AMD and Xilinx will pair up on Valentine's Day

[mongeese]

What just happened? After announcing its intent more than 15 months ago, AMD has received approval to acquire Xilinx on Monday, Valentine’s Day. It will be the second-largest deal in tech history.

Xilinx and AMD began the proceedings at the end of October 2020. It was simple from the outset; both Xilinx and AMD were publicly traded, and AMD wanted to exchange shares with Xilinx. A ratio of 1.7234 was set: for each share of Xilinx a shareholder owns, they will receive 1.7234 shares of the new AMD that has Xilinx under its umbrella.

However, the deal took several months longer to complete than anticipated, possibly due to interference from the Chinese government. Plus, during the period since the announcement, both the general nature of the tech industry and AMD and Xilinx’s specific circumstances changed significantly.

From October 2020 to last Thursday, AMD shares have risen in value by 67 percent and Xilinx shares by 82 percent. However, on Friday, both company’s shares fell by 10 percent as they announced the finalization of the acquisition.

Based on AMD’s Friday opening price, AMD will buy each Xilinx share for $217 worth of its stock, which is also the exact price that Xilinx shares opened at yesterday morning, too. Some further downturns have occurred today, but the company’s share prices will continue to move in tandem.

A programmable Xilinx board for students.

The deal is valued at approximately $53 billion at the current stock values, making it the second-largest acquisition in the tech industry. It's $14 billion behind Dell's 2015 acquisition of EMC and a similar distance ahead of the third-largest—Avago's acquisition of Broadcom also in 2015. That is, if we're not counting Microsoft's massive gaming acquisition of Activision Blizzard for $68.7 billion just a few weeks ago.

At the end of the deal, current AMD shareholders will own about 74 percent of the 'new' AMD, while current Xilinx shareholders will own the remaining 26 percent. Correspondingly, at least two Xilinx directors will join AMD's board, and Xilinx CEO Victor Peng will serve as the President of the Xilinx arm.

But that’s just the business side—for AMD, the acquisition of Xilinx is no less significant than its acquisition of ATI in 2006. Although Xilinx is not a household name, it is a major player in the enterprise market and is credited with the invention of the FPGA (field-programmable gate array).

These days, Xilinx makes broadly similar products to AMD, making their unification quite logical. For example, both produce accelerators (like GPUs) for data analysis and AI. While AMD targets the top end of the market with high-performance products, Xilinx produces cheaper, more power-efficient, and more flexible alternatives.

It might take multiple years for the fruits of this acquisition to bear, but they’re sure to be sweet.

Image credit: AMD by Maxence Pira, Xilinx Board by Sourabh Belekar

Permalink to story.

https://www.techspot.com/news/93376-amd-xilinx-pair-up-valentine-day.html

 

[psycros]

Xilinx was the mark of crappy products in the 90's and early 2000's.

 

[Sausagemeat]

The acquisition is worth $53 billion? Yeah I don’t ever want to hear anyone saying “AMD are a small company with a limited R&D budget”.

 

[Arbie]

AMD are a small company with a limited R&D budget.

[Edit: Sorry! Didn't see the prev post in time.]

 

[emmzo]

That`s why you should always pay what AMD is asking for their products, because they really need the money. You don`t. You should feel better snubbing Intel.

 

[Dimitriid]

That`s why you should always pay what AMD is asking for their products, because they really need the money. You don`t. You should feel better snubbing Intel.

We don't need the money? I don't get it you think anyone who likes tech is automatically some kind of well off, 6 figures per year person or something?

AMD is not immune to criticism, nobody is. They're just a greedy ****ing corporation like everyone else's: Buy what makes the most sense and sometimes that's what cheapest, whenever or not it has an "AMD" label on it.

 

[emmzo]

We don't need the money? I don't get it you think anyone who likes tech is automatically some kind of well off, 6 figures per year person or something?

AMD is not immune to criticism, nobody is. They're just a greedy ****ing corporation like everyone else's: Buy what makes the most sense and sometimes that's what cheapest, whenever or not it has an "AMD" label on it.

Sorry, I should have added /sarcasm. I thought it was obvious. I see that argument a lot "AMD needs the money" which is garbage. As a consumer you need competition, not fanboyism.

 

[koblongata]

Hardware accelerated software code, it is going to be huge.

 

[Tom Yum]

Xilinx was the mark of crappy products in the 90's and early 2000's.

Xilinx is the leading and largest FPGA provider in the world. Despite having access to Intel resources, Altera is still smaller than Xilinx, lower revenue, lower margin, lower marketshare. But the true value of Xilinx to AMD is likely IP like 5G, PCIe 5, 100Gb Ethernet, etc. What exact Xilinx products are crappy?

