Understanding AMD's Semi-Custom Strategy
by Anand Lal Shimpi on September 5, 2013 5:51 PM EST
With Microsoft’s acquisition of Nokia we are now living in a world where all three major client OS providers (Apple, Google and MS) all own/are device companies. Samsung gets an honorary seat at the table by virtue of making devices as well as most of the components inside those devices (and fabbing the ones they don’t make), on top of adding a fair bit of software customization of their own. That makes four companies with the ability to control most of the market, what happens to everyone else?
Both Google and Microsoft have stated publicly that acquiring device companies doesn’t mean that they want to stop working with outside partners. Naturally that’s what you’d expect Google and Microsoft to say to avoid overly angering existing partners shipping existing devices. In the long run however, companies like Acer, ASUS, Dell, HP, HTC, Lenovo and LG (among others) will find themselves in an interesting position, competing against their OS suppliers that also happen to be vertically integrated device makers.
I don’t expect that we’ll immediately transition to a world where there are only four brands of smartphones, tablets and notebooks. I suspect that all of the existing players, vertically integrated or not, will do their best to maintain/grow marketshare despite the threat of real consolidation. After all, a threat alone isn’t enough to force companies out of a market.
When faced with a very large competitor, you need a broad set of partners. It’s all of these other companies that AMD views as potential candidates (including Samsung it seems) for its semi-custom approach to silicon manufacturing.
Last year AMD’s new CEO Rory Read talked about delivering semi-custom silicon, similar to what AMD does in game consoles, to OEMs who want to differentiate. At the time I didn’t really get how this strategy would work but I think I now have better insight into the theory. If we assume that Apple and Samsung will remain major players in the consumer device space, both of these companies have heavily invested in developing their own software and SoCs. If you’re an OEM looking to compete with either (or both), you need the same assets. As a company that isn’t Apple or Samsung however, you don’t have the resources to truly integrate vertically and become a silicon manufacturer as well as run a profitable business. AMD hopes that there will be at least a couple of enlightened players in the device space who realize they need a silicon strategy to differentiate but don’t have the resources/expertise/desire to become a silicon player.
The idea here is that AMD would then provide those OEMs with a semi-custom SoC, where they could choose their own IP blocks (video decode/encode, CPU cores, GPU, etc...), and help give them competitive parity or ideally even an advantage over the larger players. Counting on the bigger guys having higher overhead, being less agile and having to make larger profits should, at least on paper, give the smaller players a fighting chance.
It took me a while to understand AMD’s semi-custom strategy, but seeing how things have shaken up over the past few months I think I can understand it a lot better than what it was first floated at AMD’s Financial Analyst Day last year.
The other thing I think I understand a lot better now is why former Lenovo President & COO, Rory Read, was chosen as the man to run AMD. I’ve met most of the new guard at AMD and generally came away quite impressed. The challenge set in front of them is nothing short of insane, but the company has put together a good combination of leaders and visionaries to at least stage a comeback. The big unknown for me was always Rory. Until a few months ago I’d never even met the guy - I’d only heard him speak to analysts, and the nature of those conversations just wasn’t up my alley.
I finally had the opportunity to speak with Rory a few months ago in Austin in a much more candid setting. It was during that meeting that he laid out his strategy for AMD, and it was then that I understood why he was there.
Rory’s playbook for AMD is actually very similar to how he ran things at Lenovo. Lenovo was stuck in a similar position not too long ago: it had a relatively high margin enterprise business that it used to fund and grow a much lower margin consumer business. The new AMD strategy is quite similar.
Traditional margins on x86 CPUs are nothing short of tremendous. The fabless semiconductor manufacturers that compete with Intel don’t operate on anywhere near the same margins. AMD used to play in the same space Intel did, and as a result was often viewed as disappointing as their margins wouldn’t hold up. The problem was that although AMD shed its fab burden, its margins were always viewed as needing to be up at Intel levels. That perception has to change.
Rory’s strategy is to use high margin revenues from existing markets (e.g. PC client, GPUs) and use it to fund a low cost structure expansion into new markets. On paper it’s a sound approach, but the unexpected quick decline of AMD PC sales/shipments threw a wrench in the plan.
The company had to shrink in order to deal with lower than expected revenues than what Rory had planned on when he first took the job almost two years ago. Based on what Raja Koduri told me after he joined a few months ago, it seems like it’s working:
“Raja returns to a very different AMD than the one he left. I asked him what’s different and he responded by saying the AMD he left acted like a company that was 10x its size. Today, AMD is a much smaller and more agile company. Raja believes AMD is in a better position to take advantage of new opportunities vs. being in the hopeless position of never being able to catch up in mature markets.”
