Electronics Weekly Electronics Design & Components Tech News
“With the 810 (Snapdragon 810) we made a conscious decision to use licensed cores to accommodate the accelerated shift to 64 bit,” says Qualcomm CEO Steve Mollenkopf, “our next premium processor will use our own 64 bit custom CPU architecture as well as the most advanced process node.”
The new processor will be, says Mollenkopf “a device that returns to our internally developed CPU with integrated modem” made on “the latest node”.
As Qualcomm expects to sample the new processor in late H2 2015, then “the latest node” could mean TSMC’s 16nm finfet process – an as yet unproven process.
Qualcomm has lost the processor slot in Samsung’s next premium Galaxy phone the S6. “We now expect that our Snapdragon 810 processor will not be in the upcoming design cycle of a large customer’s flagship device,” says Mollenkopf.
There have been reports of the 810 overheating. A recent research note from J.Pl Morgan said: “For the Snapdragon 810 we believe the issues are related to the implementation of new 64-bit ARM cores (A57) which is causing overheating when accelerating above 1.2-1.4 Ghz.” All that Mollenkopf will say about that is “the device is working the way that we expected it work.”
Mollenkopf acknowledges that Qualcomm has had “some product challenges with one of our chips in meeting some of the more demanding design points of those tiers” referring to high and medium tier products from China’s handset manufacturers.
For the 810, Qualcomm used an SOC design by ARM. For its next Snapdragon Qualcomm will utilise its ARM architecture licence to design its own SOC.
Maybe there’s significance in the fact that Qualcomm’s R&D budget increased by less than 2% last year when its revenues grew 7.3%. Qualcomm has been ahead of the technology curve in mobile for so long, that it seems curious for it not to be increasing its investment in R&D at least as fast as its revenue growth.
When you’re in the lead it’s unwise to take your foot off the accelerator.
Doing so is particularly unwise when You have has $30 billion in cash. It seems crazy to be under-investing in R&D when Qualcomm’s product performance is coming under criticism.
And it’s weirder still when Qualcomm spent $2.4 billion on share buy-backs and dividends in Q4.
Seeing a technology leader like Qualcomm stumble over its product performance comes as a shock, and now it’s also stumbling in the market with a forecast that its revenues will decline between 4% and 10% this year.
The Bank of Nova Scotia reports Qualcomm figures that Qualcomm ICs were in 286 million devices in calendar Q4 2014 versus 278 million in Q4 2013 while ASPs fell to $197.30 from $221.50. Not good.
Meanwhile Qualcomm’s regulatory disputes in China remain unresolved.
This is a different Qualcomm to what we’re used to – a Qualcomm that is losing its way – apparently trading product performance for share price performance.
Ah Yes, thanks Stooriefit, that explains some of it. But there are always bright sparks coming out of the universities and always new ideas coming out of research institutions and you’d want to get the brightest sparks and pursue the best ideas if you were in Qualcomm’s position. A declining growth line for the R&D spend looks a bit like eye-off-the-ball syndrome.
When you get big, the amount you have to spend to grow more than organically becomes disproportionately large, unless you are prepared to drop your quality threshold for recruitment.
It gets expensive because:
if you recruit you have to head-hunt people and then up the compensation of the equivalent staff you already have who are a flight risk.
if you grow R&D by acquisition the owners of your target see the size of your wad and suddenly you have the Fulham Antique Shop effect – they saw you coming.
Yes absolutely SEPAM the iPhone 5 caught everyone by surprise and I expect Qualcomm went into over-drive to get a 64-bit chip and opted for an ARM-designed SOC. The only other reason for using an ARM-designed SOC would be to save development money but that seems crazy when you have $30bn cash. On the other hsnd this paltry rise in R&D suggests a worryingly complacent attitude inside Qualcomm. They used to blitz this industry with their R&D spend and now they seem to prefer to spend their money buying their shares. It’s incomprehensible to me.
There has been a lot of strange stories about Qualcomm processors lately, especially the alleged Taipan cores. Some say a new set is on its way and others say the leak is fake.
The overheating story is questionable since it should, if true, have been reported in their financial statement but this was not done.
It does seem like they were caught by surprise when Apple launched a 64-bit ARM processor and the latest Qualcomm 64-bit CPU is most likely a stop gap to enter the 64-bit arena. Most likely they are amongst the ARM architecture licencees and a 64-bit “son of Krait” is most likely in the works.
I must admit I am surprised to see a low growth in R&D. A quick check for patent statistics shows a continuous growth in international patent applications filed with nearly 2500 filed last year.