Safety Tips
- Meet in a safe, public place
- Check the item before buying
- Pay only after collecting the item
Description
Looking for custom ASIC design and AI data center solutions that scale? MediaTek delivers end-to-end AI data center infrastructure, from custom silicon all the way to the rack, with every layer optimized for performance, efficiency, and scale.
What MediaTek offers
Custom ASIC design and XPU development: Let MediaTek architect your compute from the ground up, or integrate your own AI accelerators, logic IP, and microarchitectures into one unified system.
Advanced packaging (2.5D and 3.5D): Combine compute dies, memory interface dies, and HBM into a single high-bandwidth package built for AI training and inference.
High-speed interconnect: A full hierarchy of die-to-die, chip-to-chip, board-to-board, and rack-to-rack links, including UCIe, high-speed SerDes, and co-packaged optics (CPO).
Rack-level AI compute systems: Hardware and software co-optimization built around performance per watt and performance per TCO.
NVIDIA NVLink Fusion ecosystem: Connect custom XPUs into NVIDIA's ecosystem for full-stack, rack-level data center solutions.
Next-gen Active Optical Cable (AOC): A proof of concept with Microsoft, built on MicroLED light sources, that targets major power savings and longer reach for AI clusters.
Why choose MediaTek for AI data centers?
Early adoption of leading-edge process nodes, with 10+ tape-outs on N3 and 2+ on N2.
Design-Technology Co-Optimization (DTCO) and System-Technology Co-Optimization (STCO) to get the most out of the silicon.
Mobile-proven low-voltage design experience, applied to data center chips to improve compute efficiency and tokens per watt.
Partnerships with hyperscalers, cloud service providers, and AI platform companies.
Explore MediaTek's Data Center Solutions and talk to the team about your next AI infrastructure project: https://www.mediatek.com/products/data-center
What MediaTek offers
Custom ASIC design and XPU development: Let MediaTek architect your compute from the ground up, or integrate your own AI accelerators, logic IP, and microarchitectures into one unified system.
Advanced packaging (2.5D and 3.5D): Combine compute dies, memory interface dies, and HBM into a single high-bandwidth package built for AI training and inference.
High-speed interconnect: A full hierarchy of die-to-die, chip-to-chip, board-to-board, and rack-to-rack links, including UCIe, high-speed SerDes, and co-packaged optics (CPO).
Rack-level AI compute systems: Hardware and software co-optimization built around performance per watt and performance per TCO.
NVIDIA NVLink Fusion ecosystem: Connect custom XPUs into NVIDIA's ecosystem for full-stack, rack-level data center solutions.
Next-gen Active Optical Cable (AOC): A proof of concept with Microsoft, built on MicroLED light sources, that targets major power savings and longer reach for AI clusters.
Why choose MediaTek for AI data centers?
Early adoption of leading-edge process nodes, with 10+ tape-outs on N3 and 2+ on N2.
Design-Technology Co-Optimization (DTCO) and System-Technology Co-Optimization (STCO) to get the most out of the silicon.
Mobile-proven low-voltage design experience, applied to data center chips to improve compute efficiency and tokens per watt.
Partnerships with hyperscalers, cloud service providers, and AI platform companies.
Explore MediaTek's Data Center Solutions and talk to the team about your next AI infrastructure project: https://www.mediatek.com/products/data-center
Location
Phillaur, Punjab