Capacity development

Power grid spending surges as US braces for data centre and AI boom

US electric utilities are set to spend nearly $208 billion on the power grid in 2025 and more than $1.1 trillion over the next five years, according to the Edison Electric Institute. The surge in investment reflects rising demand from data centres, artificial intelligence, and wider electrification across the economy.

EEI data shows that investor-owned utilities spent $765 billion on capital projects in the five years to 2024. The new spending represents a significant increase and is aimed at upgrading and expanding infrastructure to keep pace with the accelerating demand for electricity.

The growing investment comes as demand from energy-intensive technologies continues to rise. Data centres and AI workloads are driving sustained growth in US power consumption, placing unprecedented pressure on existing infrastructure and prompting utilities to scale up their spending plans.

David Weeks, supply chain industry practice lead at Moody’s, warned that the escalating energy crisis could become a limiting factor across multiple industries. He said grid constraints and permitting delays must be factored into corporate supply chain strategies to avoid future disruptions.

As electrification spreads across the economy, grid reliability and capacity are becoming critical considerations for companies. The planned investment underscores the urgency of modernising the power grid to support economic growth while adapting to new technological demands.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

Machine learning helps prevent disruptions in fusion devices

Researchers at MIT have developed a predictive model that could make fusion power plants more reliable and safe. The approach uses machine learning and physics-based simulations to predict plasma instabilities and prevent damage during tokamak shutdowns.

Experimental tokamaks use strong magnets to contain plasma hotter than the sun’s core. They often face challenges in safely ramping down plasma currents that circulate at extreme speeds and temperatures.

The model was trained and tested on data from the Swiss TCV tokamak. Combining neural networks with physics simulations, the team achieved accurate predictions using few plasma pulses, saving costs and overcoming limited experimental data.

The system can now generate practical ‘trajectories’ for controllers to adjust magnets and temperatures, helping to safely manage plasma during shutdowns.

Researchers say the method could be particularly important as fusion devices scale up to grid-level energy production. High-energy plasmas in larger reactors pose greater risks, and uncontrolled terminations could damage the machine.

The new model allows operators to carefully balance rampdowns, avoiding disruptions and ensuring safer, more efficient operation.

Work on the predictive model is part of wider collaboration with Commonwealth Fusion Systems and supported by the EUROfusion Consortium and Swiss research institutions. Scientists see it as a crucial step toward making fusion a practical, reliable, and sustainable energy source.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot! 

MIT AI reveals how antibiotic targets Crohn’s bacteria

MIT and McMaster researchers used AI to map how a narrow-spectrum antibiotic attacks harmful gut bacteria. Enterololin targets E. coli linked to Crohn’s flares while preserving most of the microbiome, providing a precise alternative to broad-spectrum antibiotics.

AI accelerated the process of identifying the drug’s mechanism of action, reducing a task that usually takes years to just months.

The team used DiffDock, a generative AI tool developed at MIT, to predict how enterololin binds to a protein complex called LolCDE in E. coli. Laboratory experiments, including mutant evolution, RNA sequencing, and CRISPR knockdowns, confirmed the AI predictions.

The method demonstrates how AI can provide mechanistic insights, guide experiments, and speed up early-stage antibiotic development.

Enterololin improved recovery and preserved the microbiome in mouse models compared with conventional treatments. Researchers aim to develop derivatives against resistant pathogens like Klebsiella pneumoniae, with early work underway at spinout company Stoked Bio.

The study highlights broader implications for precision antibiotics, which could treat infections without disrupting beneficial microbes. AI-driven mechanism mapping could speed up drug discovery, cut costs, and help tackle antimicrobial resistance.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot! 

Google expands Gemini-powered AI Search across the globe

Google has expanded its AI Mode in Search, supporting over 35 new languages and 40 more countries and territories. The rollout expands access across Europe and other regions, reaching over 200 countries and territories worldwide.

The update aims to make AI-powered Search more accessible globally, allowing people to interact with Search in their native language. Expanding language support, Google will enable users to ask questions and access information in their preferred language.

AI Mode is powered by Google’s latest Gemini models, which deliver advanced reasoning and multimodal understanding. These capabilities help the system grasp the subtleties of local languages and provide relevant, context-aware answers, making AI Mode genuinely useful across diverse regions.

According to Google, people using AI Mode tend to explore topics in far greater depth, with queries nearly three times longer than traditional searches. The enhanced experience will continue to roll out globally over the coming week.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Nobel Prize awarded for tunnelling in superconducting circuits

The 2025 Nobel Prize in Physics has been awarded to John Clarke, Michel Devoret and John Martinis for their experiments that brought quantum mechanical effects into macroscopic systems.

The Royal Swedish Academy of Sciences cited their ‘discovery of macroscopic quantum mechanical tunnelling and energy quantisation in an electric circuit’.

In classic quantum mechanics, particles can sometimes cross through energy barriers via a process known as tunnelling, behaviour that typically occurs at atomic or subatomic scales.

The laureates’ work showed that such quantum phenomena can appear in larger electrical circuits using superconductors and Josephson junctions, systems that were thought to be firmly in the domain of classical physics.

Their experiments (conducted in the mid-1980s) involved circuits made of superconducting materials separated by a thin insulating barrier. