Tuesday, January 19th 2016
AMD Reports 2015 Fourth Quarter and Annual Results
AMD (NASDAQ: AMD) today announced revenue for the fourth quarter of 2015 of $958 million, operating loss of $49 million and net loss of $102 million, or $0.13 per share. Non-GAAP(1) operating loss was $39 million, non-GAAP net loss was $79 million and non-GAAP(1) loss per share was $0.10.
"AMD closed 2015 with solid execution fueled by the second straight quarter of double-digit percentage revenue growth in our Computing and Graphics segment and record annual semi-custom unit shipments," said Dr. Lisa Su, AMD president and CEO. "While 2015 was challenging from a financial perspective, key R&D investments and a sharpened focus on innovation position us well to deliver great products, improved financial results and share gains in 2016."
2015 Annual Results
"AMD closed 2015 with solid execution fueled by the second straight quarter of double-digit percentage revenue growth in our Computing and Graphics segment and record annual semi-custom unit shipments," said Dr. Lisa Su, AMD president and CEO. "While 2015 was challenging from a financial perspective, key R&D investments and a sharpened focus on innovation position us well to deliver great products, improved financial results and share gains in 2016."
- Revenue of $3.99 billion, down 28 percent year-over-year, primarily due to lower client processor sales.
- Gross margin of 27 percent, down 6 percentage points year-over-year and non-GAAP gross margin of 28 percent, down 7 percentage points year-over-year. The year-over-year declines were primarily due to lower unit volumes and product mix.
- Operating loss of $481 million and non-GAAP operating loss of $253 million, compared to a loss of $155 million and non-GAAP(1) operating income of $316 million in 2014 primarily due to lower revenue and gross margin.
- Net loss of $660 million, loss per share of $0.84, and non-GAAP net loss of $419 million, non-GAAP loss per share of $0.54, compared to a net loss of $403 million, loss per share of $0.53, and non-GAAP(1) net income of $132 million, non-GAAP earnings per share of $0.16 in 2014.
- Revenue of $958 million, down 10 percent sequentially primarily driven by seasonally lower sales of semi-custom SoCs and down 23 percent year-over-year, primarily due to lower client processor sales.
- Gross margin of 30 percent, up 7 percentage points sequentially. Q3 2015 gross margin was negatively impacted by an inventory write-down of $65 million, or 6 percentage points. Excluding the Q3 2015 inventory write-down, non-GAAP gross margin improved 1 percentage point sequentially, primarily due to improved product mix in the Computing and Graphics segment.
- Operating loss of $49 million, compared to an operating loss of $158 million for the prior quarter. Non-GAAP(1) operating loss of $39 million, compared to non-GAAP(1) operating loss of $97 million in Q3 2015, primarily due to higher gross margin and lower operating expenses.
- Net loss of $102 million, loss per share of $0.13, and non-GAAP(1) net loss of $79 million, non-GAAP(1) loss per share of $0.10, compared to a net loss of $197 million, loss per share of $0.25 and non-GAAP(1) net loss of $136 million, non-GAAP(1) loss per share of $0.17 in Q3 2015.
- Cash and cash equivalents were $785 million at the end of the quarter, up $30 million from the end of the prior quarter, primarily due to improved operating cash flow.
- Total debt at the end of the quarter was $2.26 billion, flat from the end of the prior quarter.
- Computing and Graphics segment revenue of $470 million, an increase of 11 percent sequentially and a decrease of 29 percent from Q4 2014. The sequential increase was primarily due to higher notebook processor sales, and the year-over-year decrease was primarily driven by lower client processor sales.
o Operating loss was $99 million, compared to an operating loss of $181 million in Q3 2015 and an operating loss of $56 million in Q4 2014. The sequential improvement was driven primarily by higher sales and the absence of a Q3 2015 inventory write-down and the year-over-year decrease was primarily driven by lower sales.
o Client average selling price (ASP) increased sequentially driven by a richer notebook processor product mix and decreased year-over-year due to a lower notebook processor ASP.
o GPU ASP increased sequentially and year-over-year primarily due to a higher AIB channel ASP. - Enterprise, Embedded and Semi-Custom segment revenue of $488 million, a decrease of 23 percent sequentially primarily driven by seasonally lower sales of semi-custom SoCs. Revenue decreased 15 percent from Q4 2014 primarily driven by lower game console royalties, and server and embedded revenue.
