Xilinx Virtex is a powerful FPGA series that can also use HBM2 memory Source: Xilinx
Xilinx Versal, 7nm FPGA Source: Xilinx
Xilinx Alveo U25: FPN based SmartNIC Source: Xilinx
Xilinx Virtex with board supplying FPGA chip connectivity and other infrastructure Source: Xilinx
Xilinx Spartan: a line of cheaper relatively available FPGAs Source: Xilinx
Xilinx Alveo U50 programmable accelerator for various purposes of AI compression networking and other applications Source: Xilinx
Xilinx ZynQ: programmable SoC with ARM cores Source: Xilinx
Xilinx Virtex Source: Xilinx
It is said that AMD, like Intel, wants to buy the FPGA manufacturer in 2015. Can it establish itself in the server market, or is it pure insanity for $ 30 billion?
One of the key, but also a lot of surprising events of this year (if we are talking about computers and IT) was that Nvidia buys ARM and with it these processors and platform, which could theoretically become a truly giant player like Intel.
Not only Nvidia has ambitions. It seems that AMD may also be growing, which, although still not rich, probably also wants to bite some big acquisition after the poor decade of 2006–2016.
After years of hardship, AMD has become hard: it is looking for a huge acquisition
The Wall Street Journal reported yesterday that AMD is reportedly negotiating to buy Xilinx, a leading manufacturer of FPGA programmable gate arrays. Bloomberg also published the same information about this planned acquisition. This probably immediately draws parallels with that In 2015, Intel bought Alter, which was in the same market competition of Xilinx (probably can be said to be number two in the market). Let’s hope AMD has a more thoughtful strategy than just “following what the competition is doing.”
It is a message from apparently several “people familiar with the matter.” However, according to the Wall Street Journal, this is not just some acquisition gossip, because the negotiations are allegedly in an advanced stage of “slapping” could be announced very soon, for example next week.
But neither is there anything guaranteed, because before the negotiations took place for a long time and it happened that they got stuck and the companies lost interest (probably because the conditions offered were not good enough) – however, then they probably returned to the negotiations. Xilinx may be interested in joining, but it’s about money that AMD doesn’t have much.
The transaction could cost over $ 30 billion, making it a giant IT business (compare with acquisition of Mellanox Nvidia – there was a cost of only $ 6.9 billion, ARM costs Nvidia at $ 40 billion). For AMD, the biggest acquisition to date was the $ 5.4 billion strategic purchase of ATi in 2006, which created AMD as we now know it, but it has unhappily met with the deteriorating economic situation of both companies and AMD has long recovered from the debt.
That’s why the $ 30 billion acquisition looks crazy. AMD had only $ 157 million in net income and cash or cash equivalents of $ 1.775 billion in the last quarter, but needs up to $ 1 billion for its own operations. The company has also only recently repaid historical debts (and billions in losses over the last 15-20 years have been amortized by the issuance of shares). The big acquisition in this situation thus gives not only a bold but rather blind impression – let’s hope that it is not behind it that the management’s short-term success has come to the fore.
Possible explanation: the magic of shopping for your own stocks
However, it is theoretically possible that AMD would pay a significant portion of the price with its shares, which are at record highs above $ 80 and the company had a market capitalization of $ 97 billion yesterday. Then perhaps such an acquisition could be relatively painless and perhaps relatively repayable even with the relatively small profits that AMD now has. However, it is still true that the market capitalization of AMD is still not nearly as high as in Nvidia and also probably quite uncertain. The company would have to issue significantly more shares than Nvidia in relation to the existing number in order to cover a large part of the 30 billion.
Xilinx had sales of 828 million in the last quarter of 2019 (for comparison: AMD had sales of 1.93 billion in Q2 2020) and its market capitalization is $ 26 billion.
What would Xilinx bring?
You probably know Xilinx primarily as an FPGA manufacturer, so in AMD’s eyes, this product must be strategically important to complement its server portfolio. Using its own FPGAs (which should be market leaders from Xilinx), it can support the Epyce platform against Intel and Nvidia, as FPGAs have recently been used more frequently as accelerators in server, network infrastructure and similar locations. There is also some interest in their use in AI calculations, an area where AMD has to catch up with competitors (mainly Nvidia).
At the same time, however, FPGAs are also used for networking, so-called SmartNIC. Xilinx is dealing with these, and it is theoretically possible that in this way AMD will also check a little that Mellanox got into the hands of Nvidia. FPGAs are also used in 5G networks and telecommunications infrastructure, as well as in cars, so AMD would enter markets where there is not much now.
It is probably possible that this investment and connection will gradually return in “synergy”, with Xilinx products selling more AMD server processors and connecting them to supporting Xilinx FPGA sales. If this strengthens AMD’s position in data centers and establishes it as a stronger competitor against Intel (and Nvidia), it may pay off.
But there are probably considerable risks – on the one hand, that the financial burden will plunge AMD back into crisis, and on the other hand, there is also the question of whether 30 billion could not have been invested better. Even if the entire transaction can be done “for free” with shares, the question is whether the two businesses will really be successfully integrated and there will be no case where the acquisition is detrimental and will not benefit the acquirer, while the acquired company has problems after the acquisition and change of owner it will only destroy its potential.
AMD he is not very lucky to acquire (Although buying an ATi was probably necessary, though painful), let’s hope it doesn’t turn out badly here.
Galerie: FPGA Xilinx