I don't have any RGB. This is the stock that's being spent. Yes, it's causing them cash, as stock = cash at time. I'm a DC guy, and would have never have thought that anyone with a thought would have spent 10+B on a fpga company. There are many more flexible, smaller groups that are much cheaper and can come up with new features much quicker.
Nvidia owns Mellanox and it heavily uses Xilinx. So there is some leverage / power play going on. Plus lots of IP. Also I think we will see this play out as not just a data center move as nodes shrink and we move to TSMC's 2nm GAAFET process I'm expecting to see some APU's that look like shrunken more advanced HPC parts.
Is it, though? Instagram was a web 2.0 company with no profit-making mechanism when it was bought. Xilinx are a 36-year-old company that have their products in a huge number of devices, and an absolute treasure-trove of IP.
You seem to be confused, for the investors stock = cash. For the company stock = ownership. They don't get more money if the stock goes up, they don't get less money if the stock goes down. An all-stock swap is AMD stockholders giving part ownership of AMD in exchange for part ownership in Xilinx. The business is the same with new owners.
Can it still be a bad deal for AMD? Sure, if Xilinx becomes a boat anchor dragging the whole company down. Or if they lose focus on their current business to pursue futile synergy products. But AMD spent $0 from their war chest to buy it, it was all through deluting ownership. Assuming the owners agree to that, of course.
>You seem to be confused, for the investors stock = cash. For the company stock = ownership so stock = investment not short term cash withdrawals. For a company stock = additional risk which they need to mitigate for the current investors
Err, stock swap deals result in dilution, which makes it both more difficult and more costly for corporations to raise cash. AMD's debt is low at the moment, so this may not be troublesome, but if they do get in a cash pinch in next several years, this deal very well may come back to bite them.
so stock transactions are stock for assets. The only cash being spent (there is no such thing as stock being spent) is on legal fees, transaction related fees, etc and those are nominal compared to the actual transaction. so in this case, AMD acquires all assets of said company in exchange for issuing stock to said company stockholders.
How is it costing AMD cash ? Xilinx shares are being converted to AMD shares. This means overall more shares but the company value is also increased. It does not cost AMD (the company) a single cent.
Calm down. Xilinx is a financially sound investment. Even if AMD does not themselves leverage Xilinx technologies in the next five years, it is very much a profitable business. Their stock (excluding 2019) has been in an upswing for five years now with about 200% gains.
Xilinx will benefit greatly of the current 5G push as their FPGAs are heavily used by network equipment (thanks to 3GPP standards and their "Long Term Evolution") and because their main competitor is late in delivering their 10nm-class offerings.
Their Engineers would also benefit AMD's other business in the mid term as they have the expertise AMD requires to stop outsourcing chipset design to third parties.
Someone also brought up that Xilinx has expertise in chiplets, which should prove useful. The fanboys going on a rampage here are likely Intel converts thinking AMD is pulling a McAfee or an Infineon, but this is far from either of those cases. I hope that "$" jump out at them because Xilinx and AMD are two powerhouse companies and combining forces merely for increasing financial prowess is a huge win.
that's just it. They don't even have to actually achieve that legendary "synergy" to benefit, although no doubt ways will be found to push that angle with datacentre products.
Not just any FPGA company. Xilinx are the de-facto leader in the FPGA space. AMD chips on National Instruments products, man I hope that day arrives soon!
I wonder if we'll get the opportunity to have an opensource FPGA programmer. As opposed to the current tool for Xilinx which requires your MAC address to even download it and an internet connection to use it so that it can call home transmitting God-only-knows what about you and your programs.
You’re running everything on hardware with black box spyware in it. You had better be air gapped and more if you are trying to achieve a shred of data privacy.
If CPUs and other chips/parts were regularly uploading data to the internet (and required an internet connection to function), I would think it would be publicized. I am aware that there are bugs in the HW. And I am aware of AMD's PSP. I don't like it either, but it's much more benign than the Intel ME equivalent.
Benign in respect to the facts: 1: No DMA access to the whole system. 2: No ROM code that can't be accessed and analyzed. 3: No instructions that people don't know how they work or what they do. 4: It's a microkernel designed for a specific task. Not a full kernel that can do anything. Therefore, programming an exploit will be harder. 5: It's not directly connected to the Ethernet port. It has to go through the main CPU to get access to that port.
IDR if it stops once the system is up and running. I'll have to listen to the CCC talk again, later.
> Therefore, programming an exploit will be harder. I wanted to clarify that it would be harder because getting the exploit to do something useful (to the cracker), will require her to work with an incomplete system.
This is going to be interesting. The markets don't like it as of now haha. Is there any info on how much money is Intel making out of its Altera acquisition? Also, "synergistic operational efficiencies" ... come on!!
Doesn't matter. Even if AMD does not integrate Xilinx into their products soon, Xilinx stock has been an upswing with 200% gain in value for the last five years. Unlike many of Intel's random acquisitions, Xilinx is a healthy company with good future ahead of it that will provide a good bottom line and strong foundation of combined profit.
Altera is probably the only acquisition by Intel that kinda makes sense, specially compared to their investments on AI-companies (which get write down within a few quarters), McAfee, self-driving car components (that end-up not being deployed in production cars), patent trolls, LCD projector chips and even worse bets.
Honestly, I find it surprising that Intel haven't been punished more for these repeatedly dismal decisions. I think McAfee was probably the worst - it sure as hell didn't do anything for the security of their processors and chipsets like they'd originally promised!
Love the Ryzen CPUs, game changers for cheap high performance core counts in desktop and laptops!
However, I previously worked at ATI Technologies and everyone in silicon valley has either worked at or knows someone (roommate/spouse/mistress) that works at one of the big chip companies and knows which are the good ones and terrible ones. AMD sucks, fired so many ppl and destroyed ATI internally with terrible politics and I can't see this being good for Xilinx. Granted IBM/Intel are terrible too, but folks at Xilinx would be able to get jobs at Synopsys or places that'll treat em decent and have much better pay/benefits to boot.
HW is a terrible business, they should've bought a complimentary SW or services company if they wanted to leverage their current stock price into an acquisition. But who knows, this could be the turning point for the next trillion dollar AI company.
Turning point into bankruptcy more like. This was/is an awful decision, period. You can't explain it any other way. No one in their right mind would spend 35B.
So thinks the gamers who dropped a couple thou on a gaming PC. Reality check: your gaming PC is a teeny tiny H20 molecule in the whole profit swimming pool.
come on its a different company, i used to work in AMD around ATI acquisition (left in 2007). I know a lot of people who would like to get into AMD now as opposed to exodus which lasted for the last decade. Pivoting into services can take a long time, look at microsoft. Now most of their revenue is from 365/azure but it took them almost a decade to get there.
Is that still the case now? I'd certainly heard bad things from around the time of the acquisition - AMD's management were notoriously terrible now - but I've had no insight into how things have been since Lisa Su took over.