The acquisition is worth $53 billion? Yeah I don’t ever want to hear anyone saying “AMD are a small company with a limited R&D budget”.

This acquisition hasn't cost AMD anything, it is a share swap deal, which is what makes it far better than the ATI deal which cost them cash (most of which was borrowed) and incurred a big slab of debt that they had to pay off at a time of higher interest rates and tanking profit margins.

R&D costs actual cash, you can't pay for engineers and prototyping with equity. For that you still need positive cash flow and operating profit. A healthy x86 market requires both AMD and Intel to have both the resources and feel the pressure to maximise investment in R&D. This deal doesn't really change any of that or impact their capacity to invest in R&D as it hasn't cost any cash, nor are they buying a loss making company (so not a drain on margin). What it does do is give AMD access to a lot of already R&D'd IP that they don't need to duplicate. Their need to continue to grow scale (revenue, profit) remains the same.

 

[OortCloud]

Hardware accelerated software code, it is going to be huge.

Eh? What does that even mean?

 

[Lounds]

The author made a mistake, EMC was bought by Dell, not HP.

 

[mongeese]

The author made a mistake, EMC was bought by Dell, not HP.

Fixed, thanks.

Xilinx is the leading and largest FPGA provider in the world. Despite having access to Intel resources, Altera is still smaller than Xilinx, lower revenue, lower margin, lower marketshare. But the true value of Xilinx to AMD is likely IP like 5G, PCIe 5, 100Gb Ethernet, etc. What exact Xilinx products are crappy?



This acquisition hasn't cost AMD anything, it is a share swap deal, which is what makes it far better than the ATI deal which cost them cash (most of which was borrowed) and incurred a big slab of debt that they had to pay off at a time of higher interest rates and tanking profit margins.

R&D costs actually cash, you can't do pay for engineers and prototyping with equity. For that you still need positive cash flow and operating profit. A healthy x86 market requires both AMD and Intel to have the resources and feel the pressure to maximise investment in R&D. This deal doesn't really change any of that or impact their capacity to invest in R&D as it hasn't cost any cash, nor are they buying a loss making company (so not a drain on margin). What it does do is give AMD access to a lot of already R&D'd IP that they don't need to duplicate. Their need to continue to grow scale (revenue, profit) remains the same.

This is a great explanation!

 

[aryehsapir]

Please change: will own about 76 percent ...+...will own 26 percent (=102%).

 

[koblongata]

Eh? What does that even mean?

Check out Analogue FPGA hardware emulated gaming consoles
analogue.co

 

[koblongata]

Eh? What does that even mean?

Or Chord Hugo FPGA Digital to Analog Converter

 

[Tams80]

Excuse me, but deleting my post calling out someone else's ignorance is wrong.

The acquisition is worth $53 billion? Yeah I don’t ever want to hear anyone saying “AMD are a small company with a limited R&D budget”.

That's not how this acquisition worked. Read Tom Yum's reply to you.

And no, this isn't trolling.

 

[Sausagemeat]

Excuse me, but deleting my post calling out someone else's ignorance is wrong.

That's not how this acquisition worked. Read Tom Yum's reply to you.

And no, this isn't trolling.

I understand how the acquisition works, nobody is saying AMD have $53 billion in the bank that they transferred to the seller for this company. It’s actually somewhat disrespectful that you would assume I don’t know how this happened.

But regardless of exactly how the acquisition worked, AMD is now a big corporate player with billions at their disposal. And that’s the point I was making. Apparently you missed that..

 

[Sausagemeat]

Xilinx is the leading and largest FPGA provider in the world. Despite having access to Intel resources, Altera is still smaller than Xilinx, lower revenue, lower margin, lower marketshare. But the true value of Xilinx to AMD is likely IP like 5G, PCIe 5, 100Gb Ethernet, etc. What exact Xilinx products are crappy?



This acquisition hasn't cost AMD anything, it is a share swap deal, which is what makes it far better than the ATI deal which cost them cash (most of which was borrowed) and incurred a big slab of debt that they had to pay off at a time of higher interest rates and tanking profit margins.

R&D costs actual cash, you can't pay for engineers and prototyping with equity. For that you still need positive cash flow and operating profit. A healthy x86 market requires both AMD and Intel to have both the resources and feel the pressure to maximise investment in R&D. This deal doesn't really change any of that or impact their capacity to invest in R&D as it hasn't cost any cash, nor are they buying a loss making company (so not a drain on margin). What it does do is give AMD access to a lot of already R&D'd IP that they don't need to duplicate. Their need to continue to grow scale (revenue, profit) remains the same.

