Intel’s next wave of compact, high-performance computing for industrial and edge deployments is starting to arrive. After being shown off during Intel Tech Tour 2025, Panther Lake “Core Ultra Series 3” processors are now appearing in new ready-to-integrate modules built specifically for Edge AI and embedded systems—bringing workstation-class features to space-constrained designs.
One of the first companies out with product announcements is Congatec, which has introduced fresh COM-HPC and COM Express modules powered by Intel Core Ultra Series 3 “Panther Lake” CPUs. These modules are designed in small, “passport-sized” formats that help developers build smaller systems without sacrificing performance, connectivity, or long-term support.
On the CPU side, the modules support Intel Panther Lake-H chips with configurations scaling up to 16 cores, topping out at options such as the 388H. The baseline thermal design power is set at 25W, but system builders get a wide tuning window—allowing the same module family to be configured from 15W up to 65W depending on performance targets, cooling capacity, and use case.
Memory options are one of the most eye-catching parts of the lineup, giving embedded designers multiple paths depending on capacity needs, serviceability, and bandwidth goals:
Two module variants come with onboard LPDDR5X. The conga-MC1000 supports up to 32GB running at up to 8533 MT/s, while the larger conga-HPC scales dramatically higher with up to 96GB of LPDDR5X at similar speeds.
There are also LPCAMM2-based options, offering up to 96GB of LPDDR5X memory operating between 7466 and 8533 MT/s. This approach blends compact design with a modular memory format, which can be attractive for embedded deployments that still want some level of upgrade or configuration flexibility.
For designs that prefer traditional sockets, a SO-DIMM configuration is available with two slots supporting up to 128GB of DDR5 memory at speeds up to 7200 MT/s.
For Edge AI workloads, Congatec highlights significant AI acceleration capabilities. These Panther Lake “Core Ultra Series 3” modules are rated for up to 12 Xe3 GPU cores and up to 120 TOPS of AI compute, including a 50 TOPS NPU. That combination is aimed at enabling real-time inference and on-device intelligence in environments where cloud latency, bandwidth, or privacy requirements make local AI processing the better choice.
Display and deployment features also point to industrial and advanced embedded targets. The modules can drive 3 to 4 independent 6K displays, useful for control rooms, medical imaging stations, digital signage walls, and multi-monitor operator consoles. They’re also designed to function in extreme conditions, with operating temperature support spanning from -40°C to 85°C—an important requirement for outdoor enclosures, transportation, factory floors, and rugged edge installations.
Long lifecycle availability is another key embedded requirement, and Congatec is positioning these modules with a 10+ year availability cycle. That matters for industrial designers who need consistent sourcing, stable validation, and predictable refresh planning over many years.
With Intel Panther Lake now moving beyond demos into modular compute products, the Edge AI and embedded market is set to gain more compact system designs capable of high compute density, flexible memory configurations, robust display output, and wide power tuning—without giving up the long-term availability that industrial deployments depend on.