So you take the higher margin PC revenues, and use them to invest in lower cost products in new markets. Rory expects semi-custom silicon to see heavy use in these new markets by the way. Lower cost typically means lower margin, which is something the new AMD is ok with. Making Intel margins (or even traditional AMD margins) is tough, in these new markets AMD just needs to be making better margins than the ARM players as the company transitions from being fully PC supported, to PC + additive revenue from these new markets and finally to a position where AMD’s revenues are dominated by these new revenue sources. What are the new markets in specific? Getting anyone at AMD to answer that question today is tough, but I suspect the first place to look is among all of those other players I talked about earlier. The companies tasked with competing with these vertically integrated powerhouses could rely heavily on AMD.
At the end of the day, Rory Read is at AMD because he knows how to navigate a low margin business and run it efficiently to the point where he can use gains in one area to fund another (at Lenovo he took China and Commercial sales to buy him more wiggle room in consumer client). I finally get the plan, now all that’s left is to see if it’ll work.
Facebook
Twitter
RSS
51 Comments
View All Comments
haukionkannel - Friday, September 6, 2013 - link
Exacly!99.999% of the customers can not be wrong. Emailing, and facebook is everuthing you need to do with PC... And companies manufacture devises that most people want.
hobagman - Friday, September 6, 2013 - link
only an ignoramus would compare a cpu to a refrigerator. sure, a stupid consumer might think they are the same.anybody with half a brain would realize how much more complicated the cpu is.
jabber - Friday, September 6, 2013 - link
The irony is strong with this one!I love people that help make my point. I don't even have to pay them.
mrdude - Thursday, September 5, 2013 - link
Intel has to funnel billions into R&D and fabs just to maintain it's process lead, but they have to throw an obscene amount of money if they want to extend that lead. With respect to margins, AMD has the flexibility to operate at far lower margins than Intel does just in order to tread water. That's a huge benefit for them, but this notion that AMD can operate as a fabless semico isn't necessarily true given the nature of their agreement (marriage) with GloFo. AMD doesn't have anywhere near the flexibility of a Qualcomm, nVidia, or other fabless semico's in this respect. Frankly, those two (GloFo+AMD) are joined at the hip and that union hasn't been going well recently. When AMD had a great product in Llano, they were delayed by 6+ months and were supply constrained for 3/4 of the year. Shortly after, AMD had to cancel and their 28nm GloFo Bobcat followups, resulting in a year-long delay in order to make the transition to TSMC's process and also had to pay to get out of their GloFo contract - excess capacity that would have otherwise been filled. Fabless semico really isn't an apt description here.I think Rory&crew have maneuvered a difficult market incredibly well given what they've been tasked with thus far. The bigger questions are regarding AMD's future. Frankly, the short term looks much better than the long term does. Going forward, what can AMD offer that the other guys can't? What's the niche that AMD will fill? Qualcomm was able to take and hold (and will hold for the foreseeable future) the all-important modem corner, but what's AMD gonna' do? At this moment all I see is x86, and that really isn't selling all that well on either front given the lackluster PC sales and the epic failure of MS on the mobile side.
I believe AMD wants to focus on and leverage the graphical power of their APUs, but that's a case of chewing your own foot in order to stave off hunger; and chips like the Richland APU certainly aren't gonna do it. So...
Where's Kaveri? Where's the GDDR5 on an APU? What's the news from AMD that will make me say, "Whoa, I didn't see that one coming"?
Homeles - Thursday, September 5, 2013 - link
"Qualcomm was able to take and hold (and will hold for the foreseeable future) the all-important modem corner."I'm not so sure I agree. While it's certainly not a good idea to count chickens before they hatch, Intel *does* seem to be a big threat when it comes to the modem space. While it will take quite some time for them to port their modem IP to their own process, their upcoming LTE modems are no slouch, despite their disadvantage of being a separate package.
As for AMD, right now, Kabini is in a pretty good position. We're already seeing this with the XBO and PS4. As far as the PC space goes, there is a fun little gap between the upcoming Bay Trail and Intel's Haswell that has the potential to allow AMD to steal a bit of market share.