Interesting that the market and the comments here are reacting somewhat negatively to this news, especially when we know that AMD have been making smart bets over the past several years to get where they are today. I'm inclined to give AMD the benefit of the doubt on knowing what they are doing with Xilinx.
the thing is, what exactly are they getting out of this acquisition? what benefits would FPGAs provide them?
i don't need to make this question in regards to nvidia acquiring mellanox, both their current and future market seems clear as day and they will continue to grow with innovative products, and the synergy between high networking interconnects and GPUs couldn't be more obvious if you've ever seen any datacenter. and with mellanox at hands, their acquisition of cumulus again makes a lot of sense.
what exactly is AMD getting out of an FPGA acquisition? i just don't see anything useful here.
I do wonder how useful FPGAs with Infinity Fabric on them would be. Presumably chiplets would be just as useful for FPGAs as they are in the CPU business. Apart from anything else, it should enable larger devices than you could get with monolithic dies.
And once they've got that, they should be able to replace any of the Zen chiplets in Epyc with FPGA chiplets (at packaging time -- no new silicon!) giving you high performance CPU cores with a very fast connection to however many FPGAs the customer wants.
Xilinx has a huge portfolio of IP. Not just FPGAs, but also the things you synthesize on FPGAs. That's really useful stuff like Ethernet controllers (100Gb), DDR/HBM memory controllers, Software-Defined-Radio, Crypto accelerators, PCI-E PHYs, etc. Then there's the IP related to their SoCs: ARM CPUs and GPUs, specialized interconnects, DSPs, interposer tech, etc.
Let's also not forget that Xilinx has very competitive accelerator cards that are seeing wider industry adoption as we hit the limits of CPU/GPU computation.
This is it, from my perspective. I'm not going to go all out and say it's an *obviously* good move, but it certainly doesn't seem like a bad one, either - especially given the growing threat to the high-end x86 market from ARM and GPU computation.
Revenue diversification by acquiring a company with its own solid independent track record seems smart at this juncture, and they can definitely make a case for bundling IP from both companies into the same products a few years from now.
Do we still use **synergies** ?? I'm thinking there is room for all kinds of side-action, here ___ beyond imagination kinda stuff. 4D-IC-bending-warp-time-string-ding-yee-duper get TsMC on line one stuff.
This is a curiosly high valuation considering $40B NVidia paid for ARM, which is the de-facto CPU architecture in most of the embedded, and mobile world (anywhere outside of the desktop and server x86 market really). I haven't worked with FPGAs in nearly a decade, back then the two big vendors were Xilinx, and Altera (now Intel), who together controlled about 70% of the market, and there were a few other respectable companies (Actel, Lattice), and a few upstarts as well. I can't fathom how a small chunk of a somewhat niche market (FPGAs) is worth as much money, as the de-facto CPU architecture vendor.
While it is true that ARM is widely known just looking at their revenue Xilinx is the bigger company. ARM licenses the CPU-architecture and core design to third parties, as such they are far less involved in the whole process. Manufacturing, and implementing the design is still a huge part of it.
It's also not a buyout but rather a merger, amd were amd is valued at around 3 times Xilinx which is somewhat fair considering their current market value.
You could argue that AMD stock has a curiously high valuation as well, since it's more than doubled in the past year. They may be bargaining with Xilinx based on say, the average stock prices of the two companies over the past year. The plan for the deal certainly was not made overnight, so the current highs of AMD likely were never considered the going rate for a conversion of shares.
As I understand it ARM isn't highly valued on the market because even as the market for their processors grows, their revenue remains relatively static due to their licensing models. Their IP is *incredibly* valuable, but that doesn't lead to a dramatic rise in profits - hence investors aren't terribly excited and the stock price has remained fairly steady.
Basically ARM stock is undervalued compared to their technical value as an IP designer, AMD are relatively over-valued because of their significant growth in profits recently (apparently investors feel confident that will continue), and Xilinx are likely somewhere between the two.
I don't think FPGAs are that niche either TBH - and as others have mentioned, the way Xilinx and ARM sell their products is very different.
Its always good to expand the TAM. But Xilnix is opposite of AMD. Very little growth but profitable and in fact pays dividend. There is significant dilution as its all stock deal. But so far AMD stock is holding well. That could be due to excellent Q3 earnings. It would have otherwise gone up big after earnings.
From a stocks perspective, but take a look at the Xilinx CEOs quote and you'll see why this is happening, basically AMD is taking them in under thier umbrella and we'll see products with deeper Integration in the future as well as crosspolination that would be less likely if they were separate companies.
Also, Xilinx has already been using chiplets for years in their designs.. just FYI. And that is likely a huge deal.
I think it's a bigger play that what most people estimate. 1- Xilinx- good balance sheet, great net profit 2- FPGA and Interconnect about to explode: data centres, 5G, automotive 3- Xilinx has more advanced packaging technology, which will also allow AMD to greatly advance in their goals of integrating CPU, GPU, Accelerators, Memory... all in the same die 4- The accelerators on their own. The way that Apple customized their ARM products. Huge opportunity 5- Back to the data centre- an all integrated solution from one vendor only - CPU, GPU, FPGA, Interconnect
I can see this also allowing the new AMD to both have more sway at TSMC as well as optimally allocating wafers to what is most important to its bottom line.
It wouldn't do AMD any harm at all to be close to another company that have an established record of working closely with TSMC on the development of their latest processes.
What can an FPGA provide AMD? Ever seen Apple’s Afterburner in action?
That’s just one example. Now expand that with professional music production products and you can see how valuable in the creative arts what an FPGA can do.
You get hot n’ bothered about AVX when future custom PCIe post production or studio audio effects offloading cards, to AI cards and/or SoC features added to future Zen and RDNA/CDNA cards are obvious areas where this merger can provide
Seems like AMD just bought ATI again. This one isn't quite as bad as paying 3.5x what it's worth but it's still really bad. I wouldn't want to own it after that price and it's not worth the price now anyway before the buy. I said yesterday they might make 1/2B net. NOPE. Not even 400mil NET if info posted right now is right. I'll listen to the call later. 390mil at this point just tells you they keep charging too little and making nothing. Only expected to make about 60mil on 650mil of parts from consoles (~6mil units expected shortly). Stupid to waste wafers on this crap at $10-15 each vs. servers or HEDT or even large gpus. Too bad AMD hadn't went to the reticle limit at TSMC, as 830mm^2 would have killed NV cards stuck at 630 at samsung on a worse process. You would be charging $1600 at worst for a 3090 killer, of course you would have priced it at $800 anyway and blown the whole thing...ROFL. Bad management. Market for no income is stupid.
BAD MOVE. This is why she's telling you gpus are made up for by cpus. Why not make a good GPU too? You stupidly had a price war going with Intel for 3yrs. They are about to bite back and you still haven't made any income, just stupid people buying stock at ridiculous prices on hype. At some point you need 1B Q's heading to 1.5-2B. NET INCOME, not revenue. The stock is priced like you are making 6B a year heading to 8. But you are MILES from that and Intel is pulling in chips. New one Q1, Q2, then Q1's dead by Q3/4 with another. TSMC/INTC gpus will be flowing Q1 probably too, so whatever your gpus are doing to cause the CEO to say our cpu sales make up for them going down, well, it's going to get worse, especially knowing you went SMALL vs. NV. Welcome to gpu discount land again, while NV bangs out NET INCOME. I'll be surprised if they don't hit 1B NET this Q again (4 in a row 1B+ 2018, heading back there now). Still, wouldn't touch NV stock either...ROFL. Not even with YOUR pole. ;) I can't buy a stock making less than 2018 at 240-280. Someone explain $540? Or 590 short time ago...f'ing insane. Are you making 8.5B now? Hiding 4.5B somewhere? Because that is what it would take for me to say $500 and I'd be running then not buying it.
AMD spent 3yrs gaining share, for what? To give it back to Intel before making billions yearly? By xmas 2021 this is a whole other game, not to mention Intel's ~4 packing techs that can change the whole story too. They are about a year ahead on stacking tech vs. TSMC, so at some point adding layers will be a game changer. We can already see they can combine 4 different fab slabs in one (and not all from them)...Impressive. Intel has chiplets too, so again, I only see AMD losing at some point here. All Intel has to do is cut margin until AMD can't sell (not illegal if you're still making profits), or buy TSMC wafers at a higher price than AMD when they're contract is up. :) Easy. Too bad AMD hadn't just priced at Intel or above levels for 3yrs. Yeah, you'd have less share, but Intel wouldn't have went into BEAST MODE (killing 2 chips in a year it looks like by Xmas 2021?), and you'd have lots more INCOME. Intel can bleed billions per year, see mobile (4B+ pissed away for 4.5yrs). How can AMD compete if Intel does the same to them as mobile next year? Intel would do fine on even 10B NET INCOME, but coming from 21B+ they have a LOT to piss away on wafers at TSMC, pricing vs. AMD, gpus coming to add more income etc. Not good. Long Intel 2021+. Wake me when AMD/NV make more than before. AMD 2009 Q4 1B+ NET INCOME quarter, not YEAR. NV as noted 4Q's 1B+ 2018 at half stock price today (less).
You are out of your fucking mind. Net income isn't that important of a metric when a company is growing. Companies have to spend money to grow. You trying to tell me you wouldn't have bought Amazon stock in 2015 when they had negative net income? How about Tesla 3 years ago?
Also you brought up the Intel GPUs. Maybe you were joking. Hard to tell with the lack of sentence structure and punctuation. Midrange 2013 performance isn't going to convince anybody to buy that product. I will be a complete non-factor in the next hardware cycle. Nobody is going to get excited seeing an Intel Graphics sticker on their laptop. What is going to excite them is having a higher core count to brag about.
I really don't think AMD went into the bargain basement by choice, many of us still remember high priced consumer Athlon 64s back around 2005. AMD has not been in a position to charge more while its cpus were underperforming the competition. Pricing about Intel for 3 years?? 3 years ago they barely sold what they had at low prices. They've really only reached parity up until now, and have strong brand recognition and loyalty to fight against to boot. I think in Q4 when we finally see a strong performance advantage (for the first time in over a decade) we will see prices and margins increasing again. As far Intel goes, you seem very optimistic they will be able to deliver a 14nm part with high core counts that is as fast as 7nm+ TSMC and fits in servers and laptops without overheating. On top, Intel has been missing deadlines since finance expert (read: not a technologist) Bob Swan took over years ago. Swan inherited a huge tech lead, and has optimized Intel for the bottom line given their position. I would not be surprised he is behind Intel's aversion to expensive UV lithography development, which Intel has sadly proven /two nodes in a row/ (oh 10nm barely worked with pre-UV? How about you try at 7nm then?? XD) is absolutely necessary beyond 14nm. Don't try to tell me a 14nm sandwich is going to outperform a single plane 5nm TSMC in power, speed, price, or yield. 2021+ you say... or maybe +++ :P These architectures take ~5yrs to come to fruition, and Intel 7nm on non-UV is already more doomed than 10nm was. Good point about Intel buying TSMC wafers, yet to be seen. Tho that's a fairly speculative thing to be making your bets on - Intel playing dirty with TSMC as the puppet, and none of TSMCs customers ala Apple(!) getting upset about it. Intel just isn't the biggest fish anymore, and they'll continue their decline until they start focusing on providing value to customers instead of milking them. As for Xilinx, probably a strategic IP move against Nvidia more than anything. They're certainly doing fine independently and I don't see them dragging AMD down. You talk about how AMD stock price is valued several times higher than it should be - well then why is the 1.7:1 AMD:Xilinx share ratio such a bad deal? Seems to me like these folks know what they're doing and can separate market hype from reality. To me, looks totally reasonable.
This assclown has been whining that AMD "don't charge enough" for a while, and then when the 5000 series was announced - and the other Intel shills went into overdrive saying they're overcharging - he *carried on saying they're not charging enough*.
No offense, but i'm pretty sure Lisa Su knows what she's doing. After what she did with AMD, if she says they need to buy Xilinx, then they need to buy Xilinx.
She's smart and been a great CEO, but no one is incapable of making mistakes. Xilinx has a strong portfolio of products and technology, but I think the real question is whether its worth $35bn. Put another way, acquiring a company (hence all the shares) is the same as acquiring a few shares: The goal is to buy low, sell high, and AMD is buying when Xilinx's valuation has already skyrocketed during a pandemic whose full economic shakeout is still playing out.
Xilinx had a net income of $612M over the last 12 months compared to a net income of $609M for AMD. This is actually a down year for them compared to last year a well. It will be nice to have a steady stream of income outside of their x86 business since that hasn't been very reliable money maker over their life.
"but they balked at nvidia buying /a company/" - not just any old company! This is that low-information troll strategy of comparing different things and pretending they're the same, and as a result this is a stale, musty, cold-ass take.
Nvidia are being scrutinised because they will be buying a market-defining company that deals directly with many of their competitors. They also have a track record of using acquisitions in an anti-competitive fashion. People aren't "balking" at AMD because none of those things apply here.
I know that hardware companies tend to buy a lot of big FPGAs to run simulations of next-generation chips, but I hadn't thought it was a large enough line item to get "$300m in synergistic operational efficiencies"; Xilinx hasn't got a reputation for gold-plated executives and implausibly lax procurement.
The problem is that large chip designs have to be split across lots of FPGAs, and getting signals across the gap within the cycle means you can't use all the clever transceivers which introduce lots of delay; you end up very limited by pinout and running the design embarrassingly slowly. Particularly embarrassing if your cache is based on fancy multi-ported RAM macros and you have to use several RAM blocks and a ton of logic to implement them.
$300m is a lot and you don't get that by having employees move into smaller cubicles - my bet is that what they actually mean is that between the two companies, they expect that they'll be able to save money when negotiating prices for things that both AMD and Xilinx use, and probably save on licensing fees - there's a really good chance that both companies pay out for parts of the chiplet tech they've both started using, and they might be able to only pay one license fee going forward.
