(US and Canada) Aligned, a leading innovator in AI infrastructure and scalable machine learning solutions, today announced a strategic collaboration with AMD and the University of Southern California’s Information Sciences Institute (USC ISI) to accelerate the development and training of MEGALODON, a cutting-edge large language model (LLM). The collaboration centers on harnessing the powerful AMD Instinct™ MI300 GPUs and Aligned’s expertise in AI operations to train the model efficiently on non-NVIDIA architectures.

MEGALODON is USC ISI’s flagship effort in large-scale language modeling, exploring a novel Moving Average Equipped Gated Attention (MEGA) architecture designed to enhance long-context retention while reducing computational complexity. With model training requiring massive GPU resources and complex system orchestration, the collaboration between AMD and Aligned aims to push performance boundaries beyond traditional CUDA-based environments.

“As researchers push the limits of model size and capability, MEGALODON represents a leap forward in architecture and scalability,” said Jonathan May, research associate professor at the Viterbi School of Engineering. “This collaboration brings the computational power and infrastructure expertise needed to realize its full potential.”

Tackling AI at Scale with AMD Hardware and ROCm^TM Open Software

Aligned will play a critical role in optimizing MEGALODON’s training on AMD’s Instinct™ MI325X GPUs, leveraging AMD’s ROCm™ platform as a CUDA alternative. Aligned is providing the tools and engineering support to help enable USC ISI’s researchers to tap into the latest AMD GPU compute infrastructure.

“We’re excited to work alongside AMD and USC ISI to help innovate on large-scale AI training,” said Chris Ensey, CEO of Aligned. “Our mission is to enable powerful AI on any hardware—and this partnership exemplifies that vision.”

World-Class Infrastructure Meets World-Class Research

The project also intersects with ISI’s broader ecosystem of public and private partnerships—including DARPA, NSF, Lockheed Martin, and Chevron—demonstrating the real-world impact of foundational AI research. MEGALODON is being developed and tested using high-performance compute clusters and advanced storage technologies, with future plans to scale with more advanced AMD GPUs such as the MI355x.

Key Collaboration Highlights:

• Novel LLM Architecture: MEGALODON integrates the MEGA attention mechanism, offering improved scalability and long-context performance.
• Cross-Platform AI Enablement: First major effort to train a large-scale model on AMD MI325X via ROCm^TM open software ecosystem.
• Advanced Infrastructure: Joint access to supercomputing clusters, high-throughput storage, and custom chip prototyping labs like MOSIS.
• Funding & Institutional Backing: Supported by NSF grants and USC’s commitment to advancing open, interoperable AI research.