Multi-Source Journalism
This article synthesizes reporting from multiple credible news sources to provide comprehensive, balanced coverage.
Multi-Source Journalism
This article synthesizes reporting from multiple credible news sources to provide comprehensive, balanced coverage.
Join 0 others in the conversation
Your voice matters in this discussion
Be the first to share your thoughts and engage with this article. Your perspective matters!
Discover more articles
Cisco has released a new AI-powered data centre router, the 8223 routing system, which it claims is the industry's first fixed router capable of handling 51.2 terabits per second. This technology aims to address the growing challenge of scaling up da
Broadcom has unveiled the Thor Ultra, an 800G Ethernet Network Interface Card (NIC) designed for large data centers that power advanced AI tools. This innovative card leverages PCIe Gen6 technology to significantly improve throughput, addressing limi
Taiwanese tech manufacturers made a significant splash at the 2025 OCP Global Summit in San Jose, California, showcasing cutting-edge innovations for high-density AI computing workloads. Notably, automated robots designed to refill liquid cooling sys
OpenAI has submitted a letter to the White House, urging the government to expand the tax credit from the Chips Act to cover data centers and AI infrastructure, citing the need to alleviate bottlenecks and accelerate AI development in the US. The pro
OpenAI President Greg Brockman is spearheading a massive $1.4 trillion infrastructure push, driven by the company's ambitious Stargate Project, which aims to deploy hundreds of thousands of high-performance AMD chips across its data centers. This mon
Meta and Oracle have selected NVIDIA's Spectrum-X technology to upgrade their AI data centers, enabling efficient training and deployment of massive AI models. The open networking framework will connect millions of GPUs across compute clusters, accel
Here is a 2-3 sentence summary of the article: Despite recent advancements in GPU-powered AI, central processing units (CPUs) remain the backbone of high-performance computing, supporting up to 90% of simulation jobs. A new wave of CPU innovation, i
A photonics startup, Finchetto, has developed a groundbreaking all-optical packet switch that significantly outperforms traditional electronic systems, boasting speeds up to 1000 times faster and reduced power consumption. The innovative design uses
Cisco has developed a groundbreaking technology that bridges the gap between classical computing and quantum networks, enabling seamless communication between them. A newly designed quantum compiler makes connected qubits network-aware, paving the wa
Meta and Oracle have partnered with NVIDIA to upgrade their AI data centers using Spectrum-X technology, a open networking framework designed to connect millions of GPUs across compute clusters. This collaboration aims to improve AI training efficien
Cisco has launched a new AI data centre router, the 8223, which boasts a record-breaking 51.2 terabit per second speed, designed to connect multiple facilities running AI workloads. This move comes amidst an industry-wide challenge: managing exponent
Cisco has launched a new AI-powered data centre router, the 8223 system, which can process an industry-leading 51.2 terabits per second of data. This innovation aims to tackle the growing challenge of scaling up infrastructure to support increasingly
Cisco has developed a pioneering technology that bridges classical computing with quantum networks by creating a practical quantum networking chip. This innovation enables the integration of quantum computers into existing fiber-optic infrastructure,
Researchers at Columbia University have inadvertently created a tiny "rainbow chip" capable of producing dozens of powerful light channels simultaneously from a single laser beam. This breakthrough technology has the potential to revolutionize data c
Cisco has entered the market with its 8223 routing system, a 51.2 terabit per second fixed router designed specifically for connecting data centres running AI workloads, addressing the industry's growing need for scalable infrastructure. This move co
Elon Musk has expressed interest in space-based data centers, stating that SpaceX will utilize its Starlink V3 satellites to construct large data centers in orbit. This move significantly raises the profile of the nascent industry, which promises lim
Cisco has developed a groundbreaking technology that bridges the gap between classical computing and quantum networks, enabling the creation of more efficient and scalable quantum systems. A new quantum compiler allows connected qubits to become netw
Oracle has unveiled its Zettascale10 system, a massive cloud-based AI supercomputer that can reach 16 zettaFLOPS of peak performance. The system leverages 800,000 Nvidia GPUs spread across data centers and features custom networking design optimized
Nvidia has expanded its NVLink Fusion ecosystem by partnering with Samsung Foundry to enable the design and manufacturing of custom CPUs and XPUs. This move follows a recent collaboration with Intel, allowing x86 CPUs to connect directly to Nvidia pl
A potentially explosive AI bubble is looming, driven by massive investments in infrastructure and services that may not meet future demand. The rapid pace of AI software development is mismatched with the slow build-out of data centers, making it dif
Oracle has unveiled Zettascale10, its largest AI supercomputer in the cloud, which boasts up to 16 zettaFLOPS of peak performance. This massive system will connect hundreds of thousands of NVIDIA GPUs across multiple data centers, enabling customers
Tech giants Meta, Nvidia, and others are collaborating on a project called ESUN to develop open standards for high-performance connections within AI clusters using Ethernet. This initiative aims to challenge InfiniBand's dominance by leveraging Ether
A massive AI infrastructure project is underway, estimated to cost $7 trillion over the next five years, with major players like Nvidia investing up to $100 billion in OpenAI. This global computing system aims to revolutionize various aspects of our
NVIDIA has introduced a new solution to address the growing issue of AI data centres running out of space by connecting multiple locations into a single, high-performance network. This innovation, called Spectrum-XGS Ethernet technology, enables seam
Share & Engage Share
Share this article