News from manufacturers

HPE AI Factory With NVIDIA Expands for the Era of Agents

NVIDIA and HPE are expanding their AI factory collaboration with the introduction of the NVIDIA Vera CPU, available with HPE Private Cloud AI in 2027. The NVIDIA Agent Toolkit is also now available with HPE Private Cloud AI, and NVIDIA Confidential Computing extends across HPE AI Factory solutions. Additionally, the HPE ProLiant Compute DL380a has achieved certification for NVIDIA Confidential Computing.

“The HPE ProLiant Compute DL394 Gen12 with the NVIDIA Vera CPU will be available in 2027 with HPE Private Cloud AI” — NVIDIA

Fastest, Largest, Strongest: NVIDIA Blackwell Sweeps MLPerf Training 6.0

NVIDIA's Blackwell platform achieved the fastest training times across all seven benchmarks in MLPerf Training 6.0, including new mixture-of-experts workloads. The GB300 NVL72 system delivered up to 1.6x faster training than the GB200 NVL72. NVIDIA submitted results at scales up to 8,192 GPUs, the largest Blackwell-based submission in MLPerf Training to date.

“GB300 NVL72 delivered up to 1.6x faster training than GB200 NVL72 at the same scale.” — NVIDIA

Related: NVIDIA RTX PRO 2000 Blackwell Mobile , NVIDIA RTX PRO 5000 Blackwell Mobile , NVIDIA RTX PRO 3000 Blackwell Mobile , NVIDIA RTX PRO 4000 Blackwell Mobile , NVIDIA RTX PRO 1000 Blackwell Mobile , NVIDIA RTX PRO 500 Blackwell Mobile , NVIDIA RTX PRO 4000 Blackwell , NVIDIA RTX PRO 2000 Blackwell

NVIDIA Blackwell Leads on First Agentic AI Infrastructure Benchmark

Artificial Analysis released AgentPerf, the first benchmark for agentic AI workloads. NVIDIA's Blackwell Ultra NVL72 platform achieved leading performance, running up to 20x more agents per megawatt than the Hopper-based HGX H200 system. The benchmark measures concurrent agentic tasks per accelerator and per megawatt.

“NVIDIA GB300 NVL72 delivers the highest performance in the benchmark, running up to 20x more agents per megawatt than the NVIDIA HGX H200 system.” — NVIDIA

Related: Seagate FireCuda 540 1TB , Seagate FireCuda 540 2TB , Seagate FireCuda 530 2TB , Seagate FireCuda 530R 2TB

Save Big and Play Bigger: GeForce NOW Summer Sale Brings Major Membership Savings

NVIDIA's GeForce NOW summer sale offers discounts on 12-month memberships, with $35 off Performance and $70 off Ultimate tiers. The service streams games via cloud servers, eliminating local hardware requirements. Guild Wars 3 will be available on GeForce NOW at launch.

“Guild Wars 3 is coming to GeForce NOW, bringing its next-generation massively multiplayer online role-playing game (MMORPG) experience to the cloud at launch.” — NVIDIA

Related: Asrock Fatal1ty H270M Performance , Asrock Fatal1ty H370 Performance , Asrock Fatal1ty 970 Performance/3.1 , Asrock Fatal1ty H170 Performance/Hyper , Asrock Fatal1ty 970 Performance , Asrock Fatal1ty H97 Performance , Asrock Fatal1ty E3V5 Performance Gaming/OC , Asrock Fatal1ty H170 Performance

For Robotaxis, Safety Must Be Built In, Not Bolted On

NVIDIA announced new robotaxi collaborations at GTC Taipei, including partnerships with Uber, Autobrains, Foxconn, VinFast, and HUMAIN. The company introduced the Halos Operating System, a safety-certified platform for autonomous vehicles, which includes Halos Core (compliant with ISO 26262 ASIL D) and Halos SDK for standardized interfaces.

“Halos Core is the next generation of NVIDIA DriveOS and certified to automotive safety standards.” — NVIDIA

NVIDIA Accelerates Google DeepMind’s DiffusionGemma for Local AI

Google DeepMind released DiffusionGemma, an open model that generates multiple words in parallel for faster text generation. NVIDIA optimized it to run on GeForce RTX, RTX PRO, and DGX Spark systems, achieving up to 4x faster performance than autoregressive models at batch size 1. The model is open weights under Apache 2.0 and runs locally on RTX and DGX Spark.

“DiffusionGemma delivers 1,000 tokens/sec at batch size 1 on a single NVIDIA H100 Tensor Core GPU, 150 tokens/sec on NVIDIA DGX Spark and fastest local inference on NVIDIA DGX Station — roughly 4x faster than an equivalent autoregressive mod” — NVIDIA

Related: NVIDIA GeForce RTX 4090 , NVIDIA GeForce RTX 4080 , NVIDIA GeForce RTX3080 12 GB , NVIDIA GeForce RTX 3080 Ti , NVIDIA GeForce RTX 4060 Ti , NVIDIA GeForce RTX 2060 Max-Q , NVIDIA GeForce RTX 4090 D , NVIDIA GeForce RTX4070 GDDR6

Supermicro and Arm advance compute for the agentic AI era

At Computex, Supermicro announced servers powered by Arm's AGI CPU, designed for agentic AI workloads. The AGI CPU features up to 136 Neoverse V3 cores, 12 DDR5 channels, and PCIe Gen6 within a 300W envelope. Supermicro's portfolio includes liquid-cooled and air-cooled systems for cloud, enterprise, and edge deployments, with sampling starting in Q4 2026 and production in 2027.

“up to 2x higher performance per rack to comparable x86-based solutions, according to Arm estimates” — Arm

NVIDIA Confidential Computing to Help Expand Apple’s Private Cloud Compute

NVIDIA GPUs with Confidential Computing are now used for confidential inference in Apple's Private Cloud Compute, expanding to Google Cloud. The collaboration supports server-side inference for Apple Foundation Models using NVIDIA Blackwell GPUs. NVIDIA Confidential Computing provides hardware-based security for AI workloads, protecting data during processing.

“NVIDIA GPUs with Confidential Computing are now used for confidential inference in Apple’s Private Cloud Compute (PCC), as it expands beyond Apple’s data centers to Google Cloud.” — NVIDIA

New Smart City Pilot Network to Bridge Digital Divide for 50 U.S. Cities

Intel announces a smart city pilot network aimed at bridging the digital divide for 50 U.S. cities. The initiative focuses on providing internet access to underserved communities.

Apple unveils innovative features and intelligence experiences across services

Apple announced upcoming software releases for fall 2027, which will introduce new features to its services. The updates aim to enhance user experiences across Apple's service ecosystem.