Former executive of both AMD and Intel graphics, Raja Koduri, appears to be a tenacious creator who doesn't give up on projects with potential. This is evident in his new company, Mihira AI, which has licensed projects that Intel decided to abandon. Specifically, Mihira aims to acquire the intellectual property left behind by Intel's Project Endgame, a network-based solution that aimed to enhance local hardware's graphic capabilities at the push of a button. This information comes from a report by EE Times. Although Intel announced Project Endgame in 2022, development on the feature remained silent, even during the period when it was expected to be available in beta form. However, in July of this year (around four months after Raja Koduri left Intel to join Mihira AI), Intel tweeted that the project was on "indefinite hold". It appears that one person's discarded idea can become someone else's treasure, as Raja and Mihira AI believe that the IP from Project Endgame will be crucial in the company's ambitions to become a software provider in the highly profitable data center industry. Mihira AI aims to build heterogeneous data center architectures that offer acceleration for graphics and AI workloads. As part of their strategy, Mihira AI plans to establish a content creation studio to showcase their ability to manage data center workloads through their software stack. The ultimate goal is to provide an agnostic solution that competes with Nvidia's CUDA and AMD's ROCm.

In this context, heterogeneous workloads refer to coordinating communication and workloads between CPU compute clusters for rendering, accelerators for AI and gaming GPU workloads, and potentially a future fourth cluster optimized for low-power AI inference. Given this objective, it seems that Project Endgame could be valuable to Mihira AI, as the software solution was designed to dynamically allocate additional graphics power when necessary. At the very least, it is one piece of Mihira AI's puzzle, especially once they further develop their own IP on top of the solution. Interestingly, despite Raja's experience in hardware design and fabrication during his time at AMD and Intel, Mihira AI has chosen not to heavily focus on these aspects. Raja explains that it is a pragmatic decision to not prioritize surpassing Nvidia's performance. Instead, Mihira AI believes that the real opportunity lies in lowering barriers to entry and democratizing cloud and network-based computing. From this perspective, previously abandoned projects hold the most potential. Get the latest news and in-depth reviews of CPUs, GPUs, AI, maker hardware, and more straight to your inbox by joining the experts who read Tom's Hardware. With over 25 years of experience, we provide the inside track on enthusiast PC tech news.