The eighth Global Internet Competition attracted 1,347 projects from 31 countries, with Apache RocketMQ and Higress both winning the Open-Source Pioneer Community Award. Our judgment is that AI competition is moving from the model foreground to the system background. What enterprises truly lack is an engineering foundation capable of handling traffic, connecting tools, and controlling costs.
What this is
RocketMQ is a message queue for asynchronously passing data between systems. It originated in Alibaba’s core transaction pipeline, was donated to the Apache Software Foundation in 2016, and now has more than 22,000 GitHub Stars. For AI workloads, RocketMQ 5.5 introduced LiteTopic to support asynchronous communication between model services and agents, while supporting protocols such as MCP and A2A.
Higress is a cloud-native gateway positioned at the network entry point to centrally manage request forwarding and security policies. It grew out of Alibaba’s internal production practice and was open-sourced in 2022. It has expanded from a gateway for websites and microservices into an AI gateway that can proxy multiple foundation-model APIs under one management layer, while offering an MCP service marketplace and the intelligent routing product HiRoute.
Industry view
Both projects deserve credit for focusing on foundational capabilities required to operate AI applications at scale, rather than building a business around a single model. RocketMQ handles communication between systems, while Higress manages entry points and traffic orchestration. Both have years of production validation and real enterprise adoption; they are not merely conceptual demonstrations.
The risks are equally clear: awards and GitHub popularity do not necessarily translate into commercial returns. Open-source projects still carry maintenance costs, and the amount of revenue contributed by enterprise adopters remains unclear. In addition, the 2026 developments cited here come from the projects’ own descriptions, and some capabilities still lack independent operational data. The gateway and message-queue markets are also crowded. Cloud vendors have the advantage of existing product portfolios and may bundle services, so enterprises need to assess migration costs and long-term support capabilities.
Impact on regular people
For enterprise IT: As multiple models and agents coexist, message queues and API gateways can reduce integration and orchestration complexity. But procurement teams should not rely on community awards alone; they must also verify stability, operating costs, and vendor support.
For individual careers: These backstage capabilities may not directly create conspicuous job titles, but they can improve the delivery efficiency of AI projects. People who understand business orchestration, data flow, and system governance may be closer to implementation than those who only know how to call model APIs.
For the consumer market: More reliable model calls and lower infrastructure costs may eventually make applications such as AI assistants and intelligent customer service more stable. In the short term, however, directly noticeable improvements will still depend on model quality and product design.