<p>➀ Nvidia has transitioned from HGX B300 NVL16 to NVL8 to accommodate faster GPU-to-GPU interconnects for LLMs with sparse MoE architectures like DeepSeek and LLAMA4.</p><p>➁ NVL8 and NVL16 both have 16 GPU dies, but NVL8 integrates two GPU dies per package, connected via NV-HBI, offering up to 10 TB/s bandwidth.</p><p>➂ The shift to NVL8 will enable Nvidia to predominantly use CoWoS-L packaging, optimizing supply chain management.</p><p>➃ DGX Spark uses the GB10 chip, expected to be in mass production around June-July, with end product shipments in August-September.</p><p>➄ The N1X is Nvidia’s Windows-on-ARM processor, requiring significant software optimization for laptop use and Windows support, potentially delaying laptops into 2026.</p>
Related Articles
- OpenAI Taps Amazon AWS To Scale AI Compute On NVIDIA Blackwell In Massive $38B Deal1 day ago
- NVIDIA Hits $5 Trillion Valuation As CEO Pops AI Bubble Concerns6 days ago
- Nvidia Poised for Record $5 Trillion Market Valuation7 days ago
- Qualcomm launches accelerators for inference8 days ago
- White GeForce RTX 5080 FE Mod Looks So Good NVIDIA Should Make It Official9 days ago
- Inside Nvidia's Org Chart: See the 36 Leaders Reporting to CEO Jensen Huang10 days ago
- Intel Reaffirms 14A, Deepens Nvidia Alliance11 days ago
- This is the Massive NVIDIA 800G OSFP to 2x 400G QSFP112 Passive Splitter DAC Cable11 days ago
- Intrepid modder builds Frame Warp demo from Nvidia Reflex 2 binaries — tech remains mysteriously shelved despite greatly reducing latency12 days ago
- NVIDIA Upgrades RTX Pro 5000 Blackwell GPU With Monstrous 72GB VRAM15 days ago