LinkBack URL
About LinkBacks
No, what you want is PRICING POWER, due to producing a product that is FASTER than Intel. They need a die size for a quad zen to be roughly the size of Intel's cpu+gpu sides of their skylake etc. Use the whole cpu to dominate then charge accordingly (on top of Intel for as long as the market will pay). Then when Intel inevitably cuts price to stop market share bleed (as AMD would be ramping at TSMC, GF and even Samsung if needed as these two share processes), you drop a little as needed to make sure they're never on the shelf long. This is the only chance to redo their 3yr's of victories in everything around ~2000. The difference between now and then though, is AMD was limited back then to 20% production tops, thus Intel could sit knowing they really couldn't lose more then 20% and still charge what they wanted. Today AMD could easily (due to 2-3 fabbers) take a huge chunk of the market, but again, this is ONLY if they produced a KING. You will never take over the market with a loser that can be priced to death. Intel could bleed for a few years until dusting you again.Originally Posted by lumireleon
A HUGE die (not bigger than Intel's though, so roughly same manufacturing costs) would force Intel to temporarily plop two skylakes together (which should still be wiped, and use more watts) until a REAL chip could be designed (the same 3-4yrs it took Intel to fix the problem last time). Today though that 3+yrs would shift massive share to AMD. It is much more difficult for Intel to fight a price war (against a better chip, can't block like before forcing companies to use white boxes like ASUS did for AMD boards etc) while trying to fight samsung etc in fabs (and arguably slowing losing that war, I see others taking over in 10 or below, like below but the jig is up), fighting for modems, networking, servers etc etc. Too many people fighting them have as much or more money than Intel coming in, so any dent in cpu income would really damage their R&D, both of which would scare shareholders. That would further hurt intels ability to invest, do M&A etc as billions in company market value would drop out (leaving Intel with far less money to work with). I really hope AMD went with a HUGE quad here, as I believe anything below 1.5x Intel's cpu side of the die is a wasted opportunity they may NEVER get again. In the 3yrs of a huge die, they could pocket a Billion a year in profits (like the only Billion profit year they had once before when they had a KING cpu) and pay off their debt before Intel could recover. That may be very conservative also if they were taking large share via 3 different fab producers pumping out as much as they could sell. I for one would happily pay $50 over my Devil's canyon price for an AMD chip that was FASTER across the board. I will not buy a tie even at $50 off. I want an AMD winner that forces Intel to do a new design taking years, allowing AMD to become a far stronger company for years to come.
I'm shocked they pushed the ZEN core in favor of bringing in GPU which has no chance to blow away NV. They are both working with max transistors, same ideas etc, and basically don't have a lot of hardware differences in games. IT isn't the same as the cpu war, where Intel is dedicating more than half of their chips die to gpu now with no replacement sans gpu sitting waiting after years of AMD being totally out, and shareholders pushing for profit and less money spent on stuff like they used to spend with many designs in play at once (used to have many designs going at once just in case AMD got it right, and also trying many different things at once with many teams). Today Intel is more of a massive production company centered on very few designs, as expected with no pc competition I guess, and on top of racing down to ARM levels and ignoring cpu for gpu for years now. This is a perfect storm of sorts for AMD (with fabs to help, ARM occupying most of Intel's time and stealing 4B a year in profits from Intel as they give away chips to compete in mobile) to strike a blow that will take Intel years to recover from. Hopefully management listened to Dirk from 2011 when he left saying you need a KING CPU before doing anything else. They've had 5yrs to do it, we'll see if they fired the right bullet. If they continue to try to be a good 2nd, it's over, management should be fired immediately if the die is not significantly larger than INtel's cpu side. Quit trying to be a cheaper alternative, and start trying to be CHAMP. Champs make money, chumps get priced to death at worst, and leftovers at best. See APU's, 2nd place gpus, non existent enthusiast cpu race. NV makes mint on gpus, INtel makes it on cpus, Apu's for AMD just get squeezed by ARM coming up, Intel going down.
The only chance they have to make a billion in profit in a year is a CPU KING. Consoles might have made them rich a gen or two ago, but at today's volumes and margins (terrible AMD deal, which is why NV wisely passed to not rob R&D from CORE stuff), it isn't enough to even offset cpu/apu losses on top of interest on their debt (barely covers that). Consoles R&D should have been ZEN debut 3-4yrs ago. Management blew it. ARM will slowly kill consoles, just as Q4 has shown (down half from last year OP income). You can have a lot of fun on TV without a console or their $60 games today. It will get worse as Valve/NV release SteamOS for ARM (no doubt working on it, valve wants to sell ARM games for sure to billions of mobile users), and ARM side puts out much more capable boxes (think Shield TV rev2 or 3 at 14nm or 10nm possibly with two socs like current consoles) and games are running vulkan based unreal4/unity5 etc engine games on ARM boxes/tablets output to tv. For $60 on Android's side today I can buy a dozen great games that will take months to get through.
http://www.anandtech.com/show/9976/amd-reports-q4-2015-results-and-fy-2015-earnings
Consoles only just starting, and already producing far lower for AMD. A nintendo win (if rumors are true) won't change this. Gpus/apu's won't make them rich vs Intel/NV either (both can price those down for a year until they fix a 1yr mishap). Only a CPU king can provide a price hike that can stick as enthusiasts pay for PERF period (see titan sales, 980ti, Intel cpu top end etc).
Pray for a KING people or AMD is sunk 2yrs or so from now (less?). They are down to 750mil in cash, injection from fujitsu coming at ~370mil IIRC 2H16, but losing 400m a year will kill all that and make R&D even weaker. WE need an AMD massive cpu, and priced to take the enthusiast/rich guys money. Screw appealing to poor people, it doesn't make you rich. They come 2nd, after the king collects. I love a cheap chip too, but this 2nd place and crap pricing for everything AMD is killing them.
https://beta.finance.yahoo.com/quote/AMD/financials
1.5B+ losses last 3yrs (scroll down to net income). You can see where this goes at current burn rate. Sorry for the long post, but AMD is in dire need of a KING and HIGHER margins (think 60%, not current 30%). Kings get 56-63% like NV/INtel products. They had the engineers to produce a king, but did management let them?
Reply With Quote
