HighByte and Databricks Forge Direct Link to Unleash Industrial AI

PORTLAND, Maine – October 14, 2026 — For over a decade, the promise of Industry 4.0 has been stymied by a fundamental plumbing problem. While enterprise IT systems evolved to embrace massive cloud data lakes and sophisticated artificial intelligence, the factory floor remained trapped in a labyrinth of proprietary protocols, legacy machinery, and brittle data pipelines. Today, the industrial software landscape shifts significantly as HighByte announces a deepening of its partnership with data and AI heavyweight Databricks, aiming to finally clear the bottleneck between operational technology (OT) and enterprise IT.

The collaboration centers on the release of HighByte Intelligence Hub version 4.4, which introduces native support for Databricks’ Zerobus connector. This integration allows manufacturers to stream contextualized, low-latency operational data directly into Databricks Delta tables, bypassing the intermediate message brokers that have long complicated industrial data architectures. It represents a maturation point in industrial DataOps, moving away from piecemeal telemetry collection toward a unified, bi-directional edge-to-cloud infrastructure designed specifically to feed enterprise AI agents and digital twin frameworks.

Bridging the OT-IT Chasm Without the Middleman

Historically, industrial enterprises attempting to bring plant floor signals into modern cloud environments have relied on complex, multi-layered architectures. Data pulled from programmable logic controllers (PLCs) or supervisory control and data acquisition (SCADA) systems typically had to be routed through intermediate streaming platforms like Kafka or MQTT brokers. These brokers, while robust, require dedicated compute resources, continuous maintenance, and complex extract, transform, and load (ETL) processes to finally land the data in a usable format for analytics.

The new Zerobus connector fundamentally alters this architecture. By enabling direct streaming ingest into Delta tables, HighByte eliminates the need for these intermediate event streaming services. Furthermore, the connector supports automatic schema evolution—meaning that as data models change on the factory floor, the corresponding Delta tables in the Databricks Lakehouse automatically adapt without requiring manual database administration.

This direct pipeline offers a distinct competitive advantage when benchmarked against rival industrial DataOps platforms. While competitors like Cognite offer powerful data contextualization through proprietary knowledge graphs, they can be resource-intensive to deploy. Solutions like Litmus Edge provide broad device connectivity but often lack the highly optimized, native direct-to-Lakehouse ingestion that the Zerobus connector provides. PTC ThingWorx, a legacy giant in the space, frequently requires complex API layers to achieve similar cloud integrations. HighByte’s approach focuses purely on modeling data at the edge and pushing it directly into the analytics environment where data scientists already operate.

Forging Smarter Steel with Digital Twins

Theoretical architecture is best proven in the crucible of heavy industry. Gerdau, one of the largest steel producers in the Americas, serves as an early enterprise adopter of this unified data pipeline. Steel manufacturing is notoriously complex, characterized by extreme temperatures, massive physical assets, and decades-old brownfield equipment that was never designed to communicate with modern cloud networks.

Partnering with Brazilian industrial IoT distributor PackIOT, Gerdau deployed the HighByte Intelligence Hub to standardize its vast operational data estate. The goal was not merely to collect data, but to contextualize it at the source—tagging raw temperature readings or vibration metrics with machine IDs, production batch numbers, and timestamps before the data ever leaves the plant floor.

"HighByte has helped us simplify the integration of industrial data into Databricks and significantly reduce the time it takes for operational data to become available to our digital applications," said Felipe Amorim, Global Tech Architecture Manager at Gerdau. "By contextualizing data closer to the source and delivering it efficiently into Databricks, we can provide our digital twins with fresh, contextualized industrial data while accelerating the development and deployment of new solutions. More importantly, this creates a trusted industrial data foundation that opens new opportunities to apply AI to our operations at scale."

For Gerdau, this translates to reduced cloud compute costs, faster time-to-market for new analytics initiatives, and a scalable foundation for digital twins—virtual replicas of physical assets that allow engineers to simulate production changes without risking actual downtime.

The Reality of Autonomous Agents in Heavy Industry

While the HighByte and Databricks integration solves a critical infrastructure bottleneck, it also highlights the growing enterprise obsession with agentic AI—autonomous systems capable of making decisions and executing tasks without human intervention. However, the factory floor is not a customer service chatbot environment; the stakes are physical, and errors can be catastrophic.

According to the State of AI Agents 2026 report published by Databricks, only 19% of organizations have deployed AI agents, and mostly to a limited extent. The primary hurdles cited are data governance and data quality. In manufacturing, an AI model is only as safe as the data feeding it.

"HighByte’s integration with Databricks, now extended through native Zerobus connectivity, helps manufacturers unlock dormant factory data by transforming fragmented signals into governed, AI-ready datasets that stream directly into the Databricks Data + AI Platform," explained Shiv Trisal, Global Industrials GTM at Databricks. "This creates a consistent foundation to model cost, quality, and productivity drivers across the network, laying the groundwork for lights-out manufacturing."

Yet, the journey to lights-out operations requires navigating strict industrial frameworks. The widely adopted ISA-95 automation standard defines rigid hierarchies for integrating enterprise and control systems. Traditional industrial control systems are deterministic, guaranteeing specific actions within milliseconds. AI models, conversely, are probabilistic.

"You cannot simply hand the keys of a blast furnace over to a probabilistic language model," noted one industrial automation engineer familiar with such deployments. "The plant floor operates on deterministic control loops where milliseconds matter. AI has to be governed by strict, immutable safety guardrails."

Closing the Loop Between Insight and Action

Because of these safety constraints, the immediate future of industrial AI agents is hybrid. Agents will likely operate at higher levels of the ISA-95 model, providing optimization recommendations, predicting equipment failures, and adjusting non-critical setpoints, while human operators and deterministic PLCs maintain ultimate authority over life-safety and core machine controls.

To even reach this hybrid state, the underlying data must be impeccable. HighByte's role in contextualizing data at the edge ensures that when an AI agent in Databricks analyzes a pressure drop, it knows exactly which valve on which line is affected, what product is currently running, and what the historical failure parameters are.

"HighByte bridges the gap between Databricks' analytical and ML/AI capabilities and the real-time decisions that must happen on the factory floor," stated Tony Paine, Chief Executive Officer at HighByte. "The closed loop between the edge and the cloud is what turns analytics investments into measurable outcomes for manufacturers. That is what our customers are asking for, and it is exactly what this partnership delivers."

By eliminating the architectural friction that has historically separated OT and IT, HighByte and Databricks are providing the necessary plumbing for the next phase of industrial transformation. As data governance improves and edge-to-cloud latency drops to near-zero, the theoretical concepts of digital twins and autonomous manufacturing are rapidly becoming operational realities.