- Operating income was $59 million compared with $84 million in Q3 2015 and $109 million in Q4 2014. The sequential decrease was primarily due to seasonally lower sales of semi-custom SoCs. The year-over-year decrease was primarily due to lower game console royalties, and server and embedded sales.
- All Other operating loss was $9 million compared with operating losses of $61 million in Q3 2015 and operating loss of $383 million in Q4 2014. The sequential improvement was primarily due to Q3 2015 restructuring and other special charges and the year-over-year improvement was primarily due to the absence of a goodwill impairment charge, lower restructuring and other special charges, net and a Q4 2014 lower of cost or market inventory adjustment.
- AMD provided a glimpse at its next-generation GPU architecture and delivered innovative new graphics, embedded, and desktop component technologies.
o AMD previewed its revolutionary 14nm FinFET Polaris GPU Architecture, highlighting significant architectural improvements including High Dynamic Range (HDR) monitor support and a 2x performance-per-watt improvement over the prior generation. The GPUs deliver a remarkable generational jump in power efficiency, and are designed for fluid frame rates in graphics, gaming, VR, and multimedia applications on small form-factor thin and light computer designs.
o AMD released its re-architected graphics software suite, Radeon Software Crimson Edition, giving users 12 new or enhanced features, up to 20 percent more graphics performance2, adjustability that can nearly double generational energy efficiency3, and stability across the full spectrum of AMD graphics products.
o AMD introduced the AMD Radeon R9 380X GPU, conceived to play the most detailed and demanding games at 1080p and 1440p. The GPU offers a 256-bit interface and 4GB of high-performance GDDR5 memory and features including compatibility for both AMD FreeSync and AMD LiquidVR technologies plus Virtual Super Resolution.
o AMD announced the new AMD FirePro W4300 graphics card, its highest performing professional graphics card optimized for Computer-Aided Design (CAD) that fits in both small and full-size workstations, offering unprecedented flexibility in its class.
o AMD achieved high-end embedded performance leadership with the introduction of the AMD Embedded R-Series SOC processors designed for digital signage, retail signage, medical imaging, electronic gaming, media storage, and communications and networking.
o AMD announced the AMD FX 6330 CPU for the China market with a new, near-silent stock cooler and offering excellent 6-core performance, control, and reliability for productivity, entertainment, and multi-tasking workloads. - AMD launched its first 64-bit ARM based product -- the AMD Opteron A1100 SoC - designed to accelerate time-to-market deployment of ARM-based systems for the datacenter and improve enterprise-class ecosystem support for 64-bit ARM in key markets. AMD is working with technology partners and customers including Red Hat, Silver Lining Solutions, SoftIron, and SUSE on AMD Opteron A1100 SoC-based hardware and software solutions that provide high-speed network and storage connectivity, energy efficiency, and a balanced total cost of ownership for storage, web, and networking workloads.
- AMD collaborated with industry leaders to bring powerful new embedded, professional graphics, and gaming solutions to market.
o AMD further solidified its No. 1 position in the thin client space with the introduction of the new AMD Embedded R-Series and AMD FirePro-based HP t730, the world's first thin client with native quad UHD/4K support.
o AMD announced several new AMD FirePro professional graphics design wins with Dell, including the new Dell Precision 3510, 7510, and 7710 mobile workstations delivering exceptional graphics performance and GPU compute capability. In particular the Dell Precision 7710 features nearly 3 TFLOPS of single-precision GPU compute power for GPU-accelerated applications and workflows.
o AMD expanded its leadership position in virtual reality (VR), announcing a collaboration with Oculus and Dell to equip Oculus Ready PCs with AMD Radeon GPUs.
o Lenovo introduced the AMD FX CPU and Radeon R9 graphics-based Lenovo Y700, the first notebook validated to support AMD FreeSync technology. - AMD provided developers with new tools designed to simplify software development and more fully harness the capabilities of its GPUs.
o AMD launched the "Boltzmann Initiative", a suite of tools designed to dramatically simplify GPU computing on AMD FirePro Graphics by leveraging Heterogeneous Systems Architecture's (HSA's) ability to harness both CPU and GPU for maximum compute efficiency through software.
o AMD announced the GPUOpen initiative to help address the evolving demands of graphics and unlock game and application development through open source software. The initiative enables game developers to better harness the investments they've made on console development, introduces a new compiler for heterogeneous computing, and demonstrates AMD's renewed commitment to Linux with its Linux Open Source Strategy.
