AI Infrastructure: OpenAI, Oracle, and SoftBank Announce 5 New US Data Centers

OpenAI, Oracle, and SoftBank have announced the development of five new artificial intelligence (AI) data center locations in the United States as part of the Stargate AI supercomputer project. The new sites will be located in Shackelford County, Texas; Doña Ana County, New Mexico; Lordstown, Ohio; Milam County, Texas; and an unannounced Midwest location. This expansion brings the total planned capacity for the Stargate project to nearly 7 gigawatts, with a total investment exceeding $400 billion over the next three years. The companies are now ahead of schedule to meet their goal of a $500 billion, 10-gigawatt commitment by the end of 2025. The project is expected to generate over 25,000 onsite jobs. These new data centers are designed to provide the massive computing power required to train and operate next-generation AI models.

Sources:

Signal AI Secures $165M to Enhance Its Risk and Reputation Intelligence Platform

Signal AI, a London-based artificial intelligence startup, has successfully raised $165 million in a new growth-equity funding round. The investment was led by Battery Ventures, which will now acquire a majority stake in the company and support its next phase of growth. Signal AI’s AI-powered platform utilizes machine learning to proactively identify risks and reputational signals from a vast array of external sources, including news, regulatory filings, and social media across 226 markets. The company’s mission is to provide organizations with a clearer understanding of the complex relationships between businesses, individuals, and ongoing issues in an increasingly volatile world of risk management.

Sources:

Humanoid Robots Poised for Shipbuilding Trials by ABS and Persona AI

The American Bureau of Shipping (ABS) and robotics firm Persona AI have signed a memorandum of understanding to explore the use of humanoid robots in the maritime industry. The partnership will involve joint development projects to test robotic inspection technologies for Persona AI’s humanoid robot platform, which incorporates elements of NASA’s robotic hand technology. Unlike fixed industrial robots, humanoid models are designed to operate in spaces built for people, offering greater mobility and flexibility in confined shipyard environments. The data collected from these projects will inform new ABS standards for digital and remote surveys, potentially guiding future robot design and their application in certification processes.

MIT’s SCIGEN Tool Guides Generative AI to Discover Novel Quantum Materials

Researchers at MIT have created a new technique called SCIGEN that enables generative AI models to design materials with exotic quantum properties. While existing generative models are effective at designing stable materials, they often struggle with creating materials that exhibit unique quantum behaviors useful for applications like quantum computing. SCIGEN addresses this by incorporating specific design rules and constraints into the AI models, guiding them to generate materials with specific geometric structures, such as Archimedean lattices, which are known to give rise to quantum phenomena. Using this approach with the DiffCSP model, the researchers generated millions of candidate materials and have already successfully synthesized two new compounds, TiPdBi and TiPbSb, which exhibit exotic magnetic traits.

Sources:

Kubernetes v1.34 Beta Enhances Resource Management with Pod-Level Limits

The latest Kubernetes release, v1.34, has moved pod-level resource requests and limits to beta, enabling the feature by default. This enhancement in the container orchestration platform allows administrators to set resource boundaries for entire pods, rather than individual containers. This is particularly beneficial for multi-container setups like AI/ML workloads and microservices, where it improves scheduling accuracy, reduces resource overprovisioning, and enhances security. By defining aggregate resource needs, this feature addresses challenges where the total requirements of a pod exceed the sum of its individual container specifications, leading to better workload efficiency. This is one of 58 enhancements in the stability-focused v1.34 release.

Sources:

Google Kubernetes Engine (GKE) Expands Autopilot for Enhanced Cluster Management

Google Cloud has announced that the features of GKE Autopilot are now available to all qualifying GKE clusters, not just those in dedicated Autopilot mode. Autopilot is a managed Kubernetes service for GKE that offers a hands-off environment, automating tasks like provisioning compute capacity. This year, Autopilot’s autoscaling was upgraded to a dynamic, container-optimized compute platform that scales both horizontally and vertically. This update allows more users to benefit from simplified Kubernetes cluster operations and improved resource optimization. The feature is now generally available for clusters in the Rapid release channel running GKE version 1.33.1-gke.1107000 or later.

Sources:

IonQ Achieves Wavelength Conversion Breakthrough for Quantum Internet

IonQ has announced a significant technological milestone in the development of the Quantum Internet. In collaboration with the Air Force Research Lab (AFRL), the company has successfully demonstrated the frequency conversion of photons from the visible wavelengths used by their trapped-ion quantum computers into standard telecommunication wavelengths. This achievement is crucial for transmitting quantum information over existing long-haul fiber optic networks, paving the way for secure quantum communication between computers across vast distances. The breakthrough represents a major step in IonQ’s vision for creating distributed quantum computing networks.

Sources:

Europe Launches VLQ, Its Second Quantum Computer, in Czechia

The EuroHPC Joint Undertaking has inaugurated its second quantum computer, named VLQ, located in Ostrava, Czechia. Hosted and operated by the IT4Innovations National Supercomputing Center, the system is based on superconducting qubits and will feature 24 physical qubits in a star-shaped topology. This specific layout is designed to minimize swap operations, allowing for more efficient execution of complex quantum calculations. The VLQ will be integrated with the existing EuroHPC supercomputer, Karolina, to provide European researchers with access to a hybrid quantum-classical architecture. The system will support innovation in areas like Quantum Machine Learning (QML) and is expected to be available to users by the end of the year.

Sources:

CISA Warns of ‘Shai-Hulud’ Worm in Major Open Source Supply Chain Attack

The Cybersecurity and Infrastructure Security Agency (CISA) has issued a cybersecurity alert regarding a significant software supply chain compromise affecting the npm registry. A self-replicating worm, publicly identified as ‘Shai-Hulud’, has reportedly compromised over 500 packages within the world’s largest JavaScript registry, npmjs.com. CISA is providing guidance to address this widespread issue that has impacted the node package manager. This incident highlights the ongoing security vulnerabilities within the open-source software supply chain and the potential for broad impact across the developer community.

Sources:

WordPress Core Vulnerability (CVE-2025-58246) Exposes Sensitive Data

A medium-severity vulnerability, identified as CVE-2025-58246, has been discovered in Automattic’s WordPress content management system (CMS). The security flaw, described as an “Insertion of Sensitive Information Into Sent Data,” could allow an attacker to retrieve embedded sensitive data. According to the report, exploiting this vulnerability requires contributor-level privileges. The WordPress Core security team has acknowledged the issue and is actively working on a fix. This vulnerability affects WordPress versions up to 6.8.2.

Sources: