# Artificial Intelligence, Data Center, High-Performance Computing, Networking Published articles for Artificial Intelligence, Data Center, High-Performance Computing, Networking. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## Cornelis lands $205M to make AI networks compute, not just connect DevFeed: [Cornelis lands $205M to make AI networks compute, not just connect]() Original publisher: [Read original article]() Author: Sean Michael Kerner Published: 2026-09-14T21:13:08Z Content type: news Language: en Sources: [Network World]() Topics: [Artificial Intelligence](), [networking](), [Networks](), [InfiniBand](), [Ethernet](), [intel]() Tags: [ai](), [artificial-intelligence](), [artificial-intelligence-data-center-high-performance-computing-networking](), [compute](), [connectivity](), [cornelis-networks](), [data-center](), [ethernet](), [high-performance-computing](), [infiniband](), [latency](), [low-latency](), [networking](), [networks](), [scale-up]() ### AI overview Cornelis Networks raised $205 million and introduced Active Compute Fabric, an architecture designed to put programmable compute inside networks for AI and high-performance computing workloads. Its CN6000 platform supports both Ethernet and Omni-Path modes while retaining features intended to improve congestion management, latency, and message rates. ### Source excerpt Networks built for AI and HPC workloads have mostly focused on moving data faster between endpoints. That leaves compute idle while packets travel, a bottleneck vendors have tried to address mainly by adding bandwidth rather than changing what the fabric itself does. Cornelis Networks is taking a different approach. The company raised $205 million to fund its expansion into scale-up networking and introduced Active Compute Fabric, an architecture that puts programmable compute inside the network rather than using it only to connect endpoints. Cornelis emerged from Intel in 2020, when the team behind Intel's Omni-Path architecture spun out as an independent company. Omni-Path is a proprietary network connectivity protocol that provides a competitive alternative to InfiniBand and Ethernet. In 2025, the company shipped its first generation of standalone products, the CN5000 platform. Cornelis Networks has since continued to develop its own SuperNICs, switches and software built for AI and HPC workloads. "I want Cornelis to be synonymous with game-changing technology that improves AI efficiency and eases some of the burdens that we're facing in the world today about keeping it funded and responsibly so," Lisa Spelman, CEO of Cornelis Networks, told Network World. The 'path' forward for enabling AI Cornelis built its first generation of independent products around Omni-Path exclusively. That architecture requires both Cornelis SuperNICs and switches end to end. The company's second generation, the 800 gigabit CN6000, adds Ethernet support without giving up the proprietary architecture that differentiates its products. Spelman describes the CN6000 SuperNIC as multimode silicon capable of running either protocol. "You can use it in Ethernet mode, like RoCEv2 Ethernet mode, or you can use it as OmniPath end to end with the OmniPath switch," she explained. "It's one piece of silicon, both modes contained in it." In Ethernet mode, the CN6000 can pair with switches from other ## The race to 1.6T: Ethernet and coherent optics tackle AI's bandwidth crunch DevFeed: [The race to 1.6T: Ethernet and coherent optics tackle AI's bandwidth crunch]() Original publisher: [Read original article]() Author: Sean Michael Kerner Published: 2026-09-11T13:25:03Z Content type: news Language: en Sources: [Network World]() Topics: [Ethernet](), [Network](), [Networks](), [Hardware](), [Specifications](), [standard](), [interface]() Tags: [artificial-intelligence](), [artificial-intelligence-data-center-high-performance-computing-networking](), [data-center](), [ethernet](), [hardware](), [high-performance-computing](), [networking](), [networks](), [specifications](), [speed](), [standard]() ### AI overview This article examines the development and standardization of 1.6-terabit Ethernet, including IEEE P802.3dj, vendor hardware demonstrations, and OIF's 1600ZR coherent interface. It also distinguishes true 1.6T Ethernet from ports that aggregate multiple lower-speed connections. ### Source excerpt Networks have always been driven by the need for more speed and bandwidth, and in the AI era, that drive is accelerating the path to 1.6 terabit network connectivity. Work on 1.6T began in 2020 and became a formal IEEE study group in 2021, before AI was the industry's dominant discussion point. That work became the IEEE P802.3dj Task Force, which has been building out the specifications that are now in the balloting and formal review process before finalization, which could happen later this year. While the standard is being ratified, the demand is already present, and that's manifesting as real hardware work, not just abstract roadmaps. Later this month, the Ethernet Alliance is putting 1.6T switches, modules, and cables from more than a dozen vendors on the same show floor at ECOC (European Conference on Optical Communications) in Malaga, Spain, to see if any of it actually talks to each other. At the same event, the Optical Internetworking Forum (OIF) will be highlighting its publication of the Implementation Agreement for 1600ZR, a coherent interface built to carry a full 1.6 terabit Ethernet connection over a single wavelength, across the kind of distance that separates buildings on a campus or spans a metro area. The goal of the demonstration is to show that the technology isn't hype, it's real. "That's always the hope of the presentation and the display, that it's not just technical mumbo jumbo on a sleep-inducing IEEE specification," David Rodgers, president and chair of the Ethernet Alliance, told Network World. "It's actually live. It's happening." Where 1.6T Ethernet stands today Not everything marketed as 1.6T means the same thing. A switch port built on eight lanes can be split into different combinations of port speeds rather than a single 1.6T Ethernet connection. "There's a lot of people out there that are doing 1.6 terabit-per-second capacity, and that's not the same thing as 1.6 terabit Ethernet," John D'Ambrosia, chair of the IEEE P802.3dj Task Fo