Analyst(s): Olivier Blanchard
Publication Date: September 11, 2025
NVIDIA and MediaTek’s GB10 collaboration blends CPUs, memory, and GPUs into a superchip, sparking speculation around a potential $73 billion acquisition. Regulatory barriers could stall such a move, but the partnership has recently suggested consistent transformative potential.
What is Covered in this Article:
- NVIDIA and MediaTek collaborate on GB10, integrating CPUs, memory, and GPUs in a dual-dielet package.
- GB10’s technical design features Arm v9.2 cores, Blackwell GPU, and up to 128GB LPDDR5X memory.
- Market speculation suggests NVIDIA may consider acquiring MediaTek for $73 billion.
- According to industry reports, regulatory and geopolitical barriers make such an acquisition unlikely.
- The partnership underscores close technological and strategic ties between the duo
The News: NVIDIA and MediaTek have officially teamed up on the GB10 Grace Blackwell Superchip, blending MediaTek’s strength in CPU and memory design with NVIDIA’s GPU technology. Initially expected in July 2025 but now delayed, the GB10 is designed as a tight integration of both companies’ architectures using standard protocols.
The chip packs 20 Arm v9.2 CPU cores, a Blackwell GPU capable of 31 TFLOPs of FP32 performance, and up to 128GB of shared LPDDR5X memory in a dual-dielet 2.5D setup built on TSMC’s N3 process. Sources like TechRadar and TechNews say this chip could hint at what NVIDIA’s future ARM-based consumer chips might look like. There’s also buzz about NVIDIA possibly going after a $73 billion acquisition of MediaTek.
Could NVIDIA’s Collaboration with MediaTek Trigger a $73 Billion Acquisition Bid?
Analyst Take: The GB10 has confirmed the rumored CPU-GPU collaboration between NVIDIA and MediaTek. Although speculation around a hypothetical $73 billion buyout has been gaining steam recently, most reports point out that major regulatory, political, and ecosystem hurdles would make such a move unlikely. Still, the engineering behind GB10 shows how much NVIDIA is leaning on MediaTek’s strengths, making this partnership significant beyond just the hardware itself.
Integration of CPU, GPU, and Memory
GB10 brings together MediaTek’s energy-efficient CPU and memory IP with NVIDIA’s high-end GPU, all made possible through shared interfaces and performance planning.
The solution uses 20 Arm v9.2 CPU cores split into two 10-core clusters, with 32MB of shared L3 cache and dedicated L2 cache per core. NVIDIA’s Blackwell GPU adds 5th Gen Tensor Cores, RTX ray tracing, DLSS 4, and 24MB of extra L2 cache, hitting 31 TFLOPs FP32 and up to 1,000 TOPS FP4 compute.
Using a 256-bit LPDDR5X-9400 interface, the unified memory setup supports up to 128GB and 301 GB/s bandwidth, which scales to 600 GB/s with NVIDIA’s C2C NVLINK. GB10 is a standout example of tightly integrated CPU–GPU design in a single dual-dielet chip.
Strategic Importance of the Partnership
GB10 shows just how valuable MediaTek is to NVIDIA’s strategy: Tapping into MediaTek’s SoC and memory design has allowed NVIDIA to build a versatile superchip—first seen in the DGX Spark platform at the heart of recent professional-grade “AI supercomputer” PCs (like Acer’s Veriton GN100 AI Mini Workstation, the ASUS Ascent GX10, Dell’s Pro Max with GB10, the HP ZGX Nano AI Station, Lenovo’s ThinkStation PGX, and the MSI EdgeXpert MS-C931).
The DGX Spark platform is marketed to AI developers, data researchers, and other professional use cases (not to consumers). On one hand, the platform serves as a testbed for blending hardware and software, giving NVIDIA and MediaTek a chance to fine-tune before possibly moving into laptop form factors. On the other hand, it positions NVIDIA and MediaTek as high-end commercial segment PC players, expanding NVIDIA’s already enviable AI model training footprint from the data center to the enterprise edge, and expanding MediaTek’s PC solutions portfolio beyond Arm-based Chromebooks.
