowded and expensive and not-as-good-as-home..well, who is to blame?
If cooking at home was a lot of labour, terribly expensive, in-flexible etc etc...you might have been easily enticed by the fast-food shop. But if you are not the only one being enticed, then expect to see a queue.
Ultimately, you have the choice of waiting at the shop, or (if you have what it takes) opening up a new fast food joint yourself(if the limited scale of home cooking is seen as too expensive). In short, shut-up or start-up.