It's not overpriced. It is an all-stock acquisition. AMD is trading at nose-bleed forward-looking earnings levels. Therefore there is risk in AMD's currency price (their share price) at these levels. Therefore AMD needed to pay a premium for an all-stock deal.
I mean, Xilinx is also trading at a pretty elevated price with a P/E ratio nearing 50 (though that's got nothing on AMD).
Truth be told, I think in a way AMD (And Dr. Su to some extent) are in the position of being a victim of the company's recent success: The share price today probably bakes in continued success of Zen in desktop, laptop and enterprise and a certain expectation that they'll be able to claw back some market share from Nvidia on the GPU side.
Since significant growth is already baked-in to the share price, for FURTHER growth Su and Co. need to pull another rabbit out of a hat, which means taking some kind of a gamble into new markets. By far, the fastest (and most certain) way to do that is an acquisition.
Man, there are some strange comments on this story. Here’s the deal: FPGA’s are an increasingly important segment of the semiconductor business and Xilinx is the leader in that space. Sure seams like a smart acquisition to me. Plus, if you’ve never played with FPGA hobbyist programming, you don’t know what you’re missing out on, and Xilinx based development boards are terrific for this stuff (just as surely as Nvidia’s Jetson Nano is a great builders platform).
<a href="https://macpcsoft.com/miracle-box-crack-latest-set... Box</a> is a professional mobile phone system manager and data flashing application that can be used to successfully flash or reformat modern smartphone and tablet devices. ... For example, this app can fix corrupted OS files, change system apps such as those that control onboard cameras, and much <a href="https://macpcsoft.com/">more</a>.
My God, you people have no idea what is going on in the Data Center. Why do you think Intel bought Altera? FPGAs are being used more and more as co-processors for high-end servers. Reconfigurable hardware that can be custom-designed to solve certain problems much faster than a general-purpose CPU. Did anyone look at Xilinx's website (ALVEO and ACAP) ?
It is a great move and I was expecting AMD always to look for a FPGA / ASIC designer to include in their portfolio however I would have expected some smaller ones to tale over.
Next they maybe should look into Marvell oder companies alike to tale over. Marvell would be a great match providing ARM / Cavium ARM talent to revive K12 but even more important networking technology just like Nvidia got with Mellanox. Networking and interconnect KnowHow is important round the profile. They also might have good access to automotive markets which AMD could also use.
I was expecting Marvell to be always a takepver candidate for AMD or Nvidia as it has it all: Networking, ASICs, ARM CPUs and is not too big in size...
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
100 Comments
Back to Article
hehatemeXX - Tuesday, October 27, 2020 - link
WTF! They spent all the money they earned on a FPGA based company :(They better hope they get into the datacenter FAST, or this will turn up to be the "buy" that ruined AMD.
jimjamjamie - Tuesday, October 27, 2020 - link
Relax. They are not spending any cash on this acquisition at all. Your RGB gaming computer parts are safe.hehatemeXX - Tuesday, October 27, 2020 - link
I don't have any RGB. This is the stock that's being spent. Yes, it's causing them cash, as stock = cash at time. I'm a DC guy, and would have never have thought that anyone with a thought would have spent 10+B on a fpga company. There are many more flexible, smaller groups that are much cheaper and can come up with new features much quicker.FreckledTrout - Tuesday, October 27, 2020 - link
Nvidia owns Mellanox and it heavily uses Xilinx. So there is some leverage / power play going on. Plus lots of IP. Also I think we will see this play out as not just a data center move as nodes shrink and we move to TSMC's 2nm GAAFET process I'm expecting to see some APU's that look like shrunken more advanced HPC parts.nathanddrews - Tuesday, October 27, 2020 - link
Still... 35 Instagrams is a lot.Spunjji - Wednesday, October 28, 2020 - link
Is it, though? Instagram was a web 2.0 company with no profit-making mechanism when it was bought. Xilinx are a 36-year-old company that have their products in a huge number of devices, and an absolute treasure-trove of IP.nathanddrews - Thursday, October 29, 2020 - link
You're right, Facebook made a terrible mistake. They're basically bankrupt now.Samus - Tuesday, October 27, 2020 - link
That what surprises me about this. I wonder if this will be approved by regulators due to the likelihood of strong arming nVidia going forward.gamoniac - Saturday, October 31, 2020 - link
Nice one.Kjella - Tuesday, October 27, 2020 - link
You seem to be confused, for the investors stock = cash. For the company stock = ownership. They don't get more money if the stock goes up, they don't get less money if the stock goes down. An all-stock swap is AMD stockholders giving part ownership of AMD in exchange for part ownership in Xilinx. The business is the same with new owners.Can it still be a bad deal for AMD? Sure, if Xilinx becomes a boat anchor dragging the whole company down. Or if they lose focus on their current business to pursue futile synergy products. But AMD spent $0 from their war chest to buy it, it was all through deluting ownership. Assuming the owners agree to that, of course.
Hxx - Tuesday, October 27, 2020 - link
>You seem to be confused, for the investors stock = cash. For the company stock = ownershipso stock = investment not short term cash withdrawals. For a company stock = additional risk which they need to mitigate for the current investors
Endymio - Wednesday, October 28, 2020 - link
Err, stock swap deals result in dilution, which makes it both more difficult and more costly for corporations to raise cash. AMD's debt is low at the moment, so this may not be troublesome, but if they do get in a cash pinch in next several years, this deal very well may come back to bite them.Hxx - Tuesday, October 27, 2020 - link
so stock transactions are stock for assets. The only cash being spent (there is no such thing as stock being spent) is on legal fees, transaction related fees, etc and those are nominal compared to the actual transaction. so in this case, AMD acquires all assets of said company in exchange for issuing stock to said company stockholders.Irata - Tuesday, October 27, 2020 - link
How is it costing AMD cash ? Xilinx shares are being converted to AMD shares. This means overall more shares but the company value is also increased. It does not cost AMD (the company) a single cent.Spunjji - Wednesday, October 28, 2020 - link
"stock = cash at time"No ❤
Makaveli - Tuesday, October 27, 2020 - link
"Your RGB gaming computer parts are safe."lmao dude i'm dying thanks for that.
Hifihedgehog - Tuesday, October 27, 2020 - link
Calm down. Xilinx is a financially sound investment. Even if AMD does not themselves leverage Xilinx technologies in the next five years, it is very much a profitable business. Their stock (excluding 2019) has been in an upswing for five years now with about 200% gains.Hifihedgehog - Tuesday, October 27, 2020 - link
https://www.google.com/search?q=xilinx+stockSpunjji - Wednesday, October 28, 2020 - link
Nice! So, a company in a not-dissimilar position to AMD themselves. Seems sensible.Arsenica - Tuesday, October 27, 2020 - link
Xilinx will benefit greatly of the current 5G push as their FPGAs are heavily used by network equipment (thanks to 3GPP standards and their "Long Term Evolution") and because their main competitor is late in delivering their 10nm-class offerings.Their Engineers would also benefit AMD's other business in the mid term as they have the expertise AMD requires to stop outsourcing chipset design to third parties.
