Jason Collier

OpenAI, Broadcom Make $10 Billion Deal for Custom AI Chips Key Insight OpenAI has signed a $10 billion deal with Broadcom to develop custom AI chips, making OpenAI Broadcom’s fourth major AI chip customer and aiming to alleviate the processor shortages that have slowed the release of new ChatGPT versions. The new chips are expected to start shipping in 2026, reducing OpenAI’s dependence on Nvidia and directly supporting future upgrades of models like ChatGPT and GPT-5. Broadcom’s stock jumped nearly 11% after the announcement, signaling the market’s confidence in custom AI hardware as a growth driver and shifting industry dynamics. Background: On September 5, 2025, the Wall Street Journal reported OpenAI’s agreement with Broadcom for a one-time $10 billion order of AI server racks using Broadcom’s chips, following months of delays caused by a shortage of high-performance GPUs and intensifying competition with Nvidia in the AI hardware market. Credit Wall Street Journal

26

The Texas Advanced Computing Center (TACC) is charging forward with the Horizon supercomputer, set to deliver 10 times the performance of today's top supercomputers for simulations. Check out the full scoop in Axios: https://lnkd.in/g9TeTjjb #Supercomputing #HPC

2

That's a great point by AMD. For businesses optimizing SQL Server costs is key. We at Megahertz help clients save on licenses with high-performance AMD EPYC servers. #Servers #AMDEPYC #SQLServerAMD

1

1 Comment

Here's a great writeup from Taryn Plumb for Network World, taking a deeper look at how Google has started to migrate many of its internal workloads to its Arm-based servers. A few details worth noting: 🔸 These servers, built on the (Google) designed Axion CPU are now supporting about 30K of Google's 100K internal workloads 🔸 The goal is to have these workloads on both Arm & x86 simultaneously 🔸 Workloads inlcude YouTube, gmail, BiqQuery 🔸 Goal is to move everything to this shared model 🔸 Multiarch a guiding and design principle for Google developers Really interesting article that goes deeper into the port process - well worth the read. Article below ⬇️ https://lnkd.in/gxbyqh28

7

You’d think erasing NVMe storage would be quick and easy. However, erasing NVMe SSDs at scale has proven to be a slow, power-hungry, and inflexible process. A leading integrator decided to change that, and devised a cutting-edge NVMe erasure platform built around ATTO ExpressNVM Switch Host Adapters. With ExpressNVM, their solution securely wipes NVMe drives with maximum efficiency, reduced complexity, hot-plug support, and includes the ability to scale as needed. Check out this case study for details: https://lnkd.in/eHGXFUGR #DataCenterEfficiency #NVMesecurity #ITADsolutions #StorageInnovation #NVMe #storage #SSD #connectivitymatters #perfromance ATTO Technology, Inc.

23

As enterprises push agentic LLM workflows into production, challenges around data quality, security, compliance, and reliability remain top priorities. Using Nemotron NIM, Plurai, developed a solution that represents another step toward standardizing enterprise-grade benchmarking for LLMs in agentic use cases. 💡 Learn more in our technical blog: https://bit.ly/46jDYPV

21