Tag: gloflo

Globalfoundries will lose cash for years

The Abu Dhabi government’s investment outfit, Mubadala seems content to underwrite huge loses at GloFo for the foreseeable future.

Mubadala reported that its semiconductor unit ATIC had an accumulated deficit of $1.12 billion as of the end of 2011 and made losses in the past two years.

For those who came in late, ATIC is GloFo’s parent company and is the main focus of its semiconductor work. It does have small interests in Calxeda which is a developer of low power server chips.

To make matters worse, Mubadala admitted that it could not be certain that ATIC will be profitable in 2012 or in subsequent years.

Part of the problem is that ATIC was making further investments in boosting capacity and research in a bid to become a “profitable catalyst” for the emirate’s economic development.

While heavy losses at Globalfoundries might be seen as the fact that AMD has appeared to walk away from the company, that impact might be overstated.

AMD expects to spend more on wafers with Globalfoundries in 2012 than it did in 2011. The fact that it is not part of AMD also means it can try to get deals from the company’s rivals.

It would appear that much of the loss is the huge amount of investment ATIC is chucking at GloFo. It is not clear what its payoff will ultimately be and it appears to be happy to not make cash on its investment for years yet. 

AMD should move from GloFo to Samsung

While the rumours are that AMD is looking to shifting a large chunk of its chip production away from GloFlo to TSMC, it seems that the fabless chip maker might have a better long term back-up plan.

Kitguru  claims that AMD is considering a more radical shift and is actually looking towards Samsung to provide its chips in 2013 and 2014.

Samsung has everything that AMD would need in a production partner. If the two cosy up, AMD could end up in millions of Samsung branded technology devices each year.

In fact, Samsung could do well from Fusion/APU technology. At the moment Samsung is relying on its wielding of four Cortex A9 cores at 1.5GHz with proprietary graphics. This limits it to four pixel processors that munch 57 million triangles a second and it is starting to look like it will face a bit of competition in the future.

Samsung also has a fully functional 28nm production process while TSMC is flat out trying to cope with demand. Meanwhile Samsung makes practically all its own components and is able to give its rivals a kicking because it not only sells the notebook but also the hard drive, screen and memory. And jet planes. And ships.

It is a compelling argument, particularly as TSMC’s production costs are likely to rise as it has to spend money appearing to make its Chinese factories look less like sweatshops. 

Another GloFo suit flees the fab

GlobalFoundries has lost its chief financial officer and managed to find another one down the back of the sofa.

According to a company statement, GlobalFoundries has appointed Daniel Durn, a former executive at the firm’s parent company, to the position of chief financial officer and assigned two other senior executives to key positions.

Krakauer, the statement said, had left Globalfoundries to pursue other interests. We guess it is less interesting for a bloke to watch his company sit and haemorrhage cash, its long term relationship with AMD, and see its yield rates falling.

Meanwhile Greg Bartlett, who was GlobalFoundries’ senior vice president of technology and R&D, has been promoted to chief technology officer. KC Ang, formerly vice president of the firm’s 300-mm foundry operations, has been named as the outfit’s head of customer engineering and quality.

He is going to be in charge of GlobalFoundries’ Fab 8 in upstate New York, which is expected to ramp to mass production next year.

It has been a major suit reshuffle at GloFo lately. Last month Ajit Manocha, who had been acting as the company’s interim CEO, took the position on a permanent basis.

Krakauer is the second member of GlobalFoundries’ senior managers to walk away from he job.  

Doug Grose, who had been GlobalFoundries’ CEO since it started, cleaned out his desk and became a consultant to the outfit. We guess his new office was the lift. Grose’s exit was claimed to be triggered by problems with manufacturing yields at the 32-nm node.

According to the annoucement, the GlobalFoundries appointments are part of an effort by the company to “align operations and technology to meet customer needs and market opportunity”.

Translated that means sorting out the mess.

The change is also supposed to separate “technology strategy and its roadmap from technology development.”

Loosely translated, that means that people have not been planning the future correctly and probably confused their arses with their elbows.

“Our customers want us to be the preferred choice for them, so we are taking a hard look at our structure to ensure we are organised to meet their needs,” Manocha said through a statement. In other words “AMD has read us the riot act”. 

TSMC might not be ready for AMD

While pundits are jumping on good rumours which suggest AMD is moving away from GlobalFoundries and towards TSMC, no one has commented whether the outfit is up to the task either.

According to DigiTimes, while TSMC is happy to see orders heat up for advanced 28nm technology, particularly during a a general slowdown in the semiconductor industry, it is not clear if it can manage the extra pressure.

Altera, AMD, Nvidia, Qualcomm and Xilinx have all contracted TSMC to manufacture their 28nm products. Broadcom, LSI Logic and STMicroelectronics reportedly are among potential clients for TSMC’s 28nm technology.

At the moment TSMC cannot manage that sort of pressure. Its wafer output using 28nm processes is projected to top 20,000 units a month by the end of 2011.

In fact it can only consider taking on more work in 2012 when new capacity at Fab 15 comes online. Fab 15 is TSMC’s third 12-inch fab, will begin volume production in the first quarter of 2012, and ultimately raise its monthly capacity to the designed level of 100,000 wafers per month.

DigiTimes says that the new operation has order visibility of more than six months.

While that might be OK, there is a lot that can go wrong. If Fab 15 does not open in time there could be a lot of customers facing a huge backlog.

It is looking like any moves by AMD to dump GloFo to avoid production delays might be a jump from the frying pan to the fire and face more delays. 

TSMC's 40 nano flub to cause GPU price hike

Just when you thought you’d heard quite enough about TSMC’s ongoing 40nm problems, more rumblings and grumblings tip up, forcing us to bring up the whole sorry state of affairs once more.

Despite recent proclamations and declarations by the Taiwanese chipmaker that it had now sorted everything out and that its chambers were no longer helplessly mismatched, Digitimes, that source of all Asian semiconductor knowledge, begs to differ.

Indeed, Digitimes’ sources reckon that “the current shortage of graphics chips, which is being caused largely by low yields of the 40nm process at Taiwan Semiconductor Manufacturing Company (TSMC), is unlikely to be completely solved before May.” Sheeshkebab. 

We wouldn’t mind so much, if TSMC hadn’t been so adamant that its 40nm defect density had “already dropped from 0.3-0.4 per square inch to 0.1-0.3.” That, say Digitimes’ mates in the graphics card making business, is all well and good but “the improvement in overall yield still needs more time before catching up with market demand.”  

And when market demand outpaces market supply, it doesn’t take an economic genius to figure out what AMD, Nvidia and their respective partners are going to do to their 40nm GPU prices. Yep, get your hiking shoes on, folks, because you’re in for a steep climb.

An industry source close to both graphics card makers told TechEye that “the lack of 40nm capacity is absolutely impacting GPU prices. 

“TSMC holds a monopoly on the GPU market until the next generation so the market will suffer through continued constraints in 2010 and early 2011.”

TSMC, however, denied this, giving TechEye the following statement:

“Comparing with our 65nm yield ramping record, 40nm has reached the same level of yield. Our 40nm yields continue to improve at an even faster pace in the last three months.”

Faster than what, though, the firm fails to say.

Meanwhile, Glofo is sitting back and licking its fab chops at the prospect of perhaps landing both Nvidia and AMD in the near future – if Nvidia has the sense not to entrust its livelihood into TSMC’s hands yet again, that is.

 “We plan to aggressively target GPU products at the 28nm technology node and offer increased choice and competition to this high-performance segment of the market,” Glofo’s newest VP told TechEye.

Still, we bet the AMD spin-off is kicking itself that it wasn’t able to cash in on TSMC’s big 40nm flop a little earlier though. But, better later than never, eh?