Follow the AI industry.

ByteDance has launched its Seed2.1 model family across Volcano Engine and Doubao, targeting complex agent tasks that require coordinated tool use and cross-environment desktop interaction.

Reliance Industries will build the facility in Gujarat with expansion options, supported by seawater cooling and 925 MW in separate clean-energy procurement agreements.

Huawei Cloud unveiled its Agentic Infra architecture, combining ultra-large cluster networking, tiered context memory, and unified scheduling with an open-beta agent orchestration platform and open-source core.

Apple’s redesigned assistant is intended to use personal context and act across apps, with initial access depending on device support, language, location and beta timing.

The Meitner Energy Center will pair a Google data center with new clean-power generation in Gray and Roberts Counties. Google plans air cooling to limit water consumption.

NVIDIA has introduced an open humanoid reference robot for academic research, pairing a Unitree chassis and Sharpa tactile hands with Jetson Thor compute ahead of planned system availability in late 2026.

OpenAI and partners Oracle, Related Digital, and Walbridge have started construction on "The Barn," pairing a planned 1 GW infrastructure buildout with utility-protection pledges and student software credits.

NVIDIA is pairing third-party robotics hardware with its Jetson Thor onboard compute and Isaac GR00T open software to standardize humanoid robotics research. Full reference robots are scheduled to ship through Unitree in late 2026, while software workflows for existing Unitree G1 humanoids are slated for an earlier release.

The partnership targets logistics operations in Reno, Nevada, serving retail brands such as JCPenney, Brooks Brothers, and Aéropostale with Brookfield identified as an investor in both companies.

The modular codebase provides unified configurations and standardized interfaces spanning data processing, simulation, and real-world deployment across diverse robot hardware.

The name “Alibaba Token Hub” sounds like strategy-deck language. The org chart beneath it is more serious: model research, cloud delivery and AI applications now share a group, with CEO Eddie Wu personally in charge.

Micron has started volume shipments of its 12-high, 36GB HBM4 memory stacks for Nvidia Vera Rubin platforms, pairing on-package accelerator bandwidth with separate high-capacity system memory and PCIe Gen6 storage tiers.

Meta has outlined plans for four custom accelerator generations developed with Broadcom, shifting internal hardware from recommendation training toward low-precision generative AI inference while sharing rack designs.

Amazon will invest up to $50 billion in OpenAI while securing an expanded eight-year, $100 billion commercial compute commitment tied to AWS Trainium silicon.

Meta has signed a multi-year infrastructure agreement with AMD to support its AI infrastructure with up to 6 gigawatts of AMD Instinct GPUs, aiming to start physical deliveries in the second half of 2026.

The fashionable consumer-AI product begins with an empty chat box. Baidu’s new Personal Super Intelligence group starts somewhere less glamorous and potentially more useful: the documents and folders people already have.

Agreements with Vistra, Oklo, and TerraPower combine operational plant extensions and uprates with long-range advanced reactor builds to feed regional grids serving Meta's domestic data centers.

OpenAI and SoftBank Group are committing $1 billion in developer SB Energy alongside a 1.2 GW lease in Milam County, Texas, tying large-scale cluster expansion to new power generation.

Boston Dynamics has unveiled the commercial production version of its electric Atlas humanoid robot, reserving all 2026 fleet deliveries for Hyundai and Google DeepMind ahead of broader commercial availability planned for 2027.