← 목록으로
IT/기술

NVIDIA Hands Its Vera Rubin Rack Designs To OCP, Stripping Compute Trays Of Every Cable, Hose, And Fan

· 읽기 20분

🔗 원문 전체 보기 → Wccftech

NVIDIA Hands Its Vera Rubin Rack Designs To OCP, Stripping Compute Trays Of Every Cable, Hose, And Fan

요약

NVIDIA has announced that it is contributing its Vera Rubin platform to OCP, driving the advancement of AI factories. NVIDIA's OCP Announcement Sees Opening Up of The Vera Rubin Platform To More Partners NVIDIA has made a lot of updates to its VR200 "Vera Rub…

본문

AI Hardware Industry NVIDIA Hands Its Vera Rubin Rack Designs To OCP, Stripping Compute Trays Of Every Cable, Hose, And Fan Hassan Mujtaba • Oct 12, 2026 at 11:00am EDT Comments Add to Preferres Sources
A slide titled 'Announcing NVIDIA Contributes Vera Rubin MGX Innovations' features elements like a 'Cableless, Hoseless, Fanless Tray,' 'Liquid Cooled Busbar,' and 'In-tray MQD/MQDB.'

NVIDIA has announced that it is contributing its Vera Rubin platform to OCP, driving the advancement of AI factories.

NVIDIA's OCP Announcement Sees Opening Up of The Vera Rubin Platform To More Partners

NVIDIA has made a lot of updates to its VR200 "Vera Rubin" platform versus the GB300 "Grace Blackwell Ultra" platform. The company's OCP announcements include:

Related Story NVIDIA DLSS 5 Runs 30% Faster Now, Reduces VRAM Use By 20% & Coming To Hogwarts Legacy & Where Winds Meet In November

● NVIDIA and hyperscalers published open standards for next-generation AI factories,
including AI factory design standards and grid resilience.
● NVIDIA launched new open source software components for NVIDIA DSX AI Factory.
● Meta and NVIDIA advanced open AI networking with NVIDIA Spectrum-X Ethernet
SuperNIC and Spectrum-6 innovations.
● Visionbay introduced an 800V DC AI factory design for the Vera Rubin era.
● NVIDIA and Cisco announced the Ultra-Dense Small Form Factor Pluggable (USFP)
standard, the industry’s first module for 400Gb/s-based networking fabrics.
● NVIDIA and Google announced Falcon Transport support on future NVIDIA ConnectX-
10 SuperNICs and BlueField DPUs.

For starters, the compute tray has no cables, no hoses, no fans. Everything is liquid-cooled and modular, and that turns hours into minutes for the tray assembly, speeding up the serviceability and also the overall rack build times. Additionally, liquid cooling is added to the busbar, enabling 5,000 amps on the liquid-cooled busbar without making the rack bigger. This brings more compute density, more power density in the same space.

A digital render of a server room displays three screens showing a data center, a map interface, and technical specifications, with visible computer hardware components below.

And another important contribution that NVIDIA has for this year with OCP is the MQD. The acronym is MGX Quick Disconnect. And there's also an MQDB. The B stands for blind mating. Both of these are new, and there are small liquid cooling connectors inside the tray. And it means that you can take a cold plate and swap it out with easier serviceability and without concern for leaks. These are some of the many contributions to the OCP MGX 1.2 specification.

NVIDIA is also announcing that Vera is shipping to market in multiple form factors. Vera is built upon the same 3rd Gen MGX rack, which means the same architecture and design principles that unite the compute, networking, power, and cooling into one modular platform.

A presentation slide titled 'NVIDIA Full-Stack AI Factory Platform' displays components like Vera Rubin NVL72 and Groq 3 LPX under '3rd Gen MGX Rack Architecture' alongside software such as 'Dynamo' and 'TensorRT-LLM.'

And that means that partners can deploy custom Vera CPU infrastructure while still benefiting from a shared ecosystem and one consistent design. So, this allows them to be highly customizable for any customer need and use cases such as inference, agentic, storage, and more.

The company also shared some performance aspects of its Vera CPU architecture, and its tailor-made nature aimed at Agentic AI. In real-world deployments, partners running Vera are seeing a big gain in performance.

Meta has measured 2x performance for Hyperscale Cloud using DCPerf. The Texas Advanced Computing Center has run the HPC GoldenSuite, and they're seeing a 2x uplift, and Red Panda has measured 3.5x better streaming data performance on Apache Kafka.

NVIDIA announces the 'Vera CPU MGX Rack' now available, showcasing customizable trays for faster assembly and ease of servicing, with logos of partner companies like Dell Technologies, HPE, ASUS, and Lenovo.

In addition to the Vera and Rubin, Meta is also adopting NVIDIA's Spectrum networking ASIC, which enables the first use of a 102.4 Tb/s Ethernet ASIC. This is being deployed by Meta in its next-generation AI factory infrastructure, which connects Meta's Vera Rubin deployments.

Spectrum 6 also lets Meta build Spectrum X multi-plane networks. With multi-plane networks, when a component fails, the SuperNIC is going to detect that in hardware, and it's going to reroute the traffic, and do that in less than 3 milliseconds. So, that's 30 times faster than alternatives, which are all software-based.

MetaSwitch, which is called NicePack, runs on OpenStack, & it's built from the OCP contributions for that stack, including the FBOS operating system that was contributed by Meta themselves, and the switch abstraction interface, an OCP standard that NVIDIA helped to develop.

A gray unbranded server with multiple front cooling vents and visible network ports is placed on a black background.

NVIDIA has also announced that Google's Falcon Transport is supported for the Spectrum X Ethernet. This will be on future ConnectX 10 SuperNICs, and there's a roadmap extending to Bluefield.

At OCP, Visionbay unveiled plans for K-1, a next-generation AI factory platform designed around an 800 VDC architecture and built with future NVIDIA Vera Rubin deployments in mind. The project combines NVIDIA's 800 VDC reference architecture with power and thermal technologies from Vertiv to create a scalable foundation for high-density AI infrastructure.

The design addresses a growing industry challenge. As AI training, inference and reasoning workloads expand, rack power densities continue to rise. Traditional low-voltage power distribution becomes increasingly difficult to scale, requiring more copper, more cabling and more power-conversion equipment.

By adopting NVIDIA’s 800 VDC architecture, K-1 creates a path toward higher-density AI infrastructure while preserving flexibility for existing deployments. Rather than requiring a complete transition on day one, the platform is designed to phase in 800 VDC distribution where it delivers the greatest benefit, helping operators prepare for future generations of accelerated computing without extensive facility redesign.

The architecture is designed around cluster-level power centers that centralize AC-to-DC conversion and distribute 800 VDC closer to compute resources. The approach simplifies downstream power delivery, supports modular expansion and creates repeatable building blocks for future AI factory growth.

NVIDIA’s OCP contributions open the Vera Rubin platform and third-generation MGX designs, delivering cable-free liquid-cooled trays, higher power density, and modular serviceability while shipping the Vera CPU purpose-built for agentic workloads with up to 2× performance gains. Together with Meta’s adoption of Spectrum 6, Falcon Transport support, and the new high-density USFP optical module co-defined with Cisco, these open standards accelerate AI factory deployment worldwide.

By contributing at every layer of the stack and growing the MGX ecosystem to nearly 100 partners, NVIDIA helps the industry scale faster, innovate collaboratively, and unlock the full potential of agentic AI.

Hassan Mujtaba Photo

About the author: A Software Engineer by training and a PC enthusiast by passion, Hassan Mujtaba serves as Wccftech's Senior Editor for hardware section. With years of experience in the industry, he specializes in deep-dive technical analysis of next-generation CPU and GPU architectures, motherboards, and cooling solutions. His work involves not only breaking news on upcoming technologies but also extensive hands-on reviews and benchmarking.

Follow Wccftech on Google to get more of our news coverage in your feeds.

Deal of the Day

Further Reading

NVIDIA RTX 5080 24 GB GPU Expected To Arrive In Q1 2027 As The New Flagship Blackwell GeForce GPU

A close-up of an NVIDIA component with visible branding, featuring a metallic design and PCIe connection.

NVIDIA And Cisco Cram 4x OSFP’s Bandwidth Density Into An OSFP-Sized Module, Pushing AI Fabrics To 12.8 Tb/s With USFP

Four NVIDIA graphics cards are installed in a computer setup with visible branding and heatsinks.

NVIDIA’s 9-Year-Old Tesla V100 Still Packs Some Inference Muscle; 4 GPUs On An NVLink Board Can Outperform M5 Ultra, DGX Spark In Various AI Benchmarks By Up To 230%

A hand holding an AMD RDNA chip and an SK hynix GDDR7 chip against a background with the number 5.

AMD Rumored To Launch RDNA 5 In Second Half Of 2027; Also Quashes Rumors Of Next-Gen Using 4 GB GDDR7 Memory

Read all comments on NVIDIA Hands Its Vera Rubin Rack Designs To OCP, Stripping Compute Trays Of Every Cable, Hose, And Fan

Trending Stories

Popular Discussions

출처: Wccftech

작성자: Hassan Mujtaba

원문 게시일: 2026-10-12T15:00:02Z

이 글은 NewsAPI를 통해 자동으로 수집된 콘텐츠입니다. 원문은 위 링크를 통해 확인하세요.

Technology Cybersecurity Cloud IT Wccftech
← 목록으로