It is not a secret that transitions to newer chip manufacturing processes are getting more difficult from financial capital standpoint since not only research and development costs more than before, but also production equipment is getting more expensive. In a bid to keep their prices at current levels, semiconductor companies need to increase transistor densities of their process technologies at a pace faster than Moore’s law. Intel believes that it can do this at 7nm and 10nm.
That capital cost is a headache for all makers of chips because if it is too high, they either need to lower their margins or increase their prices. To offset increases of capital costs per square millimeter of a chip, makers of semiconductors need to increase transistor densitires, which is not an easy task due to leakages and other effects. However, Intel claims that it has found ways to increase those densities going forward, which will make its 7nm and 10nm process technologies economically feasible.
“We believe that through 14nm, 10nm and with some insight all the way down the 7nm we can offset that increase in capital cost per square inch of silicon by improving our density,” said Mr. Smith. “So we can keep the cost per transistor coming down at the historical curve.”
KitGuru Says: Keeping in mind that Intel controls architecture, design, technology and manufacturing aspects of its business, the company has a lot of flexibility. In general, Intel can make almost any reasonably good manufacturing process economically feasible for itself.
[ARM's latest nodes, on the other hand are not economically feasible being too expensive and suffering from bad yields and limited supplies. Only a few companies, like Apple, can afford the costs and now Apple doesn't know which way to go (Samsung or TSMC) because both are struggling mightily with the 16nm node. The path forward for Intel is clear. Sooner or later even Apple has to accept that ARM is having a fabrication crisis.
"NIce, already up 4% in one day."
[Hahahahahahahaha. Yeah, Lucy - you and AMD belong together. Is this another of your trades that you will claim to be out of as soon as it turns on you?]
Winning a supply contract with the smartphone industry heavyweight would bring some relief to TSMC, which has endured pain from its exposure to makers of lower end phones. Inventory destocking by these lower end smartphone makers will lead to lower utilization rates for TSMC’s 28nm and 20nm chip production facilities. This could lead to a weaker-than-usual performance by the company in the second quarter, argues JPMorgan analyst Gokul Hariharan.
Although you would not think it possible, AMD is still having issues that stem back from their purchase of ATi all those years ago. As most of you are sick of hearing the details of that acquisition and the gradual fall out we will not bore you here. What we will talk about are some of the issues that AMD now faces and what they mean for the consumer and AMD themselves.
Over the last year AMD has started pulling back on launching new products in both the CPU and GPU world. They have been a little more active in the APU world, but in many cases the “new” products do not have enough improvement to make them really beneficial to manufacturers. Because of this demand for AMD products has dropped off. As this demand drops the new release cycle is also impacted. Why would you make millions of parts if there is no one that is buying them? If you cannot sell inventory then you make less at the end of the quarter… you get the picture.
We have seen indications of this decline including the lack of a real GPU release for 2014-2015. AMD allowed NVIDIA to widen the performance gap in addition to simply having something new to offer their customers (manufacturers and end users alike). Now we are hearing that AMD is reducing the number of processor orders with Global Foundries without an increase to another foundry. In the simplest terms is means that AMD does not have the demand to meet the cost of producing the processors they typically order. The fact that the decrease (of about $200 Million) appears to be across the board to Global Foundries.
We are not going to say that the bell it tolling on AMD, but they cannot continue to operate like this.
AMD is learning the hard way about not having any significant product launches as it released figures which show it slipping deeper into the red again.
Worse than expected
TSMC has released Q2 sales guidance, missing the mark and disappointing many analysts.
The company said it expects a 7-8% quarterly drop in sales this quarter, which is quite a bit worse than the 5% decline analysts were expected.
Analysts believe the reduction was caused by a cutback in Qualcomm SoC orders. Samsung’s decision to use in-house 14nm parts for its flagship phones is obviously a factor, too.
TSMC previously made it clear that it expects foundry growth to slow down this year, due to the macroeconomic situation.
TSMC’s 28nm blues
While the company claims its 16nm node is on track, and coming in under budget, the short term problem for TSMC is actually its legacy 28nm node. The company reportedly lost quite a few orders to SMIC and UMC, including orders from industry heavyweights like Qualcomm and MediaTek.
We discussed why 28nm will remain a relevant node for several quarters to come in a recent feature. It will be used for low- to mid-range parts for at least another year.
But hey, at least he has a new fistful of imitation IDs.
What's poor Apple to do - what with ARM's fabrication falling apart. Samsung has low yields on its 16nm node and TSMC hasn't yet gotten 16nm in production.
Maybe Apple started believing all of those press releases from TSMC and Samsung saying that they had drawn even with Intel in the high-end fabrication race. They have except for the minor issue of being able to get volume production out the door with acceptable yields. And that's no going to happen for a very long time.
Bottom line is that the high-end is going to be in limited supply and it's going to be very, very expensive.
Given this situation, you think Apple is not going to talk to the only company who doesn't have any of these problems? Apple is already moving in Intel's direction on the modems for next year. If you can't see where this is going then you should maybe remove the bag over your head.
As a direct result of tax cuts for the rich by Republican president after Republican president. The rich will never stop trying to bleed the middle class so they can have yet more and more.
"Yes, regulations are killing us. But good thing is only about a year and 8 months maybe we will have a president."
[We'll always have a president. The question is whether we'll have one that only represents the wealthy and the dysfunctionally religious extremes or whether we'll have one that represents everyone including the poor, the middle class, the minorities, the infirm, the elderly and the very young. ]
For continuous, mindless, meaningless monitoring of the Intel ticker.
Get a life oh brainless one - Intel is still two bucks ahead of where you said it would be now. Put that in your ticket and pull on it.