Beyond this specific partnership, both companies have also partnered to create in-vehicle AI cabin solutions for next-generation software-defined vehicles. The collaboration involves MediaTek developing automotive SoCs that integrate NVIDIA GPU chiplets, AI, and graphics IP. These chips are part of MediaTek’s Dimensity Auto platform and are designed to power advanced infotainment systems with cutting-edge graphics, AI, and safety features.
MediaTek’s Kompanio series chipsets also helped the company establish itself in the Arm-based Chromebook market, particularly as Chromebooks Plus push into the AI PC ecosystem with specs echoing the Windows Copilot+ category. MediaTek’s collaboration with NVIDIA is expected to extend to this space, with rumors pointing to a jointly developed Arm-based PC chip for Windows on Arm laptops and gaming handhelds. This partnership could leverage NVIDIA’s graphics strength to compete against integrated graphics solutions from rivals AMD and Qualcomm.
Regulatory and Geopolitical Constraints
The deepening relationship between NVIDIA and MediaTek through these strategic team-ups across several critical market segments is hard to miss, and it isn’t difficult to see why it has sparked rumors of a potential acquisition. However, any potential $73 billion move by NVIDIA to acquire MediaTek would run into significant roadblocks.
For starters, think back to NVIDIA’s $40 billion bid for Arm in 2022: It was shot down due to competition concerns. Although MediaTek isn’t at all the same type of company as Arm, it would likely face similar regulatory and anticompetitive scrutiny. I suspect that the fiercely independent MediaTek and Taiwan’s government might not be keen on the idea. (Note that MediaTek is the island’s third-biggest company, and a key player in its technology fabric.) With MediaTek seen as a national asset, foreign ownership could become a political flashpoint.
My conversations with executives have also thus far not added any fuel to these rumors. For now, at least, it is fair to expect more and deeper strategic team-ups between both companies, but with no acquisition in sight.
Limited System-Level Benefits for NVIDIA
Something else to think about: NVIDIA appears to have expanded its focus from individual chips to full system-level offerings, reaching into areas like cooling and PCB design beyond just the chips themselves. Buying MediaTek might not make sense from that angle, as NVIDIA’s future is tied more to AI infrastructure and data centers than mobile chips. The company already integrates silicon, software, and cooling into its stack, making it less dependent on firms like MediaTek (which still focus primarily on Mobile and device SOCs). So while the GB10 shows how well the two can work together, a full-blown merger may not fit NVIDIA’s broader strategy or a well-integrated partnership.
Nonetheless, NVIDIA looks poised to benefit from MediaTek’s expertise and ability to scale across critical market segments (AI devices, including high-end edge compute, and automotive, especially). MediaTek, for its part, is already reaping the benefits of its increasingly conspicuous association with NVIDIA, in the form of elevated brand perception, strategic validation (particularly in energy-efficient Arm-based SOC design), and expansion into new markets (mainly Automotive, smart edge, and AI PCs).
What to Watch:
- Taiwan’s government stance will be critical in any discussion of foreign ownership of MediaTek, given its ranking behind only TSMC and Foxconn in market capitalization.
- Questions remain over whether GB10’s architecture will extend into consumer-focused N1 Series SoCs, particularly for Windows on ARM devices.
- The rollout through DGX Spark offers NVIDIA an opportunity to validate hardware–software integration before broader market expansion.
- How markets and potential deal stakeholders perceive the rumored $73 billion valuation.
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of Futurum as a whole.
Other insights from Futurum:
NVIDIA Q2 FY 2026 Earnings: Networking Steals the Spotlight and Q3 Ramp Will Be Key To Watch
Is NVIDIA’s Jetson Thor the New Brain for General Robotics?
Qualcomm’s Arm-Based Data Center CPUs To Smoothly Integrate With NVIDIA
Author Information
Olivier Blanchard is Research Director, Intelligent Devices. He covers edge semiconductors and intelligent AI-capable devices for Futurum. In addition to having co-authored several books about digital transformation and AI with Futurum Group CEO Daniel Newman, Blanchard brings considerable experience demystifying new and emerging technologies, advising clients on how best to future-proof their organizations, and helping maximize the positive impacts of technology disruption while mitigating their potentially negative effects. Follow his extended analysis on X and LinkedIn.
