45k user reviews & 1 million votes, over 29k web hosting brands & 85k plans since 2004!
Menu
WHTop → 📋 News → Arista Targets AI Cluster Bottlenecks With 1.6T Ethernet
Jun, 2026 : Arista Targets AI Cluster Bottlenecks With 1.6T Ethernet
📅 - Arista Networks has introduced its 7060XE7 Series, a 1.6-terabit Ethernet switching portfolio for rack-scale AI fabrics, as hyperscale customers try to connect larger accelerator clusters without letting power, cooling and congestion turn expensive GPUs into stranded capacity. The systems target air, liquid and hybrid-cooled environments built around AMD, Broadcom and major cloud operators today at scale globally.
The launch lands in a market where the server is no longer the only unit of AI infrastructure that matters. A single accelerator is useful. A rack of accelerators is expensive. Hundreds of thousands of XPUs connected poorly are a capital problem wearing a technical [...][... Check source for end of article ...]
Want to add a website news or press release ? Just do it, it's free! Use add web hosting news!
Related news
📅 - Gradium Raises Voice AI Funding to $100M With NVIDIA Support - Voice AI startup Gradium has expanded its financing to $100 million after bringing additional investors into its seed round, including NVIDIA. The fresh capital arrives as competition intensifies around speech models, with developers and enterprise software vendors increasingly treating voice as a core interface rather than a standalone application for future digital services and automation.
The financing says as much about where artificial intelligence investment is heading as it does about Gradium itself. Text generation has become crowded. Voice infrastructure remains comparatively open, especially for companies trying to build real-time conversational systems that can operate across [...]
📅 - AI-Native Startups Reach Unicorn Status In Half The Time - Amazon Web Services says AI native startups are reaching billion dollar valuations in 3.5 years, roughly twice as fast as pre generative AI peers, according to a new global study of 3,400 founders and senior leaders. The numbers are striking. They also arrive from a cloud provider with every reason to make AI company formation look cheaper to investors today.
The report, titled Engines of Growth, tries to separate startups that merely use AI from companies built around it from the first line of code. AWS defines the group as companies under five years old with AI at the center of the product, not just in customer support, analytics, or engineering workflows. That is a useful boundary. [...]
📅 - OpenAI and Broadcom Unveil Jalapeño Inference Chip - OpenAI and Broadcom have unveiled Jalapeño, a custom inference accelerator designed around large language models, claiming early lab results show better performance per watt than current leading systems, as OpenAI extends its infrastructure ambitions from models and products into silicon intended for gigawatt-scale data centers beginning with partners by late 2026 and, eventually, more generations of hardware.
The chip is being framed as OpenAI's first Intelligence Processor, which is grand language for a simple industrial reality: the company does not want to remain fully dependent on other people's accelerators as inference demand becomes the expensive center of the AI business. [...]
📅 - OVHcloud Public Cloud Growth Tops 20% as AI Plans Expand - OVHcloud reported faster third-quarter growth as public cloud revenue rose more than 20%, helped by starter products, corporate demand and fresh AI moves. The French cloud provider confirmed its fiscal 2026 targets, including positive levered free cash flow, while highlighting a European Commission sovereign-cloud win and planned Gladia acquisition during a crowded quarter for cloud infrastructure announcements.
For OVHcloud, the quarter is partly about numbers. €289.6 million in revenue, up 6.9% like for like. Nine-month revenue of €844.9 million, up 6.0%. Nothing explosive. But the shape is more interesting than the total.
Public Cloud is doing the work. Revenue in that segment reached [...]
📅 - NetApp Targets AI Data Bottlenecks Beyond GPU Clusters - NetApp is expanding its AI infrastructure portfolio around a growing enterprise concern: GPUs are expensive, but poorly managed data may be even more costly. The storage and data management vendor is introducing new capabilities aimed at helping organizations move, govern, and feed massive AI workloads more efficiently across hybrid environments, as enterprise deployments begin scaling beyond pilot projects.
The AI infrastructure conversation has spent the last three years obsessing over compute.
NVIDIA chips. GPU clusters. Power capacity. Cooling systems.
All important. None of them particularly useful if organizations cannot find, govern, secure, move, and prepare the data those [...]
📅 - Arista Networks has introduced its 7060XE7 Series, a 1.6-terabit Ethernet switching portfolio for rack-scale AI fabrics, as hyperscale customers try to connect larger accelerator clusters without letting power, cooling and congestion turn expensive GPUs into stranded capacity. The systems target air, liquid and hybrid-cooled environments built around AMD, Broadcom and major cloud operators today at scale globally.
Reads: 18 | Category: General | Source: Hosting Jurnalist : Hosting Jurnalist | Author: