In its October 12 announcement at the Open Compute Project (OCP) Global Summit, NVIDIA said it had contributed third-generation MGX rack and compute-tray designs to OCP for Vera Rubin-generation AI factories. The company says MGX 1.2 includes rack-scale liquid cooling, a liquid-cooled 5,000-amp busbar for higher-power systems and rack deployments of up to 250 kW. NVIDIA says the shared design is intended to simplify manufacturing and field service.
The announcement also includes open-source DSX software components for network configuration, GPU-cluster testing, rack management and bare-metal deployment. NVIDIA, Google, Meta and AMD published Open Data Center Specification v0.7 for facility design, while Google, Meta, Microsoft and NVIDIA published the OCP Grid Resilience Specification for data-center behavior during utility disturbances. NVIDIA and Cisco also introduced the USFP optical-module form factor targeting 400 Gb/s per lane. These are design and specification details reported by the companies, not guarantees of operating performance.
HABER PAPER / DATA
Behind the numbers
Inspect the source, read the data, make your own analysis.
Data table
Infrastructure specifications in NVIDIA’s OCP announcement
Design limits and targeted link characteristics announced by the company on October 12; these are not measured field results.
Infrastructure specifications in NVIDIA’s OCP announcement
Indicator (English)
Source value and unit
Period or announcement date
Data basis (English)
Numeric value (source scale)
Source unit (English)
Equality or bound qualifier
Maximum rack power stated for MGX 1.2
250 kW
2026-10-12
design specification; not measured power use
250
kW
at-most
Target rate per USFP lane
400 Gb/s
2026-10-12
design target
400
Gb/s
reported-point
Source and methodology
Haber Paper source-linked editorial data compilation; underlying sources retain their separate reuse terms.
Values are transcribed from NVIDIA’s primary announcement; planned design specifications should not be read as realized capacity. Numeric values retain source scales; bounds are distinct from exact equality. Sorting must account for units and indicator scope.
At the BUILD 2026 conference, the communications minister cited power, connectivity, cloud access and advanced chips as constraints on local AI growth.
Source: Voice of Nigeria, reporting the Ministry of Communications, Innovation and Digital Economy
Researchers from Ottawa, Columbia and Hanyang reviewed post-CMOS materials and device approaches for artificial neurons; the work is a literature review, not a new chip prototype.
Google Analytics can measure visits and site activity using cookies if you allow it. You can change your choice later.
Measurement and privacy details
Google Analytics measurement starts only if you allow it. After consent, automatic measurement may cover scrolls, outbound link clicks, downloads, video engagement, site searches and basic form interactions; search terms may be detected from URL parameters. This site’s own code does not send form field values; email redaction in automatic measurement is best-effort, not guaranteed. Our manual page-view event uses a cleaned URL. You can change your choice later. Visits that begin with consent may also send numeric LCP, INP and CLS performance measurements after Analytics loads. A first or later grant starts this on the next page load; withdrawing consent stops these performance reports for this page. The preference cookie lasts six months; advertising measurement and personalization stay off.
Your browser is sending a Global Privacy Control signal. Analytics is off by default; you may explicitly change this choice here.