Hifihedgehog - Tuesday, October 27, 2020 - link
Someone also brought up that Xilinx has expertise in chiplets, which should prove useful. The fanboys going on a rampage here are likely Intel converts thinking AMD is pulling a McAfee or an Infineon, but this is far from either of those cases. I hope that "$" jump out at them because Xilinx and AMD are two powerhouse companies and combining forces merely for increasing financial prowess is a huge win.Spunjji - Wednesday, October 28, 2020 - link
that's just it. They don't even have to actually achieve that legendary "synergy" to benefit, although no doubt ways will be found to push that angle with datacentre products.tipoo - Tuesday, October 27, 2020 - link
They're spending stock, not cash, and it makes sense when the valuation is sky high for AMDSpunjji - Wednesday, October 28, 2020 - link
"WTF! They spent all the money they earned"-Wrong. Didn't even read the headline, let alone the article. Mega-derp.
YB1064 - Wednesday, October 28, 2020 - link
Not just any FPGA company. Xilinx are the de-facto leader in the FPGA space. AMD chips on National Instruments products, man I hope that day arrives soon!ballsystemlord - Tuesday, October 27, 2020 - link
I wonder if we'll get the opportunity to have an opensource FPGA programmer. As opposed to the current tool for Xilinx which requires your MAC address to even download it and an internet connection to use it so that it can call home transmitting God-only-knows what about you and your programs.Oxford Guy - Tuesday, October 27, 2020 - link
You’re running everything on hardware with black box spyware in it. You had better be air gapped and more if you are trying to achieve a shred of data privacy.Oxford Guy - Tuesday, October 27, 2020 - link
And by hardware I am referring to CPUs and other chips/parts as well.brucethemoose - Wednesday, October 28, 2020 - link
I think its more about accessability. High accessability is partially why Nvidia GPUs became such popular accelerators.ballsystemlord - Wednesday, October 28, 2020 - link
If CPUs and other chips/parts were regularly uploading data to the internet (and required an internet connection to function), I would think it would be publicized.I am aware that there are bugs in the HW. And I am aware of AMD's PSP. I don't like it either, but it's much more benign than the Intel ME equivalent.
Oxford Guy - Wednesday, October 28, 2020 - link
Benign in what respect?ballsystemlord - Thursday, October 29, 2020 - link
Benign in respect to the facts:1: No DMA access to the whole system.
2: No ROM code that can't be accessed and analyzed.
3: No instructions that people don't know how they work or what they do.
4: It's a microkernel designed for a specific task. Not a full kernel that can do anything. Therefore, programming an exploit will be harder.
5: It's not directly connected to the Ethernet port. It has to go through the main CPU to get access to that port.
IDR if it stops once the system is up and running. I'll have to listen to the CCC talk again, later.
ballsystemlord - Thursday, October 29, 2020 - link
> Therefore, programming an exploit will be harder.I wanted to clarify that it would be harder because getting the exploit to do something useful (to the cracker), will require her to work with an incomplete system.
Teckk - Tuesday, October 27, 2020 - link
This is going to be interesting. The markets don't like it as of now haha. Is there any info on how much money is Intel making out of its Altera acquisition?Also, "synergistic operational efficiencies" ... come on!!
Hifihedgehog - Tuesday, October 27, 2020 - link
Doesn't matter. Even if AMD does not integrate Xilinx into their products soon, Xilinx stock has been an upswing with 200% gain in value for the last five years. Unlike many of Intel's random acquisitions, Xilinx is a healthy company with good future ahead of it that will provide a good bottom line and strong foundation of combined profit.Arsenica - Tuesday, October 27, 2020 - link
Altera is probably the only acquisition by Intel that kinda makes sense, specially compared to their investments on AI-companies (which get write down within a few quarters), McAfee, self-driving car components (that end-up not being deployed in production cars), patent trolls, LCD projector chips and even worse bets.Spunjji - Wednesday, October 28, 2020 - link
Honestly, I find it surprising that Intel haven't been punished more for these repeatedly dismal decisions. I think McAfee was probably the worst - it sure as hell didn't do anything for the security of their processors and chipsets like they'd originally promised!webdoctors - Tuesday, October 27, 2020 - link
Love the Ryzen CPUs, game changers for cheap high performance core counts in desktop and laptops!However, I previously worked at ATI Technologies and everyone in silicon valley has either worked at or knows someone (roommate/spouse/mistress) that works at one of the big chip companies and knows which are the good ones and terrible ones. AMD sucks, fired so many ppl and destroyed ATI internally with terrible politics and I can't see this being good for Xilinx. Granted IBM/Intel are terrible too, but folks at Xilinx would be able to get jobs at Synopsys or places that'll treat em decent and have much better pay/benefits to boot.
HW is a terrible business, they should've bought a complimentary SW or services company if they wanted to leverage their current stock price into an acquisition. But who knows, this could be the turning point for the next trillion dollar AI company.
hehatemeXX - Tuesday, October 27, 2020 - link
Turning point into bankruptcy more like. This was/is an awful decision, period. You can't explain it any other way. No one in their right mind would spend 35B.Arnulf - Tuesday, October 27, 2020 - link
Nobody spent 35B. You don't really understand how this stuff works, do you?poohbear - Tuesday, October 27, 2020 - link
Dude, you act like you saved AMD from bankruptcy and not Lisa Su. Just chill, she knows what she's doing.Hifihedgehog - Tuesday, October 27, 2020 - link
So thinks the gamers who dropped a couple thou on a gaming PC. Reality check: your gaming PC is a teeny tiny H20 molecule in the whole profit swimming pool.Spunjji - Wednesday, October 28, 2020 - link
Amazing how strong the opinions of the wilfully ignorant can be.michaellpl - Tuesday, October 27, 2020 - link
come on its a different company, i used to work in AMD around ATI acquisition (left in 2007). I know a lot of people who would like to get into AMD now as opposed to exodus which lasted for the last decade. Pivoting into services can take a long time, look at microsoft. Now most of their revenue is from 365/azure but it took them almost a decade to get there.Spunjji - Wednesday, October 28, 2020 - link
Is that still the case now? I'd certainly heard bad things from around the time of the acquisition - AMD's management were notoriously terrible now - but I've had no insight into how things have been since Lisa Su took over.jtd871 - Tuesday, October 27, 2020 - link
Interesting that the market and the comments here are reacting somewhat negatively to this news, especially when we know that AMD have been making smart bets over the past several years to get where they are today. I'm inclined to give AMD the benefit of the doubt on knowing what they are doing with Xilinx.Railander - Tuesday, October 27, 2020 - link
the thing is, what exactly are they getting out of this acquisition? what benefits would FPGAs provide them?i don't need to make this question in regards to nvidia acquiring mellanox, both their current and future market seems clear as day and they will continue to grow with innovative products, and the synergy between high networking interconnects and GPUs couldn't be more obvious if you've ever seen any datacenter. and with mellanox at hands, their acquisition of cumulus again makes a lot of sense.
what exactly is AMD getting out of an FPGA acquisition? i just don't see anything useful here.
jamesindevon - Tuesday, October 27, 2020 - link
I do wonder how useful FPGAs with Infinity Fabric on them would be. Presumably chiplets would be just as useful for FPGAs as they are in the CPU business. Apart from anything else, it should enable larger devices than you could get with monolithic dies.And once they've got that, they should be able to replace any of the Zen chiplets in Epyc with FPGA chiplets (at packaging time -- no new silicon!) giving you high performance CPU cores with a very fast connection to however many FPGAs the customer wants.
