Processing IoT and operational technology (OT) data near its source has long been a key objective for IT leaders. With advancements in technology, AI inferencing is becoming increasingly feasible at the network edge. This development is set to revolutionize applications that demand real-time responses to sensor data. According to Gartner’s analyst, Thomas Bittman, the rapid expansion of edge data, combined with the necessity for businesses to leverage AI for gain, indicates significant growth for edge AI.
Gartner projects that by 2028, over 66% of enterprise-managed data will be generated and processed outside traditional data centers or the cloud. By 2029, more than two-thirds of enterprises will adopt edge AI, a considerable increase from the current rate of just 10% in 2025. Similarly, IDC forecasts that half of all enterprise AI inference workloads will be conducted on endpoints or edge nodes by 2030, driven by robust investments in edge IT across industries like healthcare, finance, and manufacturing.
Several factors are driving the push toward edge AI:
-
Data Growth: With nearly 11.7 billion IoT devices connected in 2025, generating data that often goes unprocessed, leveraging this data is crucial. Bittman highlights that up to 90% of edge data is not utilized, resulting in pressure to filter and process it locally.
-
Data Control: In regions such as Europe, strict regulations around data privacy and residency are motivating organizations to process data at the edge. Organizations are increasingly concerned about controlling sensitive data, particularly biometric information.
-
Latency: Many applications require real-time data processing, and maintaining on-premise data reduces issues related to connectivity and transfer delays, which are critical in safety-critical scenarios.
-
Cost Efficiency: Transporting data to the cloud incurs significant costs. Edge AI allows businesses to minimize these expenses while accommodating real-time applications that necessitate local inference.
The technological landscape enabling AI at the edge has also evolved. Improvements in chip technology, such as neural processing units (NPUs), have made it feasible to process AI models on-site. These NPUs, including Google’s Tensor Processing Units and Qualcomm’s Snapdragon, are designed for efficiency, delivering immense processing capabilities without the excessive energy and heat produced by typical CPUs or GPUs.
Moreover, innovative chips that emulate brain functioning are emerging, allowing for efficient data processing by activating only when critical events occur. AI accelerators tailored for edge deployments are becoming mainstream, alongside new compact language models that maintain strong performance while operating at a smaller scale.
Edge AI has begun permeating various industries:
-
Manufacturing: Automation systems can perform real-time quality control and predictive maintenance, increasing efficiency and safety on the production floor.
-
Retail: Intelligent cameras are utilized for inventory management and theft detection.
-
Agriculture: Technology like drones and robotics enhances productivity by automating tasks such as crop management.
-
Safety Applications: Smart streetlights equipped with cameras can alert drivers of pedestrians in crosswalks, improving safety.
-
Healthcare: AI models assist in surgical procedures, while wearable technology monitors patient health metrics.
-
Mobility: Advanced driver-assistance systems leverage AI for lane monitoring and emergency response.
Adopting edge AI requires adapting to a distinctly different infrastructure than traditional data center setups. Lightweight and diverse technologies characterize edge systems, which may include packaged platforms, colocation services, or offerings from telecommunications companies.
Looking ahead, the edge AI sector is projected to grow at an impressive rate of 37% through 2030, indicating a significant shift in enterprise strategies as organizations seek to harness this technology. While challenges exist related to deployment complexity, software integration, and capital expenses, the overall momentum toward edge AI adoption is undeniable, with companies exploring its vast applications across numerous sectors.
