DriveNets

The most expensive idle asset in the world right now is a GPU waiting on the network. Following our $410 million Series D funding announcement this week, Ido Susan, CEO and co-founder at DriveNets, explores the true cost of AI infrastructure and how any network weakness can drastically reduce GPU utilization and extend job completion times. 👉 drivenets.com/blog/the-most-… DriveNets addresses this with our Scheduled Ethernet-based AI fabric. We deliver high-performance, low-latency, and reliable connectivity across large GPU clusters - supporting scale-up, scale-out, and scale-across architectures - along with front-end and storage connectivity. Having built large-scale networking solutions for the world’s leading service providers, we're brought this foundation to AI infrastructure, where GPU clusters demand massive scale, resilience, and operational efficiency. With more than $1B in secured business, learn how DriveNets is looking to support its growing AI fabric pipeline and expand our Heterogeneous AI solution.

Break through the barriers to AI infrastructure growth.​ DriveNets proves that data center power does not have to limit AI infrastructure scaling. ​ With long-distance, scale-across AI networking, clusters can span across multiple data centers.​ Read the full deployment story: drivenets.com/blog/inside-wh…

With long-distance, scale-across AI networking, clusters can span across multiple data centers - physical distance is no longer a barrier. drivenets.com/news-and-event…

Let’s meet at @AMD #AdvancingAI 2026!​ DriveNets’ AI networking solutions support full-stack optimization of AMD-based infrastructures, eliminating GPU idle time and optimizing CPMT.​ get.drivenets.com/meet-drivenets… We recently published a joint end-to-end reference architecture with AMD, that defines a validated design blueprint for building high-performance AI infrastructure using AMD Instinct™ MI350 series GPUs and DriveNets AI Fabric. This combined solution offers a high-scale, open alternative to vertically integrated AI systems.​ Meet our experts to discover more

Will you be in Vegas at Ai4? Building an AI cluster and looking for high-performance Ethernet Fabric options? Check out our Ethernet-based networking fabric, proven to deliver InfiniBand-level performance, with faster deployment and any GPU (Nvidia or AMD), any optics, and any NIC. We’ll be at Ai4 in Las Vegas (booth #840) and will be happy to meet and discuss our customers’ deployment experiences (NeoClouds and Enterprises). 👉 get.drivenets.com/ai4-2026-sched… Our end-to-end AI networking solution delivers: ✔️ Highest performance - up to 30% faster job completion time (JCT) ✔️ Open, lossless Ethernet fabric ✔️ Fastest deployment time without complex fine-tuning ✔️ The best solution for multi-site and multi-tenancy

Happy Independence Day! ​ We would like to wish a happy 4th of July to all our US friends and family. ​ May you have a safe, happy and wonderful day!​ #fourthofjuly

Meet us at the RAISE Summit​ Let’s talk the importance of networking in your AI infrastructure.​ ​get.drivenets.com/raise-summit-2… Our AI networking solutions provides: ​ 🔹 Highest performance scale-up, scale-out, scale-across, and integrated front-end and storagenetworking, for any GPU​ 🔹 Highest end-to-end performance for AMD-based infrastructures – with full-stack optimization, including AMD ROCm™, RCCL, network plugins, NICs, servers, and system-level orchestration ​ 🔹 Fastest deployment time

New1.6 Tbps platforms from @drivenets based on @Broadcom Tomahawk 6 (TH6) ASIC, support hundreds of thousands of XPUs.​ ​drivenets.com/news-and-event… Supporting:​ High-performance alternative to a single-vendor solution, with the same performance and at a lower cost​ Scale-up, scale-out, scale-across, and integrated front-end and storage​ Better token economics with full-stack performance optimizations ​100% liquid cooling and air-cooling options, for any data center

🇨🇦​Happy Canada Day // Joyeuse fête du Canada! ​ 🍁 Celebrating the values, diversity, and spirit that make Canada so special. To all our employees, partners, and customers across the country—we hope you enjoy the holiday! Wishing you a safe and joyful day filled with red, white, and plenty of Canadian pride. 🇨🇦

Let’s meet at @AMD #AdvancingAI 2026!​ ​drivenets.lodago.app/meet/e/18b1bcb…​ DriveNets’ AI networking solutions support full-stack optimization of AMD-based infrastructures, eliminating GPU idle time and optimizing CPMT.​ We recently published a joint end-to-end reference architecture with AMD, that defines a validated design blueprint for building high-performance AI infrastructure using AMD Instinct™ MI350 series GPUs and DriveNets AI Fabric. This combined solution offers a high-scale, open alternative to vertically integrated AI systems.

