As semiconductor and electronics systems generate growing volumes of IoT, log, metric, trace, and sensor data, the OpenSearch Foundation positions OpenSearch as an open data infrastructure layer for AI applications, search, observability, and security monitoring.
In an exclusive conversation with EE Times, Bianca Lewis, executive director of the OpenSearch Foundation at the Linux Foundation, described OpenSearch as an open data layer within AI infrastructure that can support multiple workloads while allowing users to choose their preferred infrastructure and service providers. “Every business produces data, and that data needs to exist somewhere because it is only useful if it can be accessed and utilized. OpenSearch is the layer that makes that data usable,” Lewis said.
That aligns with the Linux Foundation’s broader view of the AI infrastructure stack. Arpit Joshipura, general manager of networking, edge, and IoT at the Linux Foundation, told EE Times that the Linux Foundation increasingly views AI as a layered stack. Projects such as PyTorch and Linux Foundation AI & Data form the intelligence layer, with agentic and domain-specific application layers built on top.
Hybrid search for industrial workloads
AI is making modern search more effective by combining multiple search techniques to improve the relevance of results. Traditional keyword searches often fall short when users make spelling errors or use unexpected wording.
Lewis said that semantic search, powered by vectors and AI, allows systems to identify related meanings and return relevant results even when the exact keywords are absent.
She said semiconductor and electronics companies are well positioned to benefit from OpenSearch because of the volume and diversity of data they generate.
“Every device, every chip generates its own data,” she said. “Beyond that, they [devices] generate different types of data, including IoT data, log data, metric data, trace data, and sensor data.”
OpenSearch can ingest and unify these data types, providing engineers with a consolidated operational view. The platform enables organizations to analyze traces, metrics, and logs together, allowing them to understand system behavior from multiple perspectives using a unified data view.
Lewis illustrated the concept with a simple example. “If you are using only keyword-based search, you may receive incorrect results or no useful results at all,” she said. “With semantic search, powered by vectors and AI, the system understands which terms are closest in meaning.”
However, semantic search alone is not always suitable for industrial and enterprise environments that require exact identifiers, part numbers, or production-line references.
To address this, OpenSearch supports semantic, lexical, and hybrid search approaches.
“In OpenSearch, hybrid search can first filter results using keywords and then apply semantic search,” Lewis said. “This ensures that specific numbers and critical terms remain accurate while still delivering relevant results.”
The platform also supports monitoring AI applications for security, safety, and performance, and can observe AI agents with contextual information.
Lewis said this capability is moving organizations beyond traditional troubleshooting toward defining service-level objectives, performing acceptable error analysis, and achieving cost visibility by integrating cost data into traces and logs.
AI agents create new infrastructure demands
Lewis said the rapid growth of AI agents is changing infrastructure requirements across industries.
Previously, users typically executed queries one after another. Today, AI agents can execute 100,000 queries within a minute.
“If there is a security vulnerability or an operational issue, organizations no longer have time to address everything manually,” Lewis said. “They need to manage search at a scale and speed that reflects modern technology environments.”
Lewis said OpenSearch addresses this challenge by supporting large-scale search workloads while providing monitoring and governance capabilities from the initial deployment stage. These include visibility into AI agents, traces, and historical activity.
Joshipura said that the shift reflects a broader change in AI deployment, with a growing focus on inference rather than model training. He added that the Linux Foundation views AI as a layered stack built around intelligence, agentic, and domain-specific application layers.
“What you are really looking at is inference,” he said. “Training depends on the data that is provided to the model. For inference, organizations can configure systems in the way they want, using combinations of public and private data.”
Alongside performance concerns, organizations are increasingly focused on compliance and data sovereignty as regulations evolve across India, Europe, and the U.S.
To support those requirements, the OpenSearch Foundation provides long-term support versions, security scanning across all 152 repositories, and a library of the software bills of materials (SBOMs).
While acknowledging the presence of other open-source technologies, including ClickHouse, Elasticsearch, and Apache Solr, Lewis drew distinctions between their operating models and OpenSearch’s community-governed approach.
“There have always been open platforms available for use cases such as search and observability,” Lewis said, specifying that OpenSearch allows organizations to move between vendors and deployment models without becoming locked into a proprietary ecosystem.
She acknowledged that some platforms follow an open-core model under vendor ownership, while others were developed before cloud-native architectures became dominant.
OpenSearch remains free to use, although organizations must still fund the infrastructure on which it runs, whether in a data center, cloud environment, self-hosted deployment, or managed service.
India emerges as a strategic market
India has become one of the most active OpenSearch communities globally, according to Lewis. Rather than viewing India solely as a market for adoption, Lewis described the country as both a consumer and contributor to the project.
“When organizations consume and use OpenSearch, they often become contributors over time,” she said. “Companies that build on OpenSearch and depend on it for their business operations want greater control and often wish to add capabilities and features that support their specific requirements.”
As a result, organizations frequently contribute enhancements back to the project, creating a cycle of adoption and development.
Looking ahead, Lewis said she expects continued growth in the adoption of OpenSearch as a data infrastructure layer, particularly in India.
Read also:
Model Context Protocol Emerges as a Common Framework for Enterprise AI Systems
SUSE Extends Single-Kernel Linux Strategy from Edge to Data Center
Leave a Reply
You must Register or Login to post a comment.