Programmable Solutions Group
Intel’s (INTC) Altera FPGAs (field-programmable gate arrays) are central to its success in the AI (artificial intelligence) space because it would likely use these accelerators to provide AI capabilities in the data center, embedded, and automotive markets.
Intel competes with Xilinx in the FPGA market
Intel’s PSG (Programmable Solutions Group) competes with Xilinx (XLNX), the leader in the FPGA market, which is one year ahead of Intel in terms of FPGA technology. But Intel is slowly catching up by using its vast resources to secure design wins.
In 3Q17, Intel’s PSG revenues rose 10% YoY (year-over-year), or by $44 million, to $469 million, whereas Xilinx’s revenues rose 7% YoY, or by $40.8 million, to $620 million. These figures show that Intel’s PSG revenues are growing faster than Xilinx’s revenues both in dollar and percentage terms. This shows that Altera is now realizing revenue synergies from merging with Intel.
Meanwhile, Intel has secured design wins from Microsoft (MSFT) and Alibaba (BABA) for its FPGAs. Microsoft would likely use Intel’s FPGA for its accelerated deep-learning platform, codenamed Project Brainwave, and Alibaba would likely use FPGAs to accelerate its cloud service. Intel would also likely use FPGAs in its level-3 autonomous solutions for the 2018 Audi A8 and DENSO’s stereo vision system.
The design wins for FPGAs account for the ~$44 million in growth that Intel’s PSG witnessed in fiscal 3Q17. This indicates that PSG revenue is likely to grow faster in coming quarters, and there’s a possibility that Intel could reduce its revenue gap with Xilinx in the future.
Intel completed the acquisition of Mobileye in fiscal 3Q17 and included it in its All Others segment. In fiscal 3Q17, Mobileye earned $39 million in operating income on $82 million in revenues.
Intel’s All Others segment also includes Intel’s future technology research projects. It has been working on quantum computing and has found some success, and it has developed a 17-qubit superconducting test chip for quantum computing and plans to develop a 49-qubit test chip by the end of 2017. The company also unveiled Loihi, a self-learning test chip, that makes machine learning faster and more efficient by mimicking the human brain.
These future technologies have no specific timeline, but Intel’s Mobileye acquisition is an investment in future technologies with more definite results.
In the next part, we’ll see how these investments are impacting Intel’s balance sheet.