Kaveri is obviously late, but it should improve AMD's position relative to where they were with Trinity vs. Ivy Bridge. Well, possibly. Haswell is a really big deal in the mobile space, and I find it highly unlikely that Kaveri will be able to provide anywhere near power draw advantage that Haswell offers. Still, the product isn't out yet, and it's hard to evaluate its impact on the market without knowing many details about what it has to offer.
Their FX line is still an unknown, and it is rather frustrating for people that follow the tech industry, like myself, to not have any idea of whether or not we will see a successor to Vishera.
I am quite doubtful that we will see anything unpredictable out of AMD, however. I would expect their big news to be that they start becoming a profitable company sometime in the near future, for better, or for worse.
name99 - Friday, September 6, 2013 - link
It's not clear to me the extent to which Qualcomm are willing to be flexible re modems. They've built their company around a certain model of a modem, with specific functions split between analog and digital blocks. Many companies in a similar situation would rather define themselves as "we're a company that makes analog RF front ends attached to digital base bands" rather than "we're a company that makes the best damn digital modems in the world", even if that means the slide from irrelevance to bankruptcy.Intel, as incumbent, has the advantage of being much more flexible in how they do things. Today their offerings suck, but I could see them being the FIRST to offer a generic "digital communications engine", kinda like a GPU but specialized for signal processing, which would be a single chip capable of running WiFi, telco, and BT all at the same time, and doing most of the RF filtering. I could also see them as far more willing than Qualcomm (and Broadcom) to investigate and experiment with alternative ways to balance power/performance (eg in how adaptive modulation is chosen, how MIMO parameters are chosen, in how to aggregate, in when to use RTS/CTS, etc); I'd also see them as far more willing to push cognitive radio, even if that starts off just with very simple things like much aggressive detection of which 2.4 and 5GHz bands are most crowded, and rapid renegotiation to move to less crowded bands.
savagemike - Friday, September 6, 2013 - link
This may be overly simplistic but as the consumer computer world heads toward mobile Microsoft's hardware endeavors take on a whole new light. They are going to be competing with their hardware partners on what is going to be increasingly the lion share of that hardware market.Nobody like Windows 8. That's just pretty much the fact.
Linux has come a long way in recent years. What it lacks against Microsoft Windows is simply intimacy with the hardware. The capability to really produce drivers which pull the best from the hardware in both graphics and power management.
AMD should pick one primary company (Ubuntu would be the most logical) to really work hand-in-hand with to bring the same kind of fundamental base level driver support of this type of hardware using Linux which Microsoft has traditionally enjoyed. It should then present this partnership as an opportunity for the traditional hardware builders to have a play competing with Microsoft. Build desktops and laptops using AMD hardware married to Linux operating systems.
This will complement their work with enterprise too, where Linux has a strong position already. Better power management capability will become increasingly relevant there too.
Using this combination also differentiates from what Microsoft is doing and makes direct chip comparisons to Intel less interesting.
If I can buy a computer and use it day to day and get great battery life and experience smooth OS usage then I don't care so much what the chips inside do in head-to-head benchmark comparisons against Intel.
All the pieces are there I think if they want to pick them up and work together.
jabber - Friday, September 6, 2013 - link
"Nobody like Windows 8. That's just pretty much the fact."Really? I think the term is a few IT Hacks brewed up a BS storm and a lot of sheep went along with them.
I found it really embarrassing to see so-called respected IT journalists writing "It took me two weeks to work out how to shut it down!"
Sure 8 isn't perfect but then 7/Vista/XP/2000/NT4 etc. were not either.
As for Linux......"It's been a long road..getting from there to here...." Dream on.
name99 - Friday, September 6, 2013 - link
If you are right, then why are the sales figures for Win8 so lousy compared to Win7?jabber - Friday, September 6, 2013 - link
Hmm because if my experience is anything to go by a lot of people finally stumped up and bought a PC around 2010 when 7 had been out a year to replace a lot of their old XP machines. I was selling PCs a lot back in 2010. Now my customers have largely got all the laptops and PCs they need. I'm doing a lot of upgrades and servicing instead.Windows 7 was timed just right for a big PC sales boost and it did. Those folks with a Windows 7 PC are not in the market to buy another PC for at least another 3+ years.
Plus a lot of ordinary folks have realised they dont need a PC and a tablet does all the Ebay/Amazon they need.
The market has changed a lot since 2009. PC kit doesnt need upgrading quite so often as it did and a lot of people bought Windows 7 or tablets. It's not down to Windows 8, not at a ordinary consumer level.
But one can arguie blue in the face either way.