Amazon Web Services' share of contribution activities across OpenSearch repositories has dropped from 91% in the project's first year to 72% in 2025–26, according to figures shared with The New Stack by the OpenSearch Software Foundation. The data, disclosed this month alongside interviews with foundation leadership at Open Source Summit Europe, points to a gradual shift in the governance and development of the open source search and analytics platform that AWS created in 2021 as a fork of Elasticsearch. The foundation was established in September 2024 when AWS transferred the project to the Linux Foundation under vendor-neutral oversight.

Self-managed cloud deployments of OpenSearch grew sharply between 2024 and 2026, rising from 37% to 60% among surveyed users, while adoption of Amazon's managed OpenSearch Service climbed more modestly from 44% to 48%, according to the 2026 Open Data Infrastructure Report published last month by the Linux Foundation and the OpenSearch Software Foundation. Engineers affiliated with Amazon made up 95% of maintainers in the core OpenSearch repository in April 2022, compared with 71% today. Yet nearly a quarter of 71 surveyed non-adopters—24%—cited worries about the project's vendor neutrality and historical connections to AWS as reasons for not using the software. AWS still holds four of the 15 voting seats on the project's Technical Steering Committee, and an AWS employee, Andrew Ross, was reelected committee chair in October.

Bianca Lewis, executive director of the OpenSearch Software Foundation, told The New Stack that AWS's decision to donate the project to the Linux Foundation marked "a watershed moment" because the company "couldn't close OpenSearch and they couldn't change the license." She argued that the transfer made commercial sense for AWS, whose cloud business earns revenue from computing resources rather than application ownership. "AWS don't make money from applications," Lewis said. "AWS make money from compute. So operating, growing OpenSearch long term was never in AWS's plan." Lewis acknowledged that early versions of OpenSearch were essentially "Elasticsearch with a different name," sharing the same plugins and much of the same documentation, but said the software "had already well and truly diverged" by the time it joined the Linux Foundation.

The report suggests that a substantial portion of OpenSearch deployments remain experimental, with organizations evaluating the technology or using it for applications that could be replaced. Lewis pointed to the time required for companies to move away from established commercial platforms, particularly when tied to multiyear licensing agreements with vendors like Splunk or Elastic. Enterprises must weigh not only licensing costs but also the engineering effort involved in changing suppliers, moving data, and maintaining applications over time, according to the foundation. The project is positioning itself as infrastructure for applications that need to search, analyze, and retrieve data, including those built around large language models, with support for vector and hybrid search making it relevant to retrieval-augmented generation workflows.

Lewis said her goal for OpenSearch in five years is "to have a truly sustainable, fast, innovating platform, which is not dominated or heavily influenced by any single vendor in the world." She called OpenSearch "the default choice" for organizations that want their AI infrastructure on an open platform, citing improvements in the latest release, OpenSearch 3.9, aimed at helping developers search and retrieve information for AI applications. But the project's ability to attract and retain contributors beyond AWS becomes critical if organizations are choosing the platform partly to avoid dependence on individual vendors, the report indicates. The foundation's figures suggest that convincing prospective users and contributors that OpenSearch's future is no longer determined primarily by AWS remains a challenge that a change in governance alone cannot resolve. For enterprises weighing open source against proprietary alternatives, the calculus extends beyond immediate feature parity to questions of long-term platform risk and the credibility of community stewardship.