Infinigence unveils three products, including AI infrastructure for up to 100,000 GPUs, TMTPost reports

TMTPost reported on July 30, 2025, that Infinigence unveiled three AI infrastructure products at WAIC 2025. According to the report, the lineup includes Infinicloud, which supports clusters of up to 100,000 GPUs; InfiniCore, for systems with thousands of GPUs; and InfiniEdge, for edge use starting with one GPU.

If the reported lineup becomes available as described, organizations developing and operating AI could have more options for choosing infrastructure that spans large clusters and on-device computing starting with one GPU. However, the excerpt does not confirm compatibility between different environments or the scope of operational support, so buyers will need to verify supported configurations and deployment track records.

The three products stand out for their effort to cover environments with different computing scales in a single lineup. As reported, Infinicloud targets up to 100,000 GPUs, InfiniCore targets thousands of GPUs, and InfiniEdge starts with one GPU, creating tiers from cloud-scale computing to edge devices. The lineup’s evaluation is likely to depend not only on the number of GPUs but also on where models can run and how transitions between environments are handled. The article also describes compatibility with heterogeneous computing environments and support ranging from model scheduling and performance optimization to large-scale application deployment. Under that description, the products’ value depends not only on the performance of individual computers but also on how well they can support operations across different chips and architectures. However, the excerpt does not name specific supported chips, provide performance comparisons across environments or specify terms of use, making the actual scope of compatibility unclear. In a related article, we reported that AMD announced the “AMD Robotics Partner Network” on July 23, 2026, to support the development and deployment of robotics and physical AI systems AMD announces “AMD Robotics Partner Network” to accelerate robotics development. That article covered development and deployment on the robotics side, while this report covers AI computing infrastructure. No business deal or technical collaboration linking the two has been confirmed, but evaluating physical AI deployments requires assessing the development environments for robots and applications separately from the computing infrastructure that runs them. The next points to watch are when the three products will become available, their pricing, the computing resources that can actually be used, supported chips and operating track records. The excerpt provided does not confirm these details. See the original report for more information.