NVIDIA Invests USD 3.5 Billion in MediaTek to Build Next-Gen Cloud
NVIDIA Corporation and MediaTek have announced an expanded, multi-generational partnership to build the next generation of artificial intelligence computing platforms. Spanning AI cloud infrastructure, local on-device AI computing, and physical AI in automotive systems, the collaboration unites NVIDIA’s accelerated computing, AI, graphics, and software ecosystems with MediaTek’s leadership in custom silicon, power-efficient System-on-Chip (SoC) design, advanced packaging, and wireless connectivity.
As part of the expanded agreement, NVIDIA has made a landmark USD 3.5 billion investment in convertible bonds issued by MediaTek.
Concurrently, MediaTek will adopt the newly introduced NVIDIA NVLink Fusion platform, establishing a prebuilt, prevalidated design foundation that allows hyperscalers, cloud service providers (CSPs), and frontier AI model developers to build custom accelerators (XPUs) and integrate them seamlessly into NVIDIA NVLink-connected, rack-scale AI factories.
Custom AI Infrastructure: Demystifying the NVLink Fusion Platform
As artificial intelligence workloads grow in complexity and scale, hyperscalers and cloud operators are seeking to deploy semi-custom application-specific accelerators (XPUs) tailored to proprietary model architectures. However, developing custom silicon and integrating it into production data centres requires complex engineering across high-speed SerDes, advanced packaging, memory interfaces, and high-bandwidth interconnects.
The NVLink Fusion platform eliminates the need to build the surrounding infrastructure from scratch by providing a system-prevalidated foundation for multi-die XPU development.
MediaTek will offer NVLink Fusion as a core design platform, allowing customers to focus their engineering resources on core compute logic while MediaTek and NVIDIA manage connectivity, packaging, and rack-scale integration.
The NVLink Fusion architecture integrates three critical interconnect and memory technologies:
NVIDIA NVLink Fusion Chiplet: Connects custom XPUs directly into the NVIDIA NVLink scale-up fabric using optical photonics or electrical interconnects.
NVIDIA NVLink-C2C: Delivers high-bandwidth, energy-efficient die-to-die and chip-to-chip connectivity between custom XPUs, NVIDIA Rosa CPUs, and compatible host processors.
NVIDIA NVHBM: Relocates and customises high-bandwidth memory (HBM) interfaces to maximise memory bandwidth and energy efficiency while dedicating more silicon die area to active compute.
NVIDIA MGX Compatibility: Aligns custom XPU architectures with modular, rack-scale MGX server configurations, compressing time-to-market for enterprise AI factories.
Local AI Supercomputing: Advancing RTX Spark and DGX Spark
Beyond hyperscale data centres, NVIDIA and MediaTek are collaborating across multiple product generations to bring local AI supercomputing to client devices.
Building on their prior work on the GB10 Grace Blackwell Superchip (which connects a Blackwell GPU with a Grace CPU via NVLink-C2C inside NVIDIA DGX Spark developer workstations), the companies are extending this architecture into NVIDIA RTX Spark.
This joint platform combines NVIDIA’s GPU and RTX graphics technologies with MediaTek’s power-efficient Arm-based SoCs, advanced thermal management, high-bandwidth unified memory controllers (supporting up to 128GB), and integrated low-latency wireless connectivity, powering next-generation Windows AI laptops and small form-factor desktop PCs.
Automotive Innovation in the Era of Physical AI
In the automotive segment, the partnership continues to advance scalable platforms for AI-powered, software-defined vehicles (SDVs).
MediaTek Dimensity Auto platforms integrate NVIDIA graphics, AI processing engines, and software stacks to deliver intelligent digital cockpits, real-time driver-assist interfaces, and in-cabin generative AI features.
These cockpits are designed to operate alongside the NVIDIA DRIVE AGX compute platform, creating an end-to-end, high-safety compute foundation for next-generation autonomous driving.
Jensen Huang, Founder and CEO of NVIDIA, underscored how semiconductor convergence empowers customer-driven system design:
“AI is transforming every computing platform, from the world’s largest AI factories to the PC and the car. MediaTek is one of the world’s great semiconductor companies, with exceptional expertise in system-on-chip design, connectivity, leading performance and power efficiency. Together, we’re building platforms that bring NVIDIA accelerated computing to new markets and give customers the freedom to create differentiated AI systems at enormous scale.”
Rick Tsai, Vice Chairman and CEO of MediaTek, emphasised the broad scope of the multi-domain alliance:
“MediaTek and NVIDIA share a vision for making advanced AI computing pervasive across the technology landscape. NVIDIA’s investment strengthens a collaboration that spans cloud AI infrastructure, local AI computing and automotive in the era of physical AI. By combining NVIDIA’s leadership in accelerated computing and AI software ecosystem with MediaTek’s expertise in a diverse AI technology portfolio from edge to cloud, and our leadership position in custom silicon, we can accelerate innovation for our customers.”
By backing MediaTek’s custom silicon and SoC capabilities with a USD 3.5 billion investment and the NVLink Fusion platform, NVIDIA extends its proprietary interconnect architecture across third-party custom accelerators, cementing NVLink as the unifying fabric for the global AI computing ecosystem.
Read More:
AWS moves to accelerate custom AI silicon with NVLink Fusion
Online Vs. Offline: Nehru Place comments on E-commerce business model
Warranty under pressure: Nehru Place channel raises questions
2 Min read