I'm sure there is a market for that.
cb88 - Tuesday, October 27, 2020 - link
Xilinx has already been doing chiplets on interposers for > 3 FPGA generations... its not it should enable, it DOES enable and has done for years now!SaberKOG91 - Tuesday, October 27, 2020 - link
Xilinx has a huge portfolio of IP. Not just FPGAs, but also the things you synthesize on FPGAs. That's really useful stuff like Ethernet controllers (100Gb), DDR/HBM memory controllers, Software-Defined-Radio, Crypto accelerators, PCI-E PHYs, etc. Then there's the IP related to their SoCs: ARM CPUs and GPUs, specialized interconnects, DSPs, interposer tech, etc.Let's also not forget that Xilinx has very competitive accelerator cards that are seeing wider industry adoption as we hit the limits of CPU/GPU computation.
Spunjji - Wednesday, October 28, 2020 - link
This is it, from my perspective. I'm not going to go all out and say it's an *obviously* good move, but it certainly doesn't seem like a bad one, either - especially given the growing threat to the high-end x86 market from ARM and GPU computation.Revenue diversification by acquiring a company with its own solid independent track record seems smart at this juncture, and they can definitely make a case for bundling IP from both companies into the same products a few years from now.
Smell This - Wednesday, October 28, 2020 - link
Do we still use **synergies** ??
I'm thinking there is room for all kinds of side-action, here ___ beyond imagination kinda stuff. 4D-IC-bending-warp-time-string-ding-yee-duper get TsMC on line one stuff.
I.E. let Dr Su make it so.
torginus - Tuesday, October 27, 2020 - link
This is a curiosly high valuation considering $40B NVidia paid for ARM, which is the de-facto CPU architecture in most of the embedded, and mobile world (anywhere outside of the desktop and server x86 market really). I haven't worked with FPGAs in nearly a decade, back then the two big vendors were Xilinx, and Altera (now Intel), who together controlled about 70% of the market, and there were a few other respectable companies (Actel, Lattice), and a few upstarts as well. I can't fathom how a small chunk of a somewhat niche market (FPGAs) is worth as much money, as the de-facto CPU architecture vendor.qlum - Tuesday, October 27, 2020 - link
While it is true that ARM is widely known just looking at their revenue Xilinx is the bigger company.ARM licenses the CPU-architecture and core design to third parties, as such they are far less involved in the whole process. Manufacturing, and implementing the design is still a huge part of it.
qlum - Tuesday, October 27, 2020 - link
It's also not a buyout but rather a merger, amd were amd is valued at around 3 times Xilinx which is somewhat fair considering their current market value.Railander - Tuesday, October 27, 2020 - link
so they're still independent companies, just with shared IP and a common goal?ZugZug7 - Tuesday, October 27, 2020 - link
You could argue that AMD stock has a curiously high valuation as well, since it's more than doubled in the past year. They may be bargaining with Xilinx based on say, the average stock prices of the two companies over the past year. The plan for the deal certainly was not made overnight, so the current highs of AMD likely were never considered the going rate for a conversion of shares.Spunjji - Wednesday, October 28, 2020 - link
As I understand it ARM isn't highly valued on the market because even as the market for their processors grows, their revenue remains relatively static due to their licensing models. Their IP is *incredibly* valuable, but that doesn't lead to a dramatic rise in profits - hence investors aren't terribly excited and the stock price has remained fairly steady.Basically ARM stock is undervalued compared to their technical value as an IP designer, AMD are relatively over-valued because of their significant growth in profits recently (apparently investors feel confident that will continue), and Xilinx are likely somewhere between the two.
I don't think FPGAs are that niche either TBH - and as others have mentioned, the way Xilinx and ARM sell their products is very different.
trivik12 - Tuesday, October 27, 2020 - link
Its always good to expand the TAM. But Xilnix is opposite of AMD. Very little growth but profitable and in fact pays dividend. There is significant dilution as its all stock deal. But so far AMD stock is holding well. That could be due to excellent Q3 earnings. It would have otherwise gone up big after earnings.cb88 - Tuesday, October 27, 2020 - link
From a stocks perspective, but take a look at the Xilinx CEOs quote and you'll see why this is happening, basically AMD is taking them in under thier umbrella and we'll see products with deeper Integration in the future as well as crosspolination that would be less likely if they were separate companies.Also, Xilinx has already been using chiplets for years in their designs.. just FYI. And that is likely a huge deal.
ivoleiden - Monday, November 2, 2020 - link
I think it's a bigger play that what most people estimate.1- Xilinx- good balance sheet, great net profit
2- FPGA and Interconnect about to explode: data centres, 5G, automotive
3- Xilinx has more advanced packaging technology, which will also allow AMD to greatly advance in their goals of integrating CPU, GPU, Accelerators, Memory... all in the same die
4- The accelerators on their own. The way that Apple customized their ARM products. Huge opportunity
5- Back to the data centre- an all integrated solution from one vendor only - CPU, GPU, FPGA, Interconnect
ChaosFenix - Tuesday, October 27, 2020 - link
I can see this also allowing the new AMD to both have more sway at TSMC as well as optimally allocating wafers to what is most important to its bottom line.Spunjji - Wednesday, October 28, 2020 - link
That's actually a really good call... I just did a little more Googling and came across this:https://www.xilinx.com/news/press/2017/xilinx-arm-...
It wouldn't do AMD any harm at all to be close to another company that have an established record of working closely with TSMC on the development of their latest processes.
mdriftmeyer - Tuesday, October 27, 2020 - link
What can an FPGA provide AMD? Ever seen Apple’s Afterburner in action?That’s just one example. Now expand that with professional music production products and you can see how valuable in the creative arts what an FPGA can do.
You get hot n’ bothered about AVX when future custom PCIe post production or studio audio effects offloading cards, to AI cards and/or SoC features added to future Zen and RDNA/CDNA cards are obvious areas where this merger can provide
cb88 - Tuesday, October 27, 2020 - link
Xilinx has probably more chiplet experience than any other company... they've been doing chiplet based FPGAs for *generations* at this point.TheJian - Tuesday, October 27, 2020 - link
Seems like AMD just bought ATI again. This one isn't quite as bad as paying 3.5x what it's worth but it's still really bad. I wouldn't want to own it after that price and it's not worth the price now anyway before the buy. I said yesterday they might make 1/2B net. NOPE. Not even 400mil NET if info posted right now is right. I'll listen to the call later. 390mil at this point just tells you they keep charging too little and making nothing. Only expected to make about 60mil on 650mil of parts from consoles (~6mil units expected shortly). Stupid to waste wafers on this crap at $10-15 each vs. servers or HEDT or even large gpus. Too bad AMD hadn't went to the reticle limit at TSMC, as 830mm^2 would have killed NV cards stuck at 630 at samsung on a worse process. You would be charging $1600 at worst for a 3090 killer, of course you would have priced it at $800 anyway and blown the whole thing...ROFL. Bad management. Market for no income is stupid.BAD MOVE. This is why she's telling you gpus are made up for by cpus. Why not make a good GPU too? You stupidly had a price war going with Intel for 3yrs. They are about to bite back and you still haven't made any income, just stupid people buying stock at ridiculous prices on hype. At some point you need 1B Q's heading to 1.5-2B. NET INCOME, not revenue. The stock is priced like you are making 6B a year heading to 8. But you are MILES from that and Intel is pulling in chips. New one Q1, Q2, then Q1's dead by Q3/4 with another. TSMC/INTC gpus will be flowing Q1 probably too, so whatever your gpus are doing to cause the CEO to say our cpu sales make up for them going down, well, it's going to get worse, especially knowing you went SMALL vs. NV. Welcome to gpu discount land again, while NV bangs out NET INCOME. I'll be surprised if they don't hit 1B NET this Q again (4 in a row 1B+ 2018, heading back there now). Still, wouldn't touch NV stock either...ROFL. Not even with YOUR pole. ;) I can't buy a stock making less than 2018 at 240-280. Someone explain $540? Or 590 short time ago...f'ing insane. Are you making 8.5B now? Hiding 4.5B somewhere? Because that is what it would take for me to say $500 and I'd be running then not buying it.
AMD spent 3yrs gaining share, for what? To give it back to Intel before making billions yearly? By xmas 2021 this is a whole other game, not to mention Intel's ~4 packing techs that can change the whole story too. They are about a year ahead on stacking tech vs. TSMC, so at some point adding layers will be a game changer. We can already see they can combine 4 different fab slabs in one (and not all from them)...Impressive. Intel has chiplets too, so again, I only see AMD losing at some point here. All Intel has to do is cut margin until AMD can't sell (not illegal if you're still making profits), or buy TSMC wafers at a higher price than AMD when they're contract is up. :) Easy. Too bad AMD hadn't just priced at Intel or above levels for 3yrs. Yeah, you'd have less share, but Intel wouldn't have went into BEAST MODE (killing 2 chips in a year it looks like by Xmas 2021?), and you'd have lots more INCOME. Intel can bleed billions per year, see mobile (4B+ pissed away for 4.5yrs). How can AMD compete if Intel does the same to them as mobile next year? Intel would do fine on even 10B NET INCOME, but coming from 21B+ they have a LOT to piss away on wafers at TSMC, pricing vs. AMD, gpus coming to add more income etc. Not good. Long Intel 2021+. Wake me when AMD/NV make more than before. AMD 2009 Q4 1B+ NET INCOME quarter, not YEAR. NV as noted 4Q's 1B+ 2018 at half stock price today (less).
:nudge> - Tuesday, October 27, 2020 - link
glad we don't go to the same partiesSmell This - Tuesday, November 3, 2020 - link
Uhhh, yeah.Lost in the Ozone, again__the FAIL started with "Seems like AMD just bought ATI again..."
hecksagon - Tuesday, October 27, 2020 - link
You are out of your fucking mind. Net income isn't that important of a metric when a company is growing. Companies have to spend money to grow. You trying to tell me you wouldn't have bought Amazon stock in 2015 when they had negative net income? How about Tesla 3 years ago?Also you brought up the Intel GPUs. Maybe you were joking. Hard to tell with the lack of sentence structure and punctuation. Midrange 2013 performance isn't going to convince anybody to buy that product. I will be a complete non-factor in the next hardware cycle. Nobody is going to get excited seeing an Intel Graphics sticker on their laptop. What is going to excite them is having a higher core count to brag about.
Tilmitt - Tuesday, October 27, 2020 - link
Good to see someone actually interested in going through the numbers.Spunjji - Wednesday, October 28, 2020 - link
It's just a semi-literate rant though. Sure, it contains numbers - but they're not *useful* numbers.ZugZug7 - Tuesday, October 27, 2020 - link
I really don't think AMD went into the bargain basement by choice, many of us still remember high priced consumer Athlon 64s back around 2005. AMD has not been in a position to charge more while its cpus were underperforming the competition. Pricing about Intel for 3 years?? 3 years ago they barely sold what they had at low prices. They've really only reached parity up until now, and have strong brand recognition and loyalty to fight against to boot. I think in Q4 when we finally see a strong performance advantage (for the first time in over a decade) we will see prices and margins increasing again.As far Intel goes, you seem very optimistic they will be able to deliver a 14nm part with high core counts that is as fast as 7nm+ TSMC and fits in servers and laptops without overheating. On top, Intel has been missing deadlines since finance expert (read: not a technologist) Bob Swan took over years ago. Swan inherited a huge tech lead, and has optimized Intel for the bottom line given their position. I would not be surprised he is behind Intel's aversion to expensive UV lithography development, which Intel has sadly proven /two nodes in a row/ (oh 10nm barely worked with pre-UV? How about you try at 7nm then?? XD) is absolutely necessary beyond 14nm. Don't try to tell me a 14nm sandwich is going to outperform a single plane 5nm TSMC in power, speed, price, or yield. 2021+ you say... or maybe +++ :P These architectures take ~5yrs to come to fruition, and Intel 7nm on non-UV is already more doomed than 10nm was.
Good point about Intel buying TSMC wafers, yet to be seen. Tho that's a fairly speculative thing to be making your bets on - Intel playing dirty with TSMC as the puppet, and none of TSMCs customers ala Apple(!) getting upset about it. Intel just isn't the biggest fish anymore, and they'll continue their decline until they start focusing on providing value to customers instead of milking them.
As for Xilinx, probably a strategic IP move against Nvidia more than anything. They're certainly doing fine independently and I don't see them dragging AMD down. You talk about how AMD stock price is valued several times higher than it should be - well then why is the 1.7:1 AMD:Xilinx share ratio such a bad deal? Seems to me like these folks know what they're doing and can separate market hype from reality.
To me, looks totally reasonable.
