Intel rebuffs claims that 10nm chip fabrication process has been abandoned
A report claims Intel is giving up on its 10-nanometer process for chip production after enduring numerous delays and yield issues over the last few years, with Intel already declaring it as "untrue" and that it is continuing to make "good progress" on the process.
According to sources of a report within Intel, the company has effectively killed off its struggling 10-nanometer production process. Very few details are offered about the supposed termination of the node's development, with the only information provided by company moles is "the process is indeed dead."
The report by SemiAccurate has been refuted by Intel, with an official tweet calling media reports on the 10-nanometer process's demise as false. "We are making good progress on 10nm," the tweet states, with yields said to be improving at a rate declared by the chip producer during its last earnings report.
The rumor is likely to be either firmly confirmed or denied during Intel's financial results occurring on Thursday.
SemiAccurate claims the termination of the current 10-nanometer process would be good for Intel, calling it "the first adult decision we have seen from the company in several years." In reasoning this, it points to timetables in the Q1 2018 earnings being shifted around, and references to "hyperscaling" from early 2017 that were supposedly an excuse to for a timetable slide from that time.
A sudden shift of the product timeline to give up on its existing work on the process is seemingly unlikely for Intel to make, considering the amount of effort and resources that have been consumed in its development so far. It would also be a major move that would be unlikely to be performed under the watch of interim CEO Bob Swan, rather than a fully fledged CEO.
The decision would also go against what was brought up during a "Supply Update" by Swan at the end of September, where he mentioned the company was "making progress" with yields improving and volume production still anticipated for sometime in 2019. "Cannon Lake," the generation meant to include chips using the 10-nanometer process, has so far consisted of just one processor made with the process.
While Intel has endured repeated delays in 10-nanometer chip production, one that was supposed to have been introduced in 2016, the firm is now working to keep up with demand for its processors made using a 14-nanometer process. Along with increasing investment in chip manufacturing sites around the world, Intel is also reportedly outsourcing some of its 14-nanometer production to rival TSMC, including its H310 and other 300-series chipsets.
Cannon Lake is also expected to provide a number of benefits to MacBook Pro users in the future, if Apple elects to use the processors, due to their support for LPDDR4 memory. The 15-inch MacBook Pro refreshed in July includes a 32GB RAM option, but only after Apple redesigned the system to use DDR4 memory, but a processor supporting LPDDR4 would allow for the use of the less power-hungry memory, increasing battery life.
According to sources of a report within Intel, the company has effectively killed off its struggling 10-nanometer production process. Very few details are offered about the supposed termination of the node's development, with the only information provided by company moles is "the process is indeed dead."
The report by SemiAccurate has been refuted by Intel, with an official tweet calling media reports on the 10-nanometer process's demise as false. "We are making good progress on 10nm," the tweet states, with yields said to be improving at a rate declared by the chip producer during its last earnings report.
The rumor is likely to be either firmly confirmed or denied during Intel's financial results occurring on Thursday.
Media reports published today that Intel is ending work on the 10nm process are untrue. We are making good progress on 10nm. Yields are improving consistent with the timeline we shared during our last earnings report.
-- Intel News (@intelnews)
SemiAccurate claims the termination of the current 10-nanometer process would be good for Intel, calling it "the first adult decision we have seen from the company in several years." In reasoning this, it points to timetables in the Q1 2018 earnings being shifted around, and references to "hyperscaling" from early 2017 that were supposedly an excuse to for a timetable slide from that time.
A sudden shift of the product timeline to give up on its existing work on the process is seemingly unlikely for Intel to make, considering the amount of effort and resources that have been consumed in its development so far. It would also be a major move that would be unlikely to be performed under the watch of interim CEO Bob Swan, rather than a fully fledged CEO.
The decision would also go against what was brought up during a "Supply Update" by Swan at the end of September, where he mentioned the company was "making progress" with yields improving and volume production still anticipated for sometime in 2019. "Cannon Lake," the generation meant to include chips using the 10-nanometer process, has so far consisted of just one processor made with the process.
While Intel has endured repeated delays in 10-nanometer chip production, one that was supposed to have been introduced in 2016, the firm is now working to keep up with demand for its processors made using a 14-nanometer process. Along with increasing investment in chip manufacturing sites around the world, Intel is also reportedly outsourcing some of its 14-nanometer production to rival TSMC, including its H310 and other 300-series chipsets.
Cannon Lake is also expected to provide a number of benefits to MacBook Pro users in the future, if Apple elects to use the processors, due to their support for LPDDR4 memory. The 15-inch MacBook Pro refreshed in July includes a 32GB RAM option, but only after Apple redesigned the system to use DDR4 memory, but a processor supporting LPDDR4 would allow for the use of the less power-hungry memory, increasing battery life.
Comments
but it needs to be said that Intel is still in the lead in most areas. It’s well known that earlier this year, Intel proposed that companies use, instead of the current node number, 14nm, 10nm, 7nm, etc., transistor density as a measurement. It’s also well known that the idea was rebuffed by their competitors. Why was this proposed? It was because Intel’s 14++nm node is just as efficient in density, power reduction and performance as other 10nm nodes, and possibly a bit better. Competitors know this very well, and are uncomfortable with the idea of giving up their apparent advantage in node numbers.
we can also be sure that Intel is being much more careful with their 10nm node to continue this efficiency and density advantage, with their 10nm node being at least as good as anyone else’s 7nm node, possibly a bit better. The only competitor Intel has in x86, where their main source of income and profits lie, is and. While it seems as though and has some advantages right now, those advantages are shrinking with Intel’s latest products. Their price advantage is simply because at the same price as Intel, they can’t sell chips, so they accept much less profit, and even accept losses in order to continue in business. They are just now going to 7nm with TSMC for a couple of new products.
apple and others are not yet competing with Intel with their ARM products, though maybe that could change in a minor way in a year or two. Until then, Intel is in the driver’s seat. If they get commercial levels of product out of their 10nm node at the end of the 1st half of next year, they will be in great shape, but people don’t seem to want to talk about that. It’s more fun putting them down.
Still, this does not necessarily mean that they are failing overall in process---perhaps they just made poor decision on which process flow to implement. It happens....
What they are failing in is decision making ability and implementing, whatever it is. That says that current process management stinks.
and Swan is only temporary and shouldn't make a decision? If that is true, why do they have him, and what good is he. I think that he should review the situation and make a definitive decision---what it should be of course I have no idea, but they have been diddling around for years.
The diddling is a result of current process management, brilliant or not, not knowing how to manage development. They desperately need help and a new view; not a reorg using previous failures (failures in this case at least)
Sell all your Intel stock while it’s still worth something. Don’t end up owing Kodak or Blackberry till the very end.
Oh hy he way switching to transistor density does solve any issues with respect to comparing process densities. This especially with SoC which makes x various by blocks of high density logic with random logic. A large cache for instance would increase the density figure while random logic would lower it.
Apples A12 comes in at almost 84 million transistor per square mm. If it ever ships the estimate is that Intel will be able to do 100 million on its 10 no process. That is a significant boost over its current tech by 2.7 times. So given that there are truths in these numbers Apple is shipping a higher density processor in volume compared to Intel. Not by a little bit either. But again density as a measurement can be misleading due to the averaging effect. What is even more disturbing is that TSMC is on the path to a vastly improved 7nm process in 2019. If AMD can get product out using TSMC tech early in 2018 Intel is screwed.