StorageReview.com

LIVE: Micron's 245TB SSD and What Extreme Density Enables at Scale nitter.cf/i/broadcasts/1RKZzBMna…

We sent 400TB of Solidigm QLC flash to the rugged terrain in South America to help a film crew capture the majesty of nature and the vibrant community of animals that live there. 6 SSDs in each Areca RAID unit with one more cold spare. Capybaras for scale. @solidigm

Dell XPS 16 Creator Edition reviewed: 20-core RTX Spark N1X at full speed in 17.8mm, 6,631 Cinebench multi-core, 11.17 FLUX.2 images/min, and the most repairable N1X notebook we have opened. storagereview.com/review/del… #DellXPS #RTXSpark #AIPC @Dell

Oracle 26ai puts vector search in the database, which makes the block array AI storage. IBM FlashSystem ran OLTP and vector search together, and 64G FC nearly doubled vector QPS over 32G. #Oracle26ai #FlashSystem #FibreChannel storagereview.com/review/ibm… @IBMData @Broadcom @Emulex

120 to 142kW per compute rack, 242kW N+1 direct-to-chip cooling, and a 1,400A DC busbar: Vertiv SmartIT ships GB300 NVL72 and Rubin racks factory-validated on NVIDIA MGX and OCP ORV3. storagereview.com/news/verti… #Vertiv #NVIDIA #LiquidCooling @Vertiv @NVIDIAAIInfra

Is Your Storage AI-Ready? Scale, Resiliency, and HPE Alletra Storage MP nitter.cf/i/broadcasts/1qKVmyjla…

NVIDIA Minerva pulls GPU, fabric, NVMe, and scheduler telemetry into one incident queue for AI factories. Dell, Lenovo, and Supermicro will sell it; IREN runs it today. storagereview.com/news/nvidi… #NVIDIA #AIOps #AIFactory @NVIDIAAI

Great track day at COTA with @HPE

Dell Pro Precision 9 T2 review: Xeon 658X, four channels of DDR5-6400 up to 1TB, two hot-plug E3.S Gen5 SSDs, RTX PRO 6000 at 300W or 600W. 39% ahead of the Core T2 multi-core. storagereview.com/review/del… #Dell #Xeon6 #Workstation @Dell @IntelBusiness

A mockup in hand, the real one bolted into Astrolab's FLIP rover: HPE Spaceborne Computer-4, with KIOXIA SSDs inside, flies to the lunar south pole in late 2026 to run AI inference, LiDAR navigation, and HPC benchmarks on the Moon. #HPE #Spaceborne #KIOXIA @HPE_Compute

HPE Spaceborne Computer-4 is headed to the lunar south pole on Astrolab’s FLIP rover: a fanless 2.6 lb module with two 1TB KIOXIA BG6 SSDs capped at 5W, running HPL, HPCG and AI inference on the Moon. storagereview.com/news/hpe-s… #HPE #Spaceborne #KIOXIA @HPE_Compute @KIOXIAAmerica

Meta’s Prometheus supercluster runs 64 of every 72 GPUs per NVL72 rack with 8 as spares, cut interruptions 99% from ~1,000 per 8K GPUs a day, and is scaling toward 1 million GPUs. storagereview.com/news/meta-… #Meta #NVL72 #AIInfrastructure @AIatMeta @NVIDIAAI

Surface Laptop Ultra review: the RTX Spark N1X ran a 35B coding agent at 89 tok/s and won the multi-core CPU tests, but we had one critical problem. storagereview.com/review/mic… #SurfaceLaptopUltra #RTXSpark #WindowsOnArm

Single slot, 165W, 32GB: the RTX PRO 4500 Blackwell Server Edition matched a 350W L40S at short context, 1,107 vs 995 tokens/s on Ministral 3 14B FP8. Replaces the work of 3x L4 GPUs. storagereview.com/review/nvi… #NVIDIA #RTXPRO4500 #Blackwell @nvidia

245TB in one E3.L SSD. The Micron 6600 ION fits an exabyte in 6 racks and returns ~120kW of power budget at scale, enough to run a GB200 NVL72. We go live with Micron’s Bill Bollengier Thu Oct 15, 2pm ET: linkedin.com/events/75144120… #Micron #SSD #DataCenter @MicronTech

Crescent Island is now Intel GPU D-series: up to 480GB LPDDR5X at 1.5TB/s, 1.3 PFLOPS FP8, 350W PCIe, sampling now. Intel’s own card is 160GB; partners go to 480GB. Plus Intel’s 46,000-core ORv3 racks: storagereview.com/news/intel… #Intel #OCP2026 #AIInference @intel

Micron 9650 PRO SSDs in hand: PCIe Gen6 x4, E3.S 7.5mm, 7.68TB on these. Micron rates the 9650 at up to 28GB/s sequential reads, double what a Gen5 x4 drive can move. Capacities in the family run to 30.72TB. Gen6 drives need next-gen servers, more on that soon from the lab. #Micron #PCIeGen6 #DataCenter @MicronTech

32,768 Zen 6 cores in one ORv3 rack: AMD Helion mixes 256c Venice at 400W and 600W with 96c high-frequency nodes for agents, or adds up to 16 MI350P in an all-in-one build. Planned for November. storagereview.com/news/amd-h… #AMD #EPYC #AgenticAI @AMDServer

HPE ProLiant Gen13 opens with four AMD EPYC Venice servers: a 10U DL585a with 8 double-wide GPUs on PCIe Gen6, a 256-core 1U DL525 with 16 memory channels, and two ORV3 XD nodes. iLO 8 adds Auto-SED. storagereview.com/news/hpe-p… #HPE #ProLiant #AMDEPYC @HPE_Compute

Two DGX Station GB300s in a Cluster: GLM-5.3: 188 output tok/s on one Station, 5,018 on the pair. GLM-5.2: 11x at 32 streams. DeepSeek V4.1 Flash: 5,248 tok/s. Five models tested: storagereview.com/review/dgx… #DGXStation #GB300 #AIInfrastructure @NVIDIAAI