AMD partners with GlobalFoundries for Co-Packaged Optics in Instinct MI500 AI accelerators

AMD Eyes GlobalFoundries Co-Packaged Optics for Instinct MI500 Accelerators

AMD is reportedly preparing to expand its AI hardware strategy with co-packaged optics for the Instinct MI500 generation, a move that could bring GlobalFoundries back into the picture in a much more visible way. The reported plan centers on using photonic integrated circuits manufactured by GlobalFoundries, while ASE would manage packaging and assembly. Rather than focusing only on raw accelerator performance, the goal appears to be improving how future Instinct platforms move data at scale. That shift makes sense. AI infrastructure is increasingly constrained by interconnect bandwidth, power efficiency, and latency rather than compute alone. Traditional copper links remain practical, but they become more difficult to scale as system bandwidth requirements keep rising. Signal integrity, power draw, thermal load, and physical routing all become tougher to manage in dense clusters. Co-packaged optics aims to address those issues by moving optical communication much closer to the compute package, reducing electrical bottlenecks and making large multi-node systems easier to scale. For AMD, that could be an important step for the Instinct MI500 family, which is expected in 2027. The company's nearer-term roadmap remains centered on the MI400 series, but the follow-up generation gives AMD room to introduce broader platform-level changes. CPO could help increase usable bandwidth between compute nodes while lowering the power cost of data movement across racks and larger AI deployments. In hyperscale environments, that matters just as much as adding more matrix throughput or memory capacity. The GlobalFoundries angle is also notable for historical reasons. The company originated from AMD's former manufacturing operations before becoming independent, and AMD fully divested years ago. Since then, the relationship has continued in more limited form, mainly around mature-node products. A return through silicon photonics would not signal a broader foundry shift away from advanced logic suppliers, but it would highlight how AMD is selecting specialists for specific functions in the AI supply chain. GlobalFoundries has been investing in silicon photonics and specialty process technologies, which makes it relevant for optical interconnect work even without competing at cutting-edge GPU process nodes. ASE's role in the packaging side is equally important. Advanced AI products are no longer defined by a single die or even a single manufacturing source. Packaging, chip-to-chip connectivity, and interconnect integration now play a major role in system capability. Pairing photonics production from one supplier with packaging expertise from another reflects how modern accelerator design is evolving into a multi-company engineering effort. AMD is not alone in moving this way. NVIDIA is also pursuing optical interconnect technology for future AI platforms, including work tied to Rubin Ultra. That puts AMD's reported MI500 plans into a wider industry context. The next phase of AI hardware competition will depend not only on faster silicon, but also on how efficiently vendors can connect large numbers of accelerators in real deployments. If this partnership takes shape as reported, Instinct MI500 may become one of AMD's most important platform transitions yet, with optical interconnects helping the company address the scaling limits that now define the AI data center race.

AMD Taps GlobalFoundries for MI500's Co-Packaged Optics as the Silicon Photonics Race With NVIDIA Heats Up

AMD will be leveraging GlobalFoundries for the development of its MRM Co-packaged Optic solution for the next-gen Instinct MI500 AI accelerators. CPO or Co-Packaged Optics (Silicon Photonics) is the next-generation solution that reduces reliance on copper and harnesses light to transfer signals. These CPOs are packaged alongside hardware accelerators such as GPUs and will be a key solution for next-gen AI factories, offering improved interconnect latency and creating high-bandwidth connections between CPU and GPU. Both AMD and NVIDIA will be leveraging these technologies for their next-gen AI GPU accelerators. Based on the latest information, AMD will be developing an MRM-based CPO solution for its next-gen Instinct MI500 accelerator. The PIC or Photonic Integrated Circuits will be handed to GlobalFoundries for manufacturing, while ASE will be responsible for the packaging. Last year, AMD acquired photonics experts, Enosemi, to accelerate its Co-Packaged Optics innovations. Similarly, NVIDIA is also said to be working on its own CPO PIC for Vera Rubin accelerators, using TSMC for the PIC and packaging handled by SPIL. The chips themselves will be assembled at Foxconn Industrial Internet or Industrial Fulian, a subsidiary of Foxconn. For Rubin Ultra, the CPO solution will be prioritized over NPO (Near-Package Optics). Moving forward, NVIDIA is going to fully adopt Co-Packaged Optics, removing the need for NPOs with its Feynman generation of AI accelerators. As for AMD, the company has already confirmed that its MI500 series will be fabricated on an advanced 2nm process technology. The upcoming MI400 series makes use of a 2nm process technology too, but the one utilized by MI500 is more advanced, & offers various enhancements. The chips will be produced at TSMC. The MI500 accelerators will leverage the new CDNA 6 architecture (CDNA 5 for MI400 series), & HBM4E memory, which will offer even higher speeds and memory bandwidth than 19.6TB/s on HBM4-powered MI400 accelerators. Also, unlike previous reports, it looks like AMD isn't going to change to the UDNA architecture naming for Instinct GPUs. AMD has promised a big jump in AI performance with the Instinct MI500 series. The company is on the trajectory to achieve over 1000x AI performance in just four years. This is crucial to meet the growing AI demand and keep up with the competition, which is also accelerating at a rapid pace. MI500 launches in 2027.