77 Comments on AMD Reports 2015 Fourth Quarter and Annual Results
The impact of not being able to adequately profit from having superior products back then is just too big.
"oh, they failed with bulozer" - yeah. With their tiny R&D budget.That somehow balances situation when Compaq didn't dare to get AMD chips for free, eh?
AMD would get bigger market share, bigger revenue and higher margins, if not Intel's misuse of dominant market position (for which it was convicted in EU), are you seriously arguing with that?
And who cares what happened back in the very early beginning, when Intel was forced by IBM to, cough, do certain things.
Their only option is to come with something way too revolutionary, but then the old Intel tactics used to crush the K7 will for sure resurface.
As this looks like a fake competition, a solution would be to force Intel to provide x86 licenses to third parties. That's it, nobody needs inside technology, nobody needs implementation details, just the right to build a processor on this technology, and I'm pretty sure most big design houses will be able to come with something strong on the market. There are smart people everywhere.
My post wasn't about mitigating what Intel has done - I've already written as much myself, it is about putting the past into perspective - not the selective memory version where AMD were some put upon angelic entity. It is historical fact that AMD could not supply the vendors it already had on board from around 1997/98 to the mid-2000's so the range of vendors they could keep happy was limited. The post I linked to had a couple of links to 2005 shortages, but if you'd like a more comprehensive listing it's no problem - how about2003? AMD sounding out TSMC in 2001 for a JV ? AMD had supply issues almost continuously with every release (K6-II and K6-III are well documented across the board shortages). If AMD was using all its resources I (and virtually every tech historian) could understand and sympathize - but the simple truth is Jerry Sanders and his puppet Ruiz steadfastly refused to second source production even though the cross-license with Intel allowed AMD to outsource up to 20% of their production to third-party foundries. Why? Because Sanders was an old school egomaniac that believed if you weren't totally vertically integrated you had no right to be a semiconductor company - you don't sell anything you don't fab yourself as AMD ex-President and COO Atiq Raza noted in an Ars Technica article:Apologists lay the fault of AMD's predicament entirely at Intel's door, and it worked so well that the board of directors were free to run the company into the ground because blame for any bad decision could be routed through the excuse of a lack of R&D because Intel wouldn't let them sell chips. It wasn't until the old guard left and started penning their memoirs that the extent of the damage from "real men have fabs" became apparent.
Intel have been guilty of some truly heinous acts, not just against AMD but Chips & Technologies, Intergraph, Cyrix, Seeq, and their own employees among others, but to dismiss AMD's own actions in their own misdeeds and decline is at the very least short-sighted.
Companies exist to make money for investors. You just can't keep everyone happy doing that.
Make shit products: people hate you. Make good products: people still hate you. Might as well go all evil now because no matter what a successful company of that size does, it is going to make someone very unhappy.
Is your grudge more against capitalism and today's object/consumer goods centric world or what? 6700K like you said, is faster than anything competition can throw at it. It also is faster than it's previous model. And you are upset that you didn't get THE best CPU you can get for same money?
Welcome to reality, sir. Life is not a fair place.
I have been f'cked by both of these companies for as long as I can remember. Please don't take my direct language as offence, I understand your point. It is just that pretty much all of these companies are doing it. AMD included, so I see no reason to love or hate either one more than the other. :)
Doubling PCI-E bandwidth might need more/faster cache memory, wider internal buses and beefier support IPs to take full use of it. In the end, it could very well look more like Haswell-E. All that for what, couple percent performance increase with SLi/Crossfire?
You can buy that product if you really need one tho, so it is not like it is not made available.
