-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
AMD on the road to recovery.
- Thread starter humbug
- Start date
It results in improving their offerings, and allows them to compete against Intel in the FPGA market. FPGA market is currently only $2 billion a year but expected to grow by 4-5x in the next 5 years due to insane demands and is a somewhat duopoly between Intel and Xilinx (the two have 85-90% of the market). Intel have been buying smaller players in recent years to gain grounds on Xilinx. FPGA market is more profitable than the CPU market, allowing AMD to not only improve their margins and diversify their revenue streams but also gain some talent that is useful for CPU microarchitecture development.
Yup, but AMD need to refile in the US, it should just be a formality as the US had already approved its just that it ran out on the 11'th of January due to the China approval delay.
With it AMD will absorb Xilinx markets and grow new ones, AMD will also absorb Xilinx $4 Billion annual revenue making them larger than Nvidia in revenue terms.
IBM has a 2 nm transistor. The race to reduce nanometers in chips (analyticsindiamag.com)
TSMC probably a 1 nm transistor. The race to reduce nanometers in chips (analyticsindiamag.com)
Intel claims 20A and 18A (Angstrom) after its Intel 3 node. Intel Process Roadmap Through 2025: Renamed Process Nodes, Angstrom Era Begins | Tom's Hardware (tomshardware.com)
So, what happens in several years?
"Eventually the laws of physics will cause us to be unable to advance the amount of transistors we can put on a chip, and it looks like we are getting close."
What exactly is next after nanometer processors (after it gets smaller then 1 nanometer)? What is planned in the future? : hardware (reddit.com)
"As others have pointed out, we are no where close to 1 nm for actual transistor sizes.
To get to 1nm we might have to switch from Si to some other material like carbon nanotubes.
Around the 1nm mark you lose the ability to shrink electronics smaller due to quantum mechanical tunneling, effectively, you can no longer turn off a transistor.
At that point we will have to start building our computers bigger and bigger and use more energy. That means we will probably have to start putting them in space. Use solar energy for power. Eventually, there will probably come a point in the future when a sizeable proportion of the energy generated by the sun and the available materials in the solar system is just used for computing."
What exactly is next after nanometer processors (after it gets smaller then 1 nanometer)? What is planned in the future? : hardware (reddit.com)
So, I actually expect that the "Moore's law" will dramatically slow down during the next years, and the usual annual or per-year generational improvements that we have previously seen, will become per-two or even per-three years, with the whole progress ever slowing down.
TSMC probably a 1 nm transistor. The race to reduce nanometers in chips (analyticsindiamag.com)
Intel claims 20A and 18A (Angstrom) after its Intel 3 node. Intel Process Roadmap Through 2025: Renamed Process Nodes, Angstrom Era Begins | Tom's Hardware (tomshardware.com)
So, what happens in several years?
"Eventually the laws of physics will cause us to be unable to advance the amount of transistors we can put on a chip, and it looks like we are getting close."
What exactly is next after nanometer processors (after it gets smaller then 1 nanometer)? What is planned in the future? : hardware (reddit.com)
"As others have pointed out, we are no where close to 1 nm for actual transistor sizes.
To get to 1nm we might have to switch from Si to some other material like carbon nanotubes.
Around the 1nm mark you lose the ability to shrink electronics smaller due to quantum mechanical tunneling, effectively, you can no longer turn off a transistor.
At that point we will have to start building our computers bigger and bigger and use more energy. That means we will probably have to start putting them in space. Use solar energy for power. Eventually, there will probably come a point in the future when a sizeable proportion of the energy generated by the sun and the available materials in the solar system is just used for computing."
What exactly is next after nanometer processors (after it gets smaller then 1 nanometer)? What is planned in the future? : hardware (reddit.com)
So, I actually expect that the "Moore's law" will dramatically slow down during the next years, and the usual annual or per-year generational improvements that we have previously seen, will become per-two or even per-three years, with the whole progress ever slowing down.
That means reliance on IPC improvements, not on die shrinks and packing more transistors on the packages.
In any case, the point when we will be screwed all together approaches fast.
Soldato
I wonder if pc's will have built in FPGAs one day - could also explain why AMD wanted in on a manufacturer? Might be a stretch I know, and just being a complementary business makes sense too, but I like the idea...
Today we have Alder Lake with the performance and efficiency cores. In future as mentioned maybe we'll have separate 'cores' for rasterising, neural networking etc... Maybe the step after that would be to have a set of FPGA 'cores'. Could be reprogrammed dynamically on the fly - playing Far Cry 13? Well either there's a preset template to load up, or your PC will notice patterns in the operations it's doing and come up with a custom logic design itself. Basically build your own ASIC on the fly. Then specific workloads could be farmed out to the FPGA 'core', potentially massively increasing processing speed for specific workloads...
Today we have Alder Lake with the performance and efficiency cores. In future as mentioned maybe we'll have separate 'cores' for rasterising, neural networking etc... Maybe the step after that would be to have a set of FPGA 'cores'. Could be reprogrammed dynamically on the fly - playing Far Cry 13? Well either there's a preset template to load up, or your PC will notice patterns in the operations it's doing and come up with a custom logic design itself. Basically build your own ASIC on the fly. Then specific workloads could be farmed out to the FPGA 'core', potentially massively increasing processing speed for specific workloads...
Wasting money. $35B!
AMD could have just hired a new research and development department to enter direct competition with Xilinx.
