There is no shortage of options for investors looking to get into the booming cloud computing market that’s benefiting from emerging tech trends such as artificial intelligence (AI). While tech titans like NVIDIA, Alphabet, and Intel are some of the more popular AI-focused bets out there, there are less fancy ways to buy into AI-driven cloud computing opportunities.
Chipmakers, for example, could get a nice boost from the growth in AI adoption in cloud services that are becoming increasingly ubiquitous. UBS estimates that AI-related chip revenue will jump from $6 billion in 2016 to $35 billion in 2021, and Xilinx (NASDAQ: XLNX) could be a big beneficiary of this rapid growth. Let’s take a look at why.
Xilinx’s FPGAs are gaining traction in the cloud
Xilinx is a supplier of field-programmable gate arrays (FPGAs), programmable chips expected to play a crucial role in enabling AI. So far, graphics processing units (GPUs) from NVIDIA have hogged the limelight when it comes to powering AI applications like cloud computing or self-driving cars. But FPGAs have started making their mark thanks to their versatility and efficiency.
Amazon.com (NASDAQ: AMZN) has already launched a cloud-computing virtual server (known as an ‘instance’ in cloud-computing parlance) based on FPGA chips that allows the company’s cloud customers to address several use cases, such as financial analysis and real-time video processing. For our purposes, though, it’s important to note that Amazon’s F1 cloud computing instance is powered by Xilinx’s UltraScale Plus FPGAs, giving the chipmaker a break in this market already.
Amazon is increasing FPGA deployment in the cloud, too. After originally making the Xilinx-enabled F1 instance available in only one region, Amazon last quarter rolled it out to four regions across the globe, including the GovCloud, which supports the U.S. government. The instance could be made even more widely available this year, as Xilinx is aggressively training developers to deploy cloud-based applications using FPGAs.
Chinese companies Alibaba and Baidu have also decidedto use FPGAs in future cloud computing instances, in place of GPUs. Baidu, China’s biggest search engine, is tapping Xilinx’s FPGAs to power machine-learning and data-security applications in the cloud. E-commerce giant Alibaba will launch two cloud computing instances based on Xilinx FPGAs to handle machine learning and data analytics, among other workloads.
A main reason FPGAs have been gaining so much traction in cloud computing to enable AI applications in place of GPUs is that they can be programmed or reprogrammed to perform specialized tasks. By comparison, a GPU can perform only a specific task; its fixed architecture means that it cannot be used for performing a wide variety of tasks. It’s also more power hungry when compared to an FPGA. Additionally, FPGAs don’t need an operating system to run, where GPUs do, which means that there is one less thing to worry about — adding to their reliability.
So it is not surprising that FPGAs are predicted to carve out a greater slice of the deep-learning chip market, which is expected to grow at an annual pace of 42.2% through 2025, according to Tractica.
FPGAs are driving Xilinx’s growth
The growing use of FPGAs in cloud computing has helped Xilinx deliver nine consecutive quarters of revenue growth. The chipmaker reported 8% year-over-year growth during the latest quarter, driven by a stronger contribution from the advanced products segment, which includes the UltraScale series of FPGAs.
The company’s advanced products now account for 56% of its total sales as compared to 47% in the prior-year period, with sales increasing 30% year over year in the latest quarter. Looking ahead, their contribution to Xilinx’s top line will keep getting better as cloud service providers launch more FPGA-based services.
Xilinx’s current-quarter outlook calls for $650 million in revenue at the midpoint, an increase of almost 7% from the year-ago period. This means that the chipmaker looks well placed to sustain its steady growth in the near term, though it won’t be surprising if it steps on the gas in the future thanks to the growing adoption of FPGA chips.
According to one estimate, the global FPGA market could be worth almost $13 billion by 2023, growing at an annual pace of 9%. Xilinx now dominates the FPGA space with a market share of almost 60%, and it believes that it has the potential to increase to 65% over the next three years.
That type of market share growth seems possible for Xilinx given its growing traction at big cloud players. But even if the company simply holds on to its current share, Xilinx’s annual revenue could spike to almost $8 billion by 2023 based on the estimated growth of the FPGA market. (It generated just under $2.5 billion in revenue over the past year.)
So, Xilinx has huge growth potential as its FPGA lead and impressive clients put it in prime position to tap the multibillion-dollar AI chip market.