Artificial Intelligence
-
OCP says AI racks are on a path past 1 MW as the industry moves from 800 VDC toward 1,500 VDC and more tightly co-designed power, cooling and grid infrastructure.
-
Meta’s new IcePack rack uses 102.4 Tb/s IceCube switches, liquid cooling, Flattened Planar Fabric and MetaRoCE to scale open Ethernet networking for AI clusters.
-
ASUS’ new XA P2N-E2 is a 2U NVIDIA MGX server with dual Vera CPUs, up to two RTX PRO 6000 Blackwell GPUs, PCIe 6.0 and SOCAMM2 memory.
-
Astera Labs’ Orion rack architecture scales from 8 to 160 accelerators per pod while letting operators choose their own GPUs and CPUs.
-
d-Matrix launches Squadron, an air-cooled AI inference cluster scaling to 1,024 Corsair XPUs across 16 racks for megawatt-scale deployments.
-
Intel’s GPU D-Series brings up to 480GB of LPDDR5X, 1.5TB/s bandwidth and 1.3 PFLOPS FP8 to a 350W air-cooled PCIe accelerator.
-
Meta launches Muse Wonder and Muse Wonder Full with 1M-token context, long-running agent features, new API pricing and a stronger push toward persistent AI agents.
-
AI and superintelligence can both be correct: AI is the broad category, while artificial superintelligence describes a hypothetical beyond-human capability level.
-
Brookings estimates U.S. AI infrastructure investment could reach $10.3 trillion from 2025 through 2032, averaging 3.63% of GDP per year.
-
Hyperscale Data cut its ATM sales ceiling from $300 million to $100 million after raising about $49.8 million, while keeping its Michigan AI buildout central.
-
LITEON is bringing its 800V DC architecture and a 1 MW AI power-rack presentation to OCP Global Summit 2026.
-
Intel is using OCP Global Summit 2026 to argue that agentic AI should route each workload to the CPU, GPU or accelerator that fits it best.
