physical AI

Data centers are increasingly shaping the physical and economic landscape of the United States, as the boom in cloud computing and artificial intelligence drives construction of massive facilities across the country. Communities are debating their impact on land use, power grids and water resources, while states compete to attract the investment and jobs the facilities bring.

French AI startup Mistral AI has opened a new hub in Munich dedicated to physics-based and industrial AI applications. The German facility is expected to support research into AI for manufacturing, engineering and scientific use cases, strengthening Mistral's European footprint. The move signals the company's push beyond general-purpose chatbots into applied industrial technology, at a time when European AI firms are seeking to compete with US and Chinese rivals.

A Harvard study found that students using an AI tutor learned more physics in less time than peers in traditional classroom lessons. The research compared active learning methods with an AI-driven tutor and reported better results and higher engagement for the AI group. Educators are debating what this means for the future of teaching and the role of lecturers.

A Harvard study reportedly found that an AI tutor outperformed traditional in-class teaching in physics. According to the report, students learning physics with an AI tutor achieved better results than those taught through standard classroom instruction. The finding is being circulated as fresh evidence that artificial intelligence tools can match or exceed conventional teaching methods in science education, fueling debate about the future role of AI in universities and schools.

SiMa.ai, a chip developer focused on physical AI, has reached a valuation of $1.45 billion. The company builds semiconductor technology designed to bring AI processing to physical systems such as machines and edge devices. The valuation places it among the more prominent independent AI chip startups, in a sector drawing intense investor interest as companies race to supply hardware for artificial intelligence beyond data centers.

New research presented via EurekAlert describes how artificial intelligence paired with physics-based methods is advancing imaging techniques, including work that makes the invisible visible and automates the colorization of anime footage. The announcement highlights computational imaging as a growing field where machine learning and physical modeling intersect, with potential applications in science and animation production.

Black Forest Labs has released Flux 3 Action, a 7-billion-parameter open-weight world action model designed for robotics, published via Hugging Face. The model is intended to let robots perceive and act in physical environments, and its open weights make it available for researchers and developers to download, adapt and build on. Early discussion among machine learning practitioners focuses on what a major image-model lab entering robotics could mean for the field.

A Harvard study reports that students learning physics with an AI tutor outperformed peers in a traditional classroom setting. The findings suggest personalized, interactive AI instruction can deliver greater learning gains than standard lectures, reigniting debate over how AI should fit into education and what it means for the future of teaching.

Consultancy Bain & Company says artificial intelligence is reshaping retail in ways that will push underperforming stores out of the market. The firm argues retailers that fail to adopt AI-driven operations, pricing and customer insights risk losing ground to more efficient competitors, accelerating store closures across the sector.

South Korean robotics firm Angel Robotics has launched phai-x1, a wearable robot platform designed for physical AI research. The company is positioning the system as a tool for researchers developing AI that interacts with human movement, an emerging focus in robotics. Details on specifications, pricing and availability have not yet been widely reported.

Nvidia has reached a settlement in a trademark lawsuit concerning its 'Modulus' AI software, according to a Reuters report. The dispute centred on Nvidia's use of the Modulus name for its framework for building physics-based machine learning models. Terms of the settlement were not disclosed in the report, and the company has not issued a detailed public statement on the outcome.

Anthropic says its Claude AI solved a nine-loop particle physics challenge, an extremely complex calculation in quantum field theory. Nine-loop computations involve vast numbers of terms and are traditionally done with specialized software and significant human effort, so a successful AI solution would mark a milestone for AI in theoretical physics. The claim is drawing attention from researchers assessing how far AI models can go in advanced scientific work.

Faraday Future has announced a strategic transformation into a robotaxi and EAI cabin technology operator and a physical AI investment holding company. The company plans to combine its robotics business at an approximate $200 million valuation with AIxC, to be renamed FFR, for a standalone listing. The move marks a major shift in strategy for the struggling electric vehicle maker as it seeks new growth in artificial intelligence.

Bloomberg reports that driverless trucks may provide a model for how autonomous AI systems can be deployed safely without harming people. The argument is that self-driving freight has operated under strict testing, regulation and limited-use conditions, offering a template for governing more advanced robots. The piece adds to ongoing debate about how to manage AI risks as automation expands into physical industries.

An AI-assisted study using data from the Large Hadron Collider is challenging the theory that the universe has ten dimensions, a cornerstone idea from string theory. The findings suggest the extra dimensions predicted by the theory may not exist, prompting debate among physicists about the model's validity and the growing role of machine learning in fundamental physics research.

Scientific American's latest science roundup covers the rising energy demands of artificial intelligence, ongoing efforts to detect subatomic particles, and research into an amoeba that thrives at extreme temperatures. The episode ties together the environmental cost of the AI boom with fundamental physics and biology discoveries, offering a snapshot of current conversations across science news.

A project called Roboharm is asking whether frontier robot policies actually refuse unsafe instructions. The work examines how AI systems controlling robots respond to harmful commands, a safety question increasingly urgent as AI models are deployed in physical robotics. Discussion is focused on how well current safeguards carry over from language models to embodied systems.

Gecko Robotics' chief executive is arguing that humans must remain in control as artificial intelligence moves into the physical world through robots and industrial automation. The company builds robots used to inspect critical infrastructure such as bridges, power plants and pipelines. The intervention adds a robotics industry voice to the ongoing debate over how much autonomy should be granted to AI-driven machines in safety-critical settings.

The European Union's updated product liability framework is widening its scope so claims no longer hinge solely on physical damage. Under the new rules, compensation can cover data loss, psychological harm and damage from software and AI-enabled products. Commentators say the shift significantly raises exposure for manufacturers and tech firms selling into the EU market.

Anti-AI protesters rushed the stage at Data After Dark, a pro-AI and pro-data center event held in Washington, DC. The demonstrators were unable to deliver a clear statement after the crowd drowned them out with chants of 'USA! USA!'. The confrontation highlights growing tensions between opponents of AI and data center expansion and their supporters.

Waddle Labs and RoboCurve are the focus of a new analysis arguing that general-purpose large language models could outperform specialised systems in robotics. The piece contends that broad foundation models trained on diverse data may adapt better to physical tasks than narrowly built robots, a view that challenges a common assumption in the field and is drawing attention among robotics and AI investors.

A British startup is turning video game controller inputs into training data for AI models designed to navigate the physical world. The idea is that human gameplay, including mistakes and clumsy manoeuvring, offers rich examples of how people learn to control complex systems, which could help robots and other embodied AI improve. The approach is drawing attention as a cheaper alternative to real-world data collection.