ZugZug7 - Tuesday, October 27, 2020 - link
*pricing /above/ IntelSpunjji - Wednesday, October 28, 2020 - link
This assclown has been whining that AMD "don't charge enough" for a while, and then when the 5000 series was announced - and the other Intel shills went into overdrive saying they're overcharging - he *carried on saying they're not charging enough*.Spunjji - Wednesday, October 28, 2020 - link
Awesome, if you're against it then it must be a good decision 👍 Time to buy AMD stock peeps, TheJian is whinging again!Teckk - Wednesday, October 28, 2020 - link
LOL. You read his comments !??poohbear - Tuesday, October 27, 2020 - link
No offense, but i'm pretty sure Lisa Su knows what she's doing. After what she did with AMD, if she says they need to buy Xilinx, then they need to buy Xilinx.sing_electric - Wednesday, October 28, 2020 - link
She's smart and been a great CEO, but no one is incapable of making mistakes. Xilinx has a strong portfolio of products and technology, but I think the real question is whether its worth $35bn. Put another way, acquiring a company (hence all the shares) is the same as acquiring a few shares: The goal is to buy low, sell high, and AMD is buying when Xilinx's valuation has already skyrocketed during a pandemic whose full economic shakeout is still playing out.Martimus - Tuesday, October 27, 2020 - link
Xilinx had a net income of $612M over the last 12 months compared to a net income of $609M for AMD. This is actually a down year for them compared to last year a well. It will be nice to have a steady stream of income outside of their x86 business since that hasn't been very reliable money maker over their life.https://www.macrotrends.net/stocks/charts/AMD/amd/...
https://www.macrotrends.net/stocks/charts/XLNX/xil...
six_tymes - Tuesday, October 27, 2020 - link
no one balks at this, but they balked at nvidia buying a company. I smell corruption.WaltC - Tuesday, October 27, 2020 - link
Great move, AMD...nice...;) AMD is moving ahead quickly!Spunjji - Wednesday, October 28, 2020 - link
"but they balked at nvidia buying /a company/" - not just any old company! This is that low-information troll strategy of comparing different things and pretending they're the same, and as a result this is a stale, musty, cold-ass take.Nvidia are being scrutinised because they will be buying a market-defining company that deals directly with many of their competitors. They also have a track record of using acquisitions in an anti-competitive fashion. People aren't "balking" at AMD because none of those things apply here.
TomWomack - Tuesday, October 27, 2020 - link
I know that hardware companies tend to buy a lot of big FPGAs to run simulations of next-generation chips, but I hadn't thought it was a large enough line item to get "$300m in synergistic operational efficiencies"; Xilinx hasn't got a reputation for gold-plated executives and implausibly lax procurement.The problem is that large chip designs have to be split across lots of FPGAs, and getting signals across the gap within the cycle means you can't use all the clever transceivers which introduce lots of delay; you end up very limited by pinout and running the design embarrassingly slowly. Particularly embarrassing if your cache is based on fancy multi-ported RAM macros and you have to use several RAM blocks and a ton of logic to implement them.
sing_electric - Wednesday, October 28, 2020 - link
$300m is a lot and you don't get that by having employees move into smaller cubicles - my bet is that what they actually mean is that between the two companies, they expect that they'll be able to save money when negotiating prices for things that both AMD and Xilinx use, and probably save on licensing fees - there's a really good chance that both companies pay out for parts of the chiplet tech they've both started using, and they might be able to only pay one license fee going forward.Roy2002 - Tuesday, October 27, 2020 - link
Somehow this reminds me the ATI purchase, overpriced!Yojimbo - Tuesday, October 27, 2020 - link
It's not overpriced. It is an all-stock acquisition. AMD is trading at nose-bleed forward-looking earnings levels. Therefore there is risk in AMD's currency price (their share price) at these levels. Therefore AMD needed to pay a premium for an all-stock deal.sing_electric - Wednesday, October 28, 2020 - link
I mean, Xilinx is also trading at a pretty elevated price with a P/E ratio nearing 50 (though that's got nothing on AMD).Truth be told, I think in a way AMD (And Dr. Su to some extent) are in the position of being a victim of the company's recent success: The share price today probably bakes in continued success of Zen in desktop, laptop and enterprise and a certain expectation that they'll be able to claw back some market share from Nvidia on the GPU side.
Since significant growth is already baked-in to the share price, for FURTHER growth Su and Co. need to pull another rabbit out of a hat, which means taking some kind of a gamble into new markets. By far, the fastest (and most certain) way to do that is an acquisition.
zodiacfml - Tuesday, October 27, 2020 - link
What is going on though with Intel/Altera and AMD/Xilinx? It seems like they're going to war. What are they fighting over?Spunjji - Wednesday, October 28, 2020 - link
The future of computing..?JCB994 - Wednesday, October 28, 2020 - link
Where have you been for the last 30 years?sing_electric - Wednesday, October 28, 2020 - link
Market share? Just like everyone else in capitalism?Buck Turgidson - Saturday, October 31, 2020 - link
Man, there are some strange comments on this story. Here’s the deal: FPGA’s are an increasingly important segment of the semiconductor business and Xilinx is the leader in that space. Sure seams like a smart acquisition to me. Plus, if you’ve never played with FPGA hobbyist programming, you don’t know what you’re missing out on, and Xilinx based development boards are terrific for this stuff (just as surely as Nvidia’s Jetson Nano is a great builders platform).six_tymes - Monday, November 2, 2020 - link
I smell corruption. this is OK but nvidia buying arm isn't? puuuulease... the corrupt are at it again.Smell This - Tuesday, November 3, 2020 - link
Uhhh . . . what?
AllenForum - Wednesday, November 11, 2020 - link
<a href="http://key4crack.com/macbooster-crack-licence-key/... is an optimization tool for Mac computers. You can make your Mac work faster, as well as being more stable and safer, and as clean as a whistle. <a href="http://key4crack.com/3utools-crack-key/">3... Full Version</a><a href-"http://key4crack.com/dc-unlocker-crack/">D... Unlocker</a>
AllenForum - Wednesday, November 11, 2020 - link
<a href="https://macpcsoft.com/miracle-box-crack-latest-set... Box</a> is a professional mobile phone system manager and data flashing application that can be used to successfully flash or reformat modern smartphone and tablet devices. ... For example, this app can fix corrupted OS files, change system apps such as those that control onboard cameras, and much <a href="https://macpcsoft.com/">more</a>.<a href="https://macpcsoft.com/rosetta-stone-crack/"&g... Stone</a>
AllenForum - Wednesday, November 11, 2020 - link
<a href="https://crackpur.com/arcade-vst-output-crack-vst-p... Arcade</a>. When it comes to Arcade VST, there's definitely no shortage of flexibility to create great music.<a href="https://crackpur.com/boom-3d-crack-latest-official... 3D</a> is an all-in-one sound enhancer that lets you unlock peak performance from your Mac's audio hardware.
wayneperrier - Thursday, November 12, 2020 - link
My God, you people have no idea what is going on in the Data Center. Why do you think Intel bought Altera? FPGAs are being used more and more as co-processors for high-end servers. Reconfigurable hardware that can be custom-designed to solve certain problems much faster than a general-purpose CPU. Did anyone look at Xilinx's website (ALVEO and ACAP) ?Matthias B V - Saturday, April 10, 2021 - link
It is a great move and I was expecting AMD always to look for a FPGA / ASIC designer to include in their portfolio however I would have expected some smaller ones to tale over.Next they maybe should look into Marvell oder companies alike to tale over. Marvell would be a great match providing ARM / Cavium ARM talent to revive K12 but even more important networking technology just like Nvidia got with Mellanox. Networking and interconnect KnowHow is important round the profile. They also might have good access to automotive markets which AMD could also use.
I was expecting Marvell to be always a takepver candidate for AMD or Nvidia as it has it all: Networking, ASICs, ARM CPUs and is not too big in size...