Yes, there are maybe patents and royalty payments, but better than spending $35B.
AMD could have used this money to improve its CPU and GPU market shares. Like Intel - pay the partners to avoid Intel's products..
...no. The FPGAs are used in specialised high-performance computing like aviation, aerospace, medical devices, military equipment, radars, etc... There will be no point putting FPGAs inside consumer PCs. They're very expensive, not efficient, not suitable for mass production, and generally the reason computers became cheaper is because we moved to ASIC instead of FPGAs. We're not go back decades in time.
Not how any of these stuff work in reality.
Wasting money. $35B!
AMD could have just hired a new research and development department to enter direct competition with Xilinx.
Yes, there are maybe patents and royalty payments, but better than spending $35B.
AMD could have used this money to improve its CPU and GPU market shares. Like Intel - pay the partners to avoid Intel's products..
They're not "spending" $35 billion. It's a merger, the shareholders are paying for it, not AMD itself. In fact it improves AMD's balance sheet overall because Xilinx has a higher profit margin than AMD and better credit rating. This has no effect on what they can spend on their CPU/GPU side of business.
Your suggested path would have taken over a decade, tons of spending out of their balance sheet, and it would have given Intel more time to gain ground on Xilinx so AMD would have been a distant third competitor (after Xilinx and Intel) instead of being the leader with over 50% of the global market share immediately.
Last edited:
The shareholders pay?!
I guess they will start selling their shares immediately and the stock price has already lost 30% in a month! Down from $153 in December to $103 as of today!
AMD 103.39 0.79 0.77 : Advanced Micro Devices, Inc. - Yahoo Finance
The shareholders pay?!
Yes. The new entity (after the merger) will issue new shares under AMD's name to shareholders of both companies, weighted by AMD's market cap and Xilinx's $35 billion valuation. E.g. AMD is worth $125 billion, if you held 1% of AMD before the merger, that's worth $1.25 billion, and after the merger the new entity is worth (on paper) $125bn+$35bn=$160bn, and you will hold 0.78% of it. Then when the markets open the market decides if the new entity is actually worth $160 billion, or whether's it's worth more or less. The only money AMD and Xilinx actually spend here are the legal and operational fees for the merger.
So AMD is putting the financial burden and risk of the acquisition on the shareholders rather than its own balance sheet, which is the smart thing to do. This arrangement is quite typical in these type of mergers.
Nvidia is also down 27%, it's mostly because of the crypto crash and all the reports that GPU sales and prices are going down, in addition to the whole market being down (S&P500 is also down 10%).
AMD Q4 2021 and year.
Q4 2021:
$4.826 Billion
Gross Profit $2.426 Billion
Margins 50%
2021 Total
$16.434 Billion (Up 49%)
Gross Profit $7.929 Billion (Up 67%)
https://d1io3yog0oux5.cloudfront.ne...ts_Fourth_Quarter_and_Full_Year_2021_1044.pdf
Q4 2021:
$4.826 Billion
Gross Profit $2.426 Billion
Margins 50%
2021 Total
$16.434 Billion (Up 49%)
Gross Profit $7.929 Billion (Up 67%)
https://d1io3yog0oux5.cloudfront.ne...ts_Fourth_Quarter_and_Full_Year_2021_1044.pdf
Soldato
Yep, totally smashed all expectations then . Earnings call is live in 4 mins.
"As we launch Zen 4 CPU's and RDNA3 GPU's in 2022" 
Wasting money. $35B!
AMD could have just hired a new research and development department to enter direct competition with Xilinx.
Yes, there are maybe patents and royalty payments, but better than spending $35B.
AMD could have used this money to improve its CPU and GPU market shares. Like Intel - pay the partners to avoid Intel's products..
Err no.
Xilinx have some really interesting 2.5D, 3D and interconnect patents that can really help AMD as they expand the use of chiplets to more market sectors.
As others have said this is a merger paid with stock so the actual cash cost to AMD is purely the fees of merging. Further as Lisa has said many times adding Xilinx is immediately accretive, that means it will increase profit and increase gross margin as soon as the deal is closed.
Its an acquisition, AMD are buying Xilinx.
Perhaps what can be a little confusing about it is that AMD are not reaching into a pocket and handing over physical money to current owners of the company, its an all-stock transaction, current owners get AMD stock shares, to the tune of $35 Billion, AMD can do this because they are stock rich and no debts, there is about $140 Billion of shares floating around in AMD right now.
Xilinx effectively become AMD, every physical asset they own, their Intellectual Property, their workforce..... even their earnings get added to AMD's books. all of it comes under the control of AMD, much like ATI continued to trade as ATI from 2006 to 2010 under AMD as the parent company Xilinx will probably continue to trade under that name for a few years, until such time when or if AMD decide to drop the name, as with ATI which no longer exists.
Perhaps what can be a little confusing about it is that AMD are not reaching into a pocket and handing over physical money to current owners of the company, its an all-stock transaction, current owners get AMD stock shares, to the tune of $35 Billion, AMD can do this because they are stock rich and no debts, there is about $140 Billion of shares floating around in AMD right now.
Xilinx effectively become AMD, every physical asset they own, their Intellectual Property, their workforce..... even their earnings get added to AMD's books. all of it comes under the control of AMD, much like ATI continued to trade as ATI from 2006 to 2010 under AMD as the parent company Xilinx will probably continue to trade under that name for a few years, until such time when or if AMD decide to drop the name, as with ATI which no longer exists.