A sizzling potato: Samsung is not doing any main GPU-related initiatives proper now. Even the Exynos Arm SoC utilized in Galaxy telephones makes use of AMD’s RDNA structure for graphics-related duties, however modifications are coming as the corporate begins throwing cash into GPUs.
BusinessKorea just lately reported that Samsung Electronics accepted a brand new “GPU Funding Proposal” earlier this 12 months. Samsung’s administration committee ratified the proposal in March, marking the corporate’s first vital funding in a GPU-related venture in years. The place Samsung will focus these new improvement efforts remains to be a subject of hypothesis and unconfirmed theories.
The proposal marks the primary vital funding in GPU expertise since 2012. Samsung is probably going attempting to turn out to be a aggressive participant within the ever-growing market. The corporate is already the world’s largest reminiscence producer, making HBM chips for high-performance GPUs developed by third-party firms like Nvidia.
Though there are not any particular particulars in regards to the funding, Samsung’s departure from its conventional targets within the chipmaking enterprise is properly in keeping with right now’s newest market tendencies. The AI enterprise is in full swing, and GPUs are probably the most sought-after merchandise for Large Tech firms like Microsoft and smaller ventures attempting their luck with this new generative AI gold rush.
Nvidia, the corporate presently making essentially the most highly effective and succesful GPU and AI accelerator merchandise in the marketplace, is now essentially the most helpful firm on the earth. Nevertheless, Samsung could possibly be extra all in favour of enhancing its manufacturing processes reasonably than making a brand new GPU within the conventional sense.
The Korean electronics superpower’s GPU funding would possibly speed up the creation of particular digital twins, making an AI-based digital duplicate of a totally automated semiconductor plant. Samsung has already outlined a plan prior to now and is working in partnership with Nvidia to fulfill its digital twin goal by 2030.
One other potential software of Samsung’s GPU-related funding is the corporate’s Excessive-Efficiency Computing (HPC) Middle, just lately constructed on the Hwaseong campus. The HPC facility took three years to finish and now hosts large-scale servers and community tools required for semiconductor design. Samsung might purchase a big chunk of GPU {hardware} to fill the plant, utilizing the newly constructed capabilities for improved analysis efforts or different AI-related duties.