Autonomous Reinforcement Learning Robot Control with Intel's Loihi 2 Neuromorphic Hardware

An unannounced Dell laptop has surfaced on Geekbench, indicating that the company is likely revising its product naming strategy just a year after its last update. The benchmark results also provide early insights into the performance of Intel's upcoming Xe3 integrated graphics chips, which are expected to enhance the laptop's capabilities.

Taiwan Semiconductor Manufacturing Company (TSMC), Intel, and Amkor are investing tens of billions to transform Phoenix into a semiconductor hub, highlighting the complexities of executing large-scale projects in the U.S. This initiative aims to bolster domestic chip production amid global supply chain challenges.

A new quantization technique called Four Over Six (4/6) has been introduced to enhance the NVFP4 quantization algorithm, which is crucial for large language models (LLMs). This method evaluates two potential scale factors for each block of values, addressing the performance degradation often seen during training and inference due to quantization errors.

The NASA Astrobee has successfully demonstrated the first on-orbit application of reinforcement learning (RL) for autonomous control aboard the International Space Station (ISS). This achievement involved training a deep neural network using NVIDIA's Omniverse physics simulator, allowing the Astrobee to navigate effectively in microgravity environments. The results validate a new training pipeline that bridges the simulation-to-reality gap, showcasing the potential for RL in space robotics.

A new framework named throttLL'eM has been introduced to optimize energy consumption during Large Language Model (LLM) inference by utilizing GPU frequency scaling while adhering to Service-Level Objectives (SLOs). This approach addresses the growing energy demands associated with LLMs, which are heavily reliant on GPUs for processing. The framework incorporates machine learning to predict future cache usage and batch sizes, allowing for efficient performance management.

The US Naval Research Laboratory's APIARY experiment successfully demonstrated the first reinforcement learning control of a free-flying robot in space using the NASA Astrobee on the International Space Station on May 27, 2025. This experiment involved training a robust control policy in a simulated environment, enhancing the robot's ability to operate autonomously in zero-gravity conditions.

Germany has redirected funds initially allocated to Intel into new semiconductor projects, signaling a strategic shift aimed at enhancing its technological capabilities and competitiveness in the European tech sector.

Nvidia maintained a dominant position in the discrete GPU market with a 92% share in Q3 2025, despite a slight decline from 94% in the previous quarter. AMD and Intel have made modest gains, increasing their market shares to 7% and 1.4%, respectively. This shift indicates a competitive landscape as Nvidia continues to focus on AI technologies.