Interview with Christian Eder, congatec’s Director of Marketing
We are in the midst of IC shortages. How do you address this problem at congatec?
Our embedded systems are designed on dedicated processors and ICs, and there is no substitute for either AMD, Intel or NXP processor technologies. There is also no chance to switch the components within a dedicated platform. So, we rely on careful planning together with our customers and on the market regaining stability. There have always been cycles like these, so we believe that there will come a change in availabilities sooner or later. As an extra service, we can store components in advance.
IC companies like Xilinx and Nvidia are entering the computer-on-module/board market, which has been your bread and butter since you started – do you consider them serious competition?
Computer-on-modules are well established standards for x86 and ARM processor technologies, and this market is mature and served by various vendors. We have a handful of widely accepted standards hosted by the PICMG and SGET standardisation bodies. Any new vendor is welcome to participate in this community. If Xilinx offers, for example, COM HPC modules, then these will be for FPGA applications in combination with ARM or x86 server processors. With Nvidia it is similar with their GPGPU-based Jetson implementations for AI, which are also great for extending ARM- or x86-processors-based designs. So, I would say in the majority of cases it is complementary technology to x86 or ARM computer-on-modules.
Do you see any change in the demand for x86 platforms vs ARM ones?
The percentage of ARM platforms in our portfolio is increasing as they address ultra-low power areas where x86 has no answers; but the larger-scale projects and higher volumes are based on high-end x86 clients and upcoming server processor modules. For COM-HPC, big new business opportunities await in edge computing. But for the ultra-low power sector, there will be significant new business in the ARM sector as well. For example, we recently launched a new NXP i.M8 M Plus processor module with Neural Processing Unit implementation. At the same time, our portfolio now also includes COM Express and COM-HPC modules with 11th generation Intel Core, Xeon and Celeron processors based on Intel’s Tiger Lake H microarchitecture. They will deliver more revenue compared to the SMARC modules with i.MX8 processors. Ultimately, it is a question of growth and we have it in both sectors, which is great.
How is the transition from COM Express to COM-HPC progressing?
It’s just about to take off: By now, there are two COM-HPC Client module generations available on the market. One family with the latest Intel Tiger Lake H processors, and one with earlier Tiger Lake variants. Overall, there are currently 17 different flavors available. They bring PCI Gen 4 and this is a major reason why a transition is taking place. It is not that COM Express does not offer this performance today, but long-term COM-HPC is the form factor for the higher bandwidths up to edge server level. So, new designs choose COM-HPC if the roadmap is driven by performance increase. Adaptation will happen much faster with COM-HPC Server modules as they offer massively enhanced capabilities compared to COM Express Type 7 modules. But it will still take a while as first silicon for these modules is not ready yet. We hope to see first implementations early 2022.
You recently acquired software solutions provider Real-Time System – how has that been integrated into your product portfolio?
Our acquisition of Real-Time System brings software to our portfolio of boards and modules. Its hypervisor software for real-time virtual machines complements our edge and fog server offerings on the software level, creating the foundation for further OEM development and the design of highly-individualised heterogeneous edge servers and IIoT gateways. This real-time hypervisor software is a standalone product, and engineers can buy a license even without our products. Our OEM customers who need to add these IIoT gateways and cloud access to their products can do that if they want to offer new services that lead to new business models and monetisation opportunities.
Do you plan further acquisitions and what are you looking for?
We intend to extend our offerings as we want to maintain our leadership in the embedded and edge computing markets and provide best-in-class services to simplify the use of our technologies as much as possible. With the backing of controlling DBAG Fund and their focus on growing industrial businesses, congatec now also has the financing and M&A experience to take advantage of the expanding market opportunities at the edge. This will help us to further strengthen our position.
I cannot define that in advance because it is also a question of opportunities. I can only say that we have some fields where we want to improve. One sector is the remote management capabilities of our COM-HPC platforms. But not everything is a matter of acquisitions. In the future, congatec is also going to expand its offering by working with solution partners for vision, AI, VR, AR, medical, and dedicated edge computing services – to give you a few examples. Residing in the virtual machine configurations of fog servers, the latter provide IoT gateway and security functions for vulnerability, attack and anomaly detection, or cryptographic functions that are certified highest levels of security standards used in sensible applications.
How do you attract talent?
Deggendorf is a small “Silicon Valley” for embedded computing and especially computer-on-modules – all European major COM vendors are located there. We have a technical university here too, which is excellent.
Germany has many smaller cities like Deggendorf with vendors focusing on a certain market – so not all business is done in Berlin, Hamburg or Munich. Other than that we take advantage of our international setup. We are hiring technical specialists also in our subsidiaries in US and Asia – we’re not limited to Germany only.