Core Viewpoint - AWS has decided to terminate its cloud RAN project, which was initially aimed at providing telecom operators with more options in the RAN market, potentially disappointing those seeking alternatives to traditional hardware providers like Ericsson and Nokia [2][3]. Group 1: AWS's Strategy and Changes - AWS showcased a server equipped with the Graviton3 processor designed for RAN functions but has shifted focus away from physical hardware to a Container as a Service (CaaS) layer that integrates AI solutions [3][4]. - AWS confirmed the termination of its collaboration with Nokia for specific server deployment, emphasizing a pivot towards a broader CaaS approach rather than concentrating on hardware [3][4]. Group 2: Market Dynamics and Challenges - The RAN market has been shrinking, with revenues dropping from $45 billion in 2022 to an expected $35 billion in 2024 due to reduced spending by telecom operators on 5G [4]. - The Open RAN concept has not succeeded in providing alternatives to existing RAN vendors, with Huawei, Ericsson, and Nokia still dominating the market [4][5]. Group 3: Technical Compatibility and Competition - Designing RAN software compatible with Graviton3 is challenging for suppliers, as Ericsson requires a dedicated hardware accelerator for resource-intensive tasks, which AWS's solution lacks [5][6]. - Nokia has opted for a different approach by offloading all Layer 1 functions to a custom chip developed with Marvell Technology, while using Graviton3 for less demanding Layer 2 and Layer 3 functions [6][8]. Group 4: Future Prospects and Client Engagement - Despite AWS's commitment to cloud RAN, its services have not seen widespread adoption, with major clients like EchoStar's Dish Network shutting down their cloud-based RAN initiatives [9]. - AWS's focus at recent events has shifted towards Agentic AI, indicating a potential end to its expansion in the chip sector [9].
AWS放弃了一项芯片计划