The viral open-source AI project OpenClaw faces a critical identity crisis after weekly downloads plunged nearly 50% from their March peak, as internal divisions grow over whether to prioritize enterprise stability or its founding "hacker spirit."
"To understand OpenClaw, you need to understand its irregular starting point. It was released more like a product than an open-source project," said Noam Schwartz, co-founder and CEO of the network security firm Alice and a long-time OpenClaw user.
The project's rapid, sometimes daily, updates have caused significant issues for users, with many reporting that their custom-built AI agents break after upgrading. More severe incidents include accidental data deletion by agents. The instability is compounded by security risks, with cybersecurity researchers identifying a new "OpenClaw" trojan that has already compromised over 28,000 systems by exploiting the platform's extensive permissions. This has led major corporate users, including Nvidia, to delay upgrading to the latest versions for fear of technical failure.
The internal turmoil and technical instability risk ceding OpenClaw's early lead in the nascent AI agent market. While the project struggles with its direction, competitors are closing in. According to ClawCharts, which tracks open-source AI agent activity, Nous Research's Hermes has recently surpassed OpenClaw in the number of recent GitHub contributors, signaling a potential shift in developer momentum.
The Soul of a New Machine
At the heart of OpenClaw's struggle is a debate over its future path. One faction of its volunteer maintainers advocates for a more structured approach, implementing predictable update cycles and rigorous quality control to cater to the enterprise giants now using their software. They argue this is essential for the project's long-term viability and to prevent users from being "afraid to update," as volunteer maintainer Gustavo Madeira Santana put it.
Opposing them is a group that fears such formal procedures would stifle the very "hacker spirit" that powered OpenClaw's explosive growth. They contend that the AI agent space is evolving too quickly to be constrained by corporate-style roadmaps and that high-frequency iteration is necessary to stay at the forefront.
External Pressures Mount
The project's challenges are not just internal. Anthropic, whose Claude models are popular among OpenClaw users, recently adjusted its pricing, making it significantly more expensive to use its models through third-party tools. "What used to be a $200 per month Anthropic subscription could now become $200 per day," calculated Schwartz. This has forced OpenClaw to work on helping users transition to OpenAI's models.
Meanwhile, the project's founder, Peter Steinberger, joined OpenAI in February. While he reportedly still has significant leeway to guide OpenClaw, his new role and the project's increasing reliance on external donations from companies like Tencent for its newly formed foundation add another layer of complexity to its governance.
A Market in Flux
The instability has not gone unnoticed by developers and venture capitalists. "I don't understand why a non-engineer, non-tinkerer would use OpenClaw at this point," said Darian Shirazi, a general partner at Gradient Ventures, who has switched to a competitor's product citing OpenClaw's "clunky" nature and security risks.
The 50% drop in weekly downloads from their mid-March peak, back to early March levels, quantifies this sentiment. As OpenClaw's heat fades, the question is whether it can adopt the discipline needed for enterprise adoption, perhaps by following the path of Linux with long-term support versions, without losing the innovative spark that made it a phenomenon.
This article is for informational purposes only and does not constitute investment advice.