That being said they make some bad ass mini-itx rigs.
Alright, enough offtopic. I have hoped AMD get better for a very long time but still has not happened. Hopefully next year we at least have AMD to talk about. :)
1: More PCI-E means more contacts for the CPU, you need the connections for every lane and slot you're going to have.
2: The PCI-E root complex would have to be twice as big to go from 16(20) to 40.
Example:
2600, 2600k, and 2700k have a die size of 216mm^2.
The 3820 has a die size of 294mm^2.
For doubling the PCI-E lanes, doubling the width of the memory controller, and adding 2MB more of L3 contributed to a die that was 36% bigger in relation to the 2600k. On top of that, the iGPU on the 2600s and 2700k takes up considerable die space, almost the same amount as the L3 cache or almost 2 cores, something the 3820 does not have but yet, still uses that space despite being a quad core like the others.
Sorry @cdawall but, I wouldn't call that peanuts. Your claim is spurious.
Simply put, I/O adds a whole lot more to the die than you think it does, even more since you're essentially doubling it.
Side note: The 3820 is not the same die as the 6c SB-E chips and doesn't have laser cut cores, so that's not an argument either.
AMD can not live life smoke.
But overall, it was an interesting read with good insights, thanks.
Regardless, Intel was convicted for abusing it's dominant market position. That happened MAINLY vs AMD.
It doesn't make AMD an angel, not at all, although, an underdog company is pretty limited in ways it could abuse its market position.
Where it matters to me, they are normally on the right side:
IA-64 vs AMD64
GSync vs FreeSync
Etc.Yes, although not as short sighted as turning blind eye to Intel's shitty practices and putting nearly exclusive blame on AMDs management. (maybe you don't mean it, but that's impression I get from reading your posts)
What is actually behind your stance, do you have friends working @ Intel?
My stance is regarding history as a whole. People here seem to be reasonably well aware that I write historical-based tech articles. Obviously you don't, or you would see that I treat the dealings of the companies with equanimity - unlike some here hell bent on highlighting the bad behavior of some companies while completely glossing over the bad behaviour of their favoured brand.
Here are some examples from an article on the microprocessor and its effects on personal computing I wrote a little over a year ago- and I might add they have a readership orders of magnitude above what is posted in these threads:And from a history of AMD article I wrote back in late 2012 ( well before some other sites churned out their own articles on the subject):I've lost count of the number I times I've commented on Intel's misdeeds. Waging economic warfare against AMD, Seeq, NEC ( the V20/V40 case), Intergraph, Chips and Technologies, crushing Cyrix (with IBM's help), the long running battle with DEC (both sides guilty) etc., the strong-arming and intimidation of their own workforce (and the ostracization of Bob Graham the 3rd employee at Intel because Grove wanted all the marbles) etc... But the one thing that comes across is that nobody overlooks Intel's practices. The same is definitely not true of AMD. The common refrain is that is always someone else's fault that they are in predicament they are in. I would almost guarantee that you probably had no idea that AMD faked benchmarks for Barcelona until I linked to a story. I'm also pretty sure you don't have much idea about their earlier suspect marketing under Jerry Sanders and Hector Ruiz. I'm also pretty sure that you will never bring up the subject again just as many before you have buried selective parts of semiconductor history when it doesn't fit the narrative required.
Anyhow, feel free to demonize me because I take a measured view of the industry (Thirty-five years being around semiconductor topics will do that) rather than play cheerleader and reduce everything down to a simplistic good versus evil struggle.
I'll await your evidence that AMD offered free processors to Compaq. If this is indeed true then I am missing a part of semiconductor history - along with every contemporary writer and publication I've ever read.
europa.eu/rapid/press-release_IP-09-745_en.htm?locale=en
For example, rival chip manufacturer AMD offered one million free CPUs to one particular computer manufacturer. If the computer manufacturer had accepted all of these, it would have lost Intel's rebate on its many millions of remaining CPU purchases, and would have been worse off overall simply for having accepted this highly competitive offer. In the end, the computer manufacturer took only 160,000 CPUs for free.
It was compaq.
PS
Will respond to the rest of your post a bit later.