News

New Partnerships Push AI to the Edge

• By David Ramel
• 10/07/2024

Two recent partnerships show how advanced AI is moving to the edge.

This movement, similar to "edge computing" in which data is processed closer to its source, sees AI models being deployed and executed on outlying devices, allowing for real-time analysis and decision-making without relying on a central server. In essence, Edge AI is a subset of edge computing, enabling AI applications to operate more efficiently and effectively in real-time scenarios.

Edge AI has been around since dawn of the "AI era," but industry partnerships show how it's advancing as the entire space matures. Such partnerships have long served to indicate where tech trends are heading, and two similar announcements about similar topics in a short time are notable.

Last week, a company called Edgescale AI emerged from stealth to partner with Palantir to launch a new Live Edge solution. A couple weeks earlier, ZEDEDA and Edge Impulse announced a partnership providing end-to-end automation of AI model development, deployment and orchestrating at scale. Here's a look at those two developments.

Live Edge
"The next phase of AI is operating on real-world data," said Brian Mengwasser, the co-founder and CEO of newly emergent Edgescale AI, in a news release. "We've reimagined and reinvented the cloud to encompass physical devices, where data comes from and actions have real-world impact. We eliminate the friction for our customers to deploy the latest AI capabilities anywhere. We're proud to launch this breakthrough first with Palantir, the category leader in production-grade AI."

Palantir, meanwhile, announced the news in a blog post. "In today's rapidly evolving technological landscape, enterprises use centralized cloud for integrating information, improving applications, and curating AI models on flexible infrastructure. However, the full potential of AI remains untapped in physical systems at the edge where infrastructure tends to be rigid or absent. To address this, Palantir and Edgescale AI partnered to extend the value of cloud-like infrastructure and bring AI into the physical edge ecosystem — into the real world."

The platform integrates Palantir's Edge AI capabilities with Edgescale AI's distributed infrastructure, enabling real-time processing of operational data from Internet of Things (IoT) devices in physical systems. This collaboration aims to address the challenges of applying AI in industries where automation and smarter decision-making can transform operations.

At its core, the new Live Edge offering provides a cloud-native solution that automates the creation of Virtual Connected Edges (VCEs), distributed cloud environments that allow AI to directly interact with physical devices, enhancing data flow and decision-making across connected systems. These VCEs eliminate the complexity and friction of deploying AI on-site, making it easier for industries to integrate advanced AI functionalities without the infrastructure overhead typically associated with such deployments.

ZEDEDA/Edge Impulse
These companies teamed up to provide end-to-end automation of AI model development, deployment and orchestrating at scale.

ZEDEDA sells an edge orchestration and management solution delivering applications and workloads to edge devices, while Edge Impulse streamlines the creation of AI and machine learning models for edge hardware, allowing devices to make decisions and offer insight where data is gathered.

Implementing AI at the edge is often complex, requiring collaboration between IT and operational technology (OT) teams. However, with this partnership, customers can reportedly deploy their Edge Impulse AI models via the ZEDEDA Marketplace in a secure, single-click process, alongside other applications, to multiple devices. The system also includes automatic model monitoring, which feeds relevant data back for continuous retraining. Additionally, extensive partner ecosystems are said to offer a variety of technologies to further advance edge computing projects.

"We're witnessing firsthand the explosive growth of AI and data-driven initiatives throughout our customer base and have seen the challenges inherent in delivering and iterating these projects within remote and distributed environments," said Said Ouissal, ZEDEDA's CEO and founder, in a news release. "Edge Impulse's leadership in delivering Edge AI models is a great complement to ZEDEDA's expertise in orchestrating and securing edge applications and devices at scale, and I'm delighted to offer our customers a comprehensive solution to streamline and accelerate their AI transformation efforts."

The companies cited research indicating the adoption of AI/ML at the edge is surging, noting that researchfirm Gartner predicted that by 2026, at least 50% of edge computing deployments will involve machine learning, which is up from 5% in 2022. Gartner also predicted that by 2029, at least 60% of edge computing deployments will use composite AI, compared to less than 5% in 2023. The firm defines composite AI as both predictive and generative AI (GenAI).

Trends & Benefits
Speaking of research, the Cloud Native Computing Foundation (CNCF) recently published "Top 3 edge AI trends to look for in 2024," which states that during this year, "One of the most exciting developments is the convergence of AI and edge computing."

The top three trends are:

• DevEdgeOps: "The significant ascent of edge computing has resulted in a paradigm shift in data processing and utilization. Edge computing, unlike traditional cloud-based methods, places computational resources closer to the data source. This shift in architecture introduces new complexities and opportunities for DevOps."
• AI Interaction across Edge and Cloud: "The future of AI lies in the seamless integration of edge and cloud computing in forthcoming years. AI workloads will dynamically move between the edge and the cloud, leveraging each of their strengths. The cloud will train complex AI models, while the edge will handle real-time inferencing, ensuring fast responses. Next-gen Edge platforms will support end-to-end automation, delivering comprehensive solutions across multi-cloud and edge environments."
• Micro AI: "2024 will see the rise of Micro AI—lightweight, hyper-efficient AI models for edge devices like smartwatches, IoT sensors, drones, and home appliances. These tiny AI brains enable real-time data processing and decision-making without cloud reliance. Key innovations include better algorithms, enhanced energy efficiency, and broader applications. Micro AI helps enterprises in the following ways."

And earlier in the year, Wevolver published its "2024 State of Edge AI Report." It examines Edge AI across different markets along with various other aspects of the movement.

"Today, a lot of data processing is being decentralized from large cloud data centers to smaller localized data centers and edge devices," the report said. "This has enabled the emergence of Edge AI, which processes data at or near the source of data generation. Many organizations are deploying edge functionalities, resulting in energy-efficient, low-latency applications with real-time performance. Edge AI offers significant data protection and security benefits, making it an attractive proposition for organizations across sectors to use edge computing features for various use cases."

Further research was published by Cornell University in July, titled "Edge AI: A Taxonomy, Systematic Review and Future Directions," which said, " Despite being a relatively new field of research, spanning from 2014 to the present, it has shown significant and rapid development over the last five years."

Another paper, "Revisiting Edge AI: Opportunities and Challenges," was published this summer by IEEE, whose abstract reads: "Edge artificial intelligence (AI) is an innovative computing paradigm that aims to shift the training and inference of machine learning models to the edge of the network. This paradigm offers the opportunity to significantly impact our everyday lives with new services such as autonomous driving and ubiquitous personalized health care. Nevertheless, bringing intelligence to the edge involves several major challenges, which include the need to constrain model architecture designs, the secure distribution and execution of the trained models, and the substantial network load required to distribute the models and data collected for training."

With all of the above happening in such a short time, it's clear Edge AI is gaining steam, so stay tuned to see what happens next in the space.