After working together for several years, FogHorn has joined the IBM Edge Ecosystem to offer joint customers AI-based solutions at the edge.
The partnership includes using FogHorn's AI Edge software capabilities in tandem with the IBM Edge Application Edge Manager, the latter of which runs on Red Hat OpenShift. The solution runs and manages workloads on end devices, such as gateways from Dell, HP and Intel. Instead of sending the data to a date lake in a cloud and then back down, FogHorn runs the data and takes action in the edge devices themselves.
The combined solution is being engineered to run and manage workloads on virtually any edge endpoint, including devices, clusters and servers, gateways and machines supporting RHEL and other Linux operating systems, with Red Hat OpenShift, Podman and other Docker runtimes. Currently, FogHorn is running Docker containers but has plans to run Kubernetes containers as well.
FogHorn’s offerings can also be integrated with the IBM Maximo Application Suite to optimize the performance of physical assets and accelerate the transformation of maintenance, monitoring and reliability options.
"FogHorn really defines the edge as either on the device itself, or on the premise where the data is actually being created," said FogHorn COO Chris Penrose, who was formerly with A&T. "We normalize the data, and then we apply algorithms and machine learning on those data sets right as they're coming off of the machines.
"We are actually running those data sets right there on the far edge by pulling those data sets and applying those models in real time. Then we're able to take action in real time and close the loop in real time."
The type of device determines that amount of compute power that's needed to run FogHorn's edge computing AI software. Penrose said FogHorn is compute agonistic, which means it can run on Intel, Arm, and Nvidia GPUs. Since it's transport agnostic, FogHorn can run across wireline or wireless access technologies, or even without any connectivity at all.
"So think about an oil field," Penrose said. "With pressure sensors andour putting on our platform, we can look at anomalies like leaks and other things right out there in the oil field where there's no cellular connectivity at all. We can actually run our models in real time and take action in real time, even when there is no connectivity."
Penrose said FogHorn is also cloud agnostic, although it has a deeper relationship with IBM than the other major cloud providers. By working with IBM, FogHorn benefits from Big Blue's deep global relationships, but Penrose said the partnership benefits both companies' customer bases, which, in some cases, overlap.
FogHorn can publish the insights it generates at the edge in three different ways. It can push those insights up into the cloud; it can push out alerts to a technician or it can do the closed loop fixes in the devices in real time without needing signoff.
"You're basically putting in machine learning capabilities to act on the information in real time to provide better outcomes," Penrose said.
RELATED: IBM and Palantir join forces on 'low-code' AI data processing for enterprises
FogHorn was founded in 2014 and has raised $72.5 million to date. Its investors include GE, Bosch, Honeywell and Saudi Aramco, all of which are also customers, as well as traditional investors such as Intel.
FogHorn has customers primarily across manufacturing, power and water, oil and gas, renewable energy, mining, transportation, healthcare, and retail. While its partnership with IBM has broadened its reach, FogHorn already has employees in Europe and Asia.
The advent of IoT and 5G services will also be a boon to FogHorn's bottom line. Penrose, who served as president of AT&T’s IoT solutions organization, said FogHorn, which is a member of the Linux Foundation's LF Edge, is in discussions with numerous unnamed telcos.
"With 5G rolling out, there's more and more interest for using FogHorn as the one of those applications that can ride on top of these top of these 5G network to create real value for customers and for the telcos themselves," Penrose said. "Between the private networks that are being deployed, and more IoT devices going out there, we can run down to the individual device at a customer edge. It can also run potentially on a network edge.
"The flexibility that provides as a horizontal platform that can go across many, many industries and it makes it a very interesting opportunity for us to really lean into that (telco) space as well."