From its beginnings in Vermont, USA as a supplier of motherboards and components, OnLogic has grown over the last 18 years into a global industrial PC manufacturer and solution provider focused on hardware for the IoT edge. The company continued growing as businesses accelerated digital transformation efforts during the pandemic and, as executives explained to Edge Industry Review, is positioning itself for further expansion.
In August 2021, OnLogic unveiled plans to build a new $60M plus global headquarters in South Burlington Vermont; the 145,000 square foot facility will house manufacturing, warehousing, testing facilities and all other business operations. By way of comparison, the company’s current Vermont facility was expanded to 36,000 sq. feet in 2015. An OnLogic office was also recently opened in the Raleigh, North Carolina area, in part to help the company tap into a larger talent pool.
OnLogic also plans to double the size of their European office this year, as revenues in the region have tripled over the last 5 years. The company is also significantly growing its Engineering and Supply Chain teams in Malaysia and Taiwan.
EdgeIR spoke to Michael Kleiner, Vice President of Engineering for OnLogic, to discuss some of the key issues and trends impacting the edge hardware market in 2021 and beyond.
Broadly speaking, Kleiner said OnLogic’s growth is being driven by digital transformation. “Basically, our customers are using our technology to make their manufacturing, their warehouses, and their logistics solutions more intelligent. Customers are introducing a lot more automation in things like package handling, warehousing, inventory tracking and even automated guided vehicles or robots that move materials, to make operations more efficient, cost-effective, and more reliable” according to Kleiner. And that’s just one segment of OnLogic’s overall customer base.
The following comments from the interview have been edited for length and clarity.
What are the key products/technologies customers are interested in adopting this year and into 2022? Is there interest in new GPU and edge AI chip designs?
Michael Kleiner (MK): One of the exciting things we’re seeing is the diversification of technology options and the adoption of those options by customers. Intel has always been a great provider of technology solutions for the industrial and edge space. Intel’s Elkhart Lake technology, for example, is something we’re already seeing significant interest in. The combination of power efficiency and performance is extremely appealing to edge hardware users.
AMD has also begun to pick up share in the x86 space, particularly in higher compute edge devices and servers. Inference and data analysis is an area where we see strong growth. For example, we’re working with a number of the biggest companies in the world who are using our edge server systems for testing semiconductors.
In some instances, for particularly compute-intense workloads, there’s an increasing need for GPU, VPU, and TPU options to offload some of the computations or do image data analytics. This has led to similar diversification in the AI space, where folks like Nvidia are building solutions to manage increasingly large image and video data processing needs.
Where is your customer base in terms of chip technology for edge applications? Are they leaning towards ARM designs?
MK: We launched Helix 500 and 600 products last year with 10th generation Intel processors. They combine desktop CPU performance in a small form factor and they’re fanless, while still maintaining the 35 Watt processors. We’re very excited about adoption of those. As mentioned, Intel and AMD continue to be primary players in the edge space, but the particular needs of developers are leading to further diversification.
We are seeing a real push in terms of ARM in the last few years. For our customer base in industrial edge markets, we think an entry-level solution could be an interesting option. Of course, it’s always important to address the software stack question to make sure there’s an offering the customers can actually use, but on the hardware side, Raspberry Pi is one example of an ARM-based platform that’s making strides toward providing an industrial option for developers. In a few months, we will be launching our own industrial ARM solution based on the Raspberry Pi compute module 4. We’re excited to share more about that soon.
In terms of applications and solutions, how key is edge AI for growing your business versus a normal upgrade cycle, in the market for control systems, for example?
MK: Overall, AI is absolutely key in really getting the plants, logistics warehouses, factories, and farms to the next level of efficiency and intelligence, so to speak.
If you talk about upgrading control systems…we see some growth in computer-based PLCs. Some of our customers are actually using a PC to implement PLC functionality. That gives them the benefit of also using the PC to run applications, as well as collect and upload data at the source. That’s one trend we’re seeing.
Which edge AI workloads are currently the most prevalent?
MK: Computer vision is probably the biggest one because you’re detecting and identifying things based on an image. Again, that could be in factories and warehouses, but vision and machine learning are happening well outside those environments as well. We have a client who’s using ML to do fisheries monitoring onboard fishing vessels in Hawaii.
We also work with clients doing inventory management that combines machine vision and robotics, as well as production automation for assembly, materials handling and quality control applications, all relying on vision data and AI. Ultimately, if businesses are looking to make their processes more efficient, effective, reliable and repeatable, they’re increasingly turning to AI for an assist, and many of the applications require on-site edge computing.
Security and ransomware are continuing to dominate headlines. What technologies can help tackle this problem? How much of the solution is hardware, including Trusted Platform Module (TPM) and the like, versus software?
MK: We believe it’s a combination of the two. OnLogic is more on the hardware side, so TPM 2.0 as well Intel’s PTT are valuable steps toward bolstering security, but they’re only part of the equation. On the software side, you’ve seen the recent conversation around Windows 11 requiring compatible hardware to offer a TPM 2.0 solution, and we expect those kinds of requirements to pop up more and more frequently. The efforts around computer security will never be “complete”, and data integrity and protection have to be considered at every point in the infrastructure.
AMD | ARM | CPU | edge AI | edge server | GPU | industrial edge | Industry 4.0 | Intel | IoT | OnLogic