You’ve gone and made a lot of false assumptions about my post. I fully understand how this was acquired and it’s not related to how AMD can fund R&D.

However you frequently hear the tale of AMD as some kind of small player punching up against titans who are far larger than them. And that’s just not true at all. AMD make billions in profit every year. They are worth enough to be able to make one of the biggest acquisitions in the tech world. The statement “AMD have a limited R&D budget” won’t wash any longer.

 

[mbrowne5061]

Xilinx is the leading and largest FPGA provider in the world. Despite having access to Intel resources, Altera is still smaller than Xilinx, lower revenue, lower margin, lower marketshare. But the true value of Xilinx to AMD is likely IP like 5G, PCIe 5, 100Gb Ethernet, etc. What exact Xilinx products are crappy?



This acquisition hasn't cost AMD anything, it is a share swap deal, which is what makes it far better than the ATI deal which cost them cash (most of which was borrowed) and incurred a big slab of debt that they had to pay off at a time of higher interest rates and tanking profit margins.

R&D costs actual cash, you can't pay for engineers and prototyping with equity. For that you still need positive cash flow and operating profit. A healthy x86 market requires both AMD and Intel to have both the resources and feel the pressure to maximise investment in R&D. This deal doesn't really change any of that or impact their capacity to invest in R&D as it hasn't cost any cash, nor are they buying a loss making company (so not a drain on margin). What it does do is give AMD access to a lot of already R&D'd IP that they don't need to duplicate. Their need to continue to grow scale (revenue, profit) remains the same.

Altera also sucks to work with, compared to Xilinx. Xilinx has a much better workflow, and much better documentation in my experience. I can find schematics, and 'theory of operation' on their partner sites (like Digilent), and the software provided make configuration of the gates and programming a breeze (again, compared to Altera. We are still talking about FPGAs here)

Honestly, I do wonder what AMD is thinking here. FPGAs are kind of weird, specialized-generalist hardware. They're what you buy when you need a chip that doesn't yet exist, but don't need a lot of them. Small production runs and prototyping ASICs. Even if AMD is thinking they're going to diversify beyond CPUs and GPUs, buying Xilinx is kind of like buying the dairy farm just to get a glass of milk. Like, fine, if you also want to manage a dairy farm... but why?

I get the feeling that AMD has something planned here, and they want FPGA tech in their portfolio. GPU/FPGA hybrid cards, aimed specifically at the compute/AI/ML and crypto industries, maybe? I've heard of FPGAs being used to load balance multiple GPUs before. Or there will be a GPU with FPGA on-board that devs can make use of in conjunction with their GPU cores, to handle non-vector calculations efficiently without having to go to CPU?

I just hope AMD keeps up with Xilinx's existing board partners, development environment, and documentation through whatever they are planning.

 

[mbrowne5061]

Eh? What does that even mean?

Field Programmable Gate Array.

Basically, one of the first things they teach you in a digital circuits course is that you can re-create any digital circuit exclusively using NAND gates. It won't be as efficient, compared to using all the 'correct' logic gates, but it will still be more efficient than creating the logic with software running on a CPU (in terms of power usage, and cycles needed to complete the task).

So this is what an FPGA does, it provides an array of NAND gates (among other things, and probably other logic gates as well) that can be programming "in the field" to suit whatever task the designer needs it to do. Essentially, you can reconfiguring the hardware of your "CPU" (FPGA) to perfectly suit your needs, then feed it whatever code you need to feed it in the form of C. In fact, once you're satisfied with your FPGA, you can take the 'code' you developed, and send it to the foundry to get turned into an actual hardware chip.

FPGAs sit at the border between software development, and hardware development. A lot of chips - from CPUs, to GPUs, to BIOS, to crypto ASICs, etc - all probably started their life out on an FPGA.

 

[OortCloud]

Field Programmable Gate Array.

Basically....

I know what an FPGA is. It just wasn't clear that was what he was talking about from his reply 'hardware accelerated software code'. But thanks for the refresher!

 

[umbala]

That`s why you should always pay what AMD is asking for their products, because they really need the money. You don`t. You should feel better snubbing Intel.

Uh-huh, we should all be buying $2000 Intel Extreme Edition (lol) processors!

 

[Arbie]

Uh-huh, we should all be buying $2000 Intel Extreme Edition (lol) processors!

Well, they would be 10 cores.

 

[emmzo]

Uh-huh, we should all be buying $2000 Intel Extreme Edition (lol) processors!

We should all be buying what's best for our pockets, that's what I said, sometimes is AMD, sometimes is Intel, but never pay for overpriced products, including the one you listed above. I took on AMD because people think they can get away with bullsh1t because they're "poor" and fighting evil, which is hilarious.