GPU News & Updates
Latest announcements, benchmarks, and industry news about datacenter GPUs and AI accelerators
Curated links
OpenAI's First Chip, Jalapeño, Takes Aim at NVIDIA's Inference Margins
OpenAI unveiled Jalapeño, its first custom chip: an inference-only chip designed in-house and built with Broadcom, scaled over Ethernet rather than NVLink, with OpenAI's own models helping design it. OpenAI claims significantly better performance per watt than current hardware, but the chip is still in testing and it shared no TFLOPS, HBM capacity, power, or confirmed node. We separate what's confirmed from what's still marketing.
Huawei vs Nvidia: The 2026-2028 AI Accelerator Race
With Trump in Beijing and Jensen Huang pitching the H200 to Chinese buyers, we decode Huawei's Ascend 910 → 970 roadmap against Nvidia's Hopper-Blackwell-Rubin lineup. The BIS-TPP numbers most outlets cite aren't raw TFLOPS — once decoded, Huawei's 2028 flagship still sits ~8x behind Nvidia on FP4, but matches on memory capacity at the rack level.
Apple's M5 Matches the V100 — at a Fraction of the Power
Apple's latest M5 chip delivers V100-class FP32 performance from a laptop SoC. The M5 Max hits 16.59 TFLOPS FP32 at an estimated 50W — that's 7x more power-efficient than NVIDIA's V100 SXM at 300W. We've added the complete M1–M5 lineup (17 chips) to Flopper.io with full precision metrics.
OpenAI Partners with NVIDIA for 10GW AI Infrastructure Deployment
OpenAI and NVIDIA forge strategic partnership to deploy 10 gigawatts of next-generation AI compute infrastructure, with NVIDIA investing up to $100B. Initial deployments using NVIDIA Vera Rubin platform targeted for H2 2026, supporting OpenAI's AGI development mission.
NVIDIA Announces H200 with 141GB HBM3e Memory
The new H200 Tensor Core GPU features 141GB of HBM3e memory at 4.8TB/s bandwidth, offering 1.4x memory capacity and 1.7x bandwidth over H100.
AMD MI300X Delivers 192GB HBM3 for Large Language Models
AMD's latest datacenter GPU packs 192GB of HBM3 memory with 5.3TB/s bandwidth, targeting large language model training and inference workloads.
Stay Updated
Get the latest GPU news and benchmarks delivered to your inbox