[Misery Loving Lucy is sad.]
Consumer confidence improved in April to the second-highest level in more than eight years as Americans held more favorable views of the economic outlook and inflation.
The University of Michigan said Friday that its preliminary index of sentiment climbed to 95.9 this month from 93 in March. The median projection in a Bloomberg survey of economists called for an increase to 94.
Consumers were more upbeat about the economy, their current financial situations and the buying climate, indicating spending will pick up after a lackluster performance in the first quarter. At the same time, households have a dimmer view about their income prospects, meaning shoppers will keep seeking out discounts from merchants.
“We’ve seen signs that other economic activity has bounced since the weather’s gotten better, so that should feed through to consumers’ views on the outlook,” Thomas Simons, a money market economist at Jefferies LLC in New York, said before the report. “Also, a big component is gasoline prices, which stopped going up and in some cases came down a little bit again. All of that is helpful.”
PORTLAND, Ore. — Intel's first Xeon system-on-chip (SoC) is twisting ARM's microserver ambitions, Charles King, principle analyst at Pundit-IT in Hayward, California told EE Times.
"Its not just Intel's first SoC in the Xeon family," King told us. "Its the beginning of a new era at Intel — expect them to move very fast in SoCs. We are going to see many more SoCs specifically designed to combat ARM microservers plus serve many datacenter functions."
The Xeon D-1500 was actually announced last year, but Intel has taken its time to make sure its ARM microserver killer was optimized for cloud, telco service providers and web hosters. Cast in Intel's 14nm 3D FinFET process, the Xeon D family will bring down the price of microserver SoC's to under $200.
ARM's high-end fabrication continues in crisis mode. The problems are threefold:
1.) Too expensive
2.) Yields are bad
3.) Intel is too far in front and moving steadily ahead.
In spite of the marketing efforts coming out of TSMC and Samsung, the ARM world hasn't closed the technology gap with Intel and they are not close to doing so. There is only one company maintaining Moore's Law and doing so with solid economics and that's Intel. It's all about yields and costs and capacity. ARM doesn't have it and won't for years and Intel isn't going to wait on them.
Yes, Intel had some yield problems which have now been substantially solved. And now we get to see the ARM world derailed by these same problems. Closing the technology, manufacturing and economics gap with Intel will take far more than wishful thinking and continuous press releases from an overly optimistic marketing department.
ARM is facing a rapidly maturing mobility market with shrinking ASPs, margins and profits. RT has died a painful death and ARM has made no progress in cracking the server market. The WinTel alliance has been resurrected. And here comes Windows 10. Intel also has a significant advantage in memory markets and technology. And a rapidly expanding reach into new datacenter solutions.
If the Altera deal goes, then Intel will have a substantial FPGA/SOC capability that the ARM world won't be able to match.
Intel continues to consolidate while ARM's fabrication continues to cough, hiccup, sputter and coast much of the time. A while back some tech peeps said that the ARM world was going to be at 28nm for a very long time. They were spot on.
everal months ago, rumors began spreading that Apple had switched foundry partners (again) and would rely on Samsung, not TSMC, for its next iPhone and iPad SoC (logically, though unofficially, dubbed the A9). Now, an analyst is reporting that Apple, in a last-second move, has yanked a significant percentage of its orders back from Samsung due to low yields at Samsung’s partner, GlobalFoundries.
Rumors like this are fairly common, but the analyst in question, Ming-Chi Kuo, has a solid reputation for correctly predicting Apple launches and activities. According to the analyst, GlobalFoundries, which originally agreed to be a contract partner and second source fabrication facility for Samsung, wasn’t able to hit Apple’s yield targets. Initial yields at GF were at 30%, while Apple demanded 50% yield to hit certain internal requirements. Globalfoundries is serving as Samsung’s second source at 14nm, after cancelling its own 14nm-XM process.
There are several ways to explain this discrepancy. The simplest answer is that Apple chose to hedge its bets and began work on a TSMC-built version of the A9 alongside the Samsung variant. The TSMC flavor could also be used to plug longer-term demand. If Apple believes it can stockpile enough chips for the first few months of sales, but sees a capacity crunch coming near the end of the year as other companies start rolling out their own solutions, then having a TSMC part ready to produce makes good sense.
The dual-supplier strategy is common. But it hasn’t always worked for Apple, which ran into trouble with the first Retina-display MacBook Pros. At the time, Apple sourced its displays from both LG and Samsung — but consumers often found that the LG displays had persistent problems with long image retention times. If the TSMC chips and Samsung/GF chips don’t perform equivalently to each other, Apple could inadvertently create a situation where some iPhones are noticeably better than others.
"9 months ago today INTC closed at 34.65. So it is down $1.78 or about 5% since then
9 months ago today ARMH closed at 43.70. So it is up $7.42 or about 17% since then"
[Hahahahahahahaha. Lucy always finds a lame statistical comparison to try to look less like a pinhead. But fails to mention being wrong by 3 bucks recently about Intel and fails to mention that ARM lost 74 cents today while Intel was in the green.
You're a pinhead loser, Lucy. Everyone knows it so stop with the spin.
You blew the earnings call, Lucy. Accept your shame and move on.]
1.) ARM down.
2.) Qualcomm down.
3.) TSM down.
4.) AMD really down.
5.) Intel up.
Surprised those fact loving pinheads didn't mention any of THESE facts.
Lots of facts came out of earnings. Pinheads seem to have missed all of them too.
How could so many have gotten it so wrong about Intel's earnings?
Well, because all they know about it is what is happening with the PC market. As we found out (some, of course, already knew) is that it's more complex than that.