DriveNets doesn’t let AMD GPUs wait Let’s sync at AMD Advancing AI 2026: drivenets.lodago.app/meet/e/18b1bcb… With DriveNets' full-stack networking solutions, AMD-based infrastructure experiences the lowest idle time and enjoys the best CPMT (Cost per Million Tokens). We recently published a joint end-to-end reference architecture that defines a validated design blueprint for building high-performance AI infrastructure using AMD Instinct™ MI350 series GPUs and DriveNets AI Fabric.  This combined solution offers a high-scale, open alternative to vertically integrated AI systems. Our end-to-end AI networking solution delivers: - Highest Performance: Maximizes GPU utilization with lossless connectivity for AMD environments - Fastest Deployment: Automated cluster bring-up and full-stack optimization to get you live sooner

The most important part of your AI cluster is not the GPU. It's the network. Once training scales beyond a single node, performance depends on how well GPUs communicate - not just how fast they compute. If the network cannot keep up, then GPUs wait. And idle GPUs are a wasted investment. Read insights from Sani Ronen: aijourn.com/the-most-impor…

Will you be in Vegas at Ai4? Building an AI cluster and looking for high-performance Ethernet Fabric options? Check out our Ethernet-based networking fabric, proven to deliver InfiniBand-level performance, with faster deployment and any GPU (Nvidia or AMD), any optics, and any NIC. We’ll be at Ai4 in Las Vegas (booth #840) and will be happy to meet and discuss our customers’ deployment experiences (NeoClouds and Enterprises). 👉 get.drivenets.com/ai4-2026-sched… Our end-to-end AI networking solution delivers: ✔️ Highest performance - up to 30% faster job completion time (JCT) ✔️ Open, lossless Ethernet fabric ✔️ Fastest deployment time without complex fine-tuning ✔️ The best solution for multi-site and multi-tenancy

System integrators are expected to deliver ✔️ massive AI environments faster ✔️ with more flexibility ✔️ better performance ✔️ fewer deployment risks But when the architecture depends on closed systems, constrained supply chains, fragmented expertise, and most importantly, a network that was not designed for AI workloads at scale - that is no easy task. Learn how DriveNets provides a complete networking solution specifically designed to address these challenges. 👉 drivenets.com/resources/eboo…

One GPU for everything no longer works for AI infrastructure. As inference becomes the dominant cost of AI, the mismatch is getting harder to ignore: prefill needs raw compute, while decode needs memory bandwidth and low latency. The solution: Heterogeneous AI - matching each stage of inference to the hardware it actually needs. But splitting inference across different kinds of hardware puts the network on the critical path. Networking becomes a first-class part of the architecture, not something you order last, and standard datacenter network is not enough. Read the new blog from Adam Wolman and learn how Heterogeneous AI is reshaping AI infrastructure. 👉 drivenets.com/blog/the-hidde…

Recently on @theallinpod youtu.be/V0lFjTWx36I?t=… (35:00) @GavinSBaker shares insights on "Reinventing Networking" "I would just say, well you know, two names that we've been involved in publicly are @AriaNetworks and @DriveNets. And they're both in the networking space and basically as data centers get more specialized and complicated you're going to have increasingly specialized chips it's called the disaggregation of inference and pre-fill and decode. And to make all of these chips to work together, like a symphony, and have the kind of the right chip for the right job at the right time, I do think we need to reinvent networking and @AriaNetworks and @DriveNets are coming at that in a very different way."

🌈 Pride Month is a time to celebrate inclusion, authenticity, and the freedom to be who we are. Today, we had the pleasure of hosting internationally recognized comedian Shahar Cohen, known for his viral “America vs. Israel” videos and the unforgettable character, Nofar. Appearing with no-nonsense Israeli confidence, Nofar let us laugh a little at the quirks of hi-tech office life. But beyond the comedy, Shahar also shared his personal story, including the challenges he's faced growing up. It was an important reminder of the power of authenticity, and of how humor brings us closer. Thank you, Shahar, for bringing us together with laughter and honesty, and being the heart of our Pride Month celebration. 🎞️ youtube.com/watch?v=QGVdm5… #PrideMonth 🌈

𝗗𝗿𝗶𝘃𝗲𝗡𝗲𝘁𝘀 は 𝗜𝗻𝘁𝗲𝗿𝗼𝗽 𝟮𝟲 𝗧𝗼𝗸𝘆𝗼 に出展します🎉 👉ブース小間番号: 5Y15 ✅ 国内複数の通信事業者様で 商用導入が開始 ✅ Dell AI Factory に DriveNets Fabric Scheduled Ethernet (FSE) が採用 ✅ AMD Instinct GPU との Reference Architecture を公開 ✅ 直近、$410M(Series D) の資金調達を完了/ 累計調達額 $1B、企業価値 $8.5B へ 「Disaggregated Networking」が"検証フェーズ"から"商用フェーズ"へ。 AT&T・Comcast・KDDI に続き、日本でも本格展開が始まっています。 AMDも新たに出資参画し、Heterogeneous AI 時代の次のステージへ加速しています。 🔗 japan.drivenets.com/home 👉 DriveNets、Telecom Infra Project (TIP) 準拠クラスタベース DDBR コアルーターの商用運用を KDDI Corporation バックボーンで開始 get.drivenets.com/drivenets-kddi… 👉 ドライブネッツ、4 億 1,000 万ドルのシリーズ D 資金調達を実施 - 大規模 AI 導入でのイーサネット・ファブリック需要の急増に対応 get.drivenets.com/drivenets-roun…

𝗗𝗿𝗶𝘃𝗲𝗡𝗲𝘁𝘀 は 𝗜𝗻𝘁𝗲𝗿𝗼𝗽 𝟮𝟲 𝗧𝗼𝗸𝘆𝗼 に出展します🎉 👉ブース小間番号: 5Y15 ✅ 国内複数の通信事業者様で 商用導入が開始 ✅ Dell AI Factory に DriveNets Fabric Scheduled Ethernet (FSE) が採用 ✅ AMD Instinct GPU との Reference Architecture を公開 ✅ 直近、$410M(Series D) の資金調達を完了/ 累計調達額 $1B、企業価値 $8.5B へ 「Disaggregated Networking」が"検証フェーズ"から"商用フェーズ"へ。 AT&T・Comcast・KDDI に続き、日本でも本格展開が始まっています。 AMDも新たに出資参画し、Heterogeneous AI 時代の次のステージへ加速しています。 レガシールーターからオープンネットワークへ：Tier-1 事業者のための現実的アプローチ 佐々木 俊輔 | J3-03 | 06.12（金）11:25〜12:05　展示会場内 J（7A09） forest.f2ff.jp/introduction/1… 🔗 japan.drivenets.com/home 👉 DriveNets、Telecom Infra Project (TIP) 準拠クラスタベース DDBR コアルーターの商用運用を KDDI Corporation バックボーンで開始 get.drivenets.com/drivenets-kddi… 👉 ドライブネッツ、4 億 1,000 万ドルのシリーズ D 資金調達を実施 - 大規模 AI 導入でのイーサネット・ファブリック需要の急増に対応 get.drivenets.com/drivenets-roun… #Interop2026

A single standalone router runs out of ports faster than the site does. A fully distributed chassis or a Clos cluster is overkill. Most real deployments live in the space between. And this is what DriveNets Back-to-Back is built to fill. Read the blog: drivenets.com/blog/drivenets… In our latest blog post, Hen Haklai explores how the DriveNets Back-to-Back architecture is designed for aggregation, edge, pre-aggregation, and peering sites that have outgrown a single standalone router but do not need a full cluster.

GPUs are essential, but they are not the system. If the network can’t keep up, GPUs wait, and waiting GPUs are the most expensive kind of inefficiency. aijourn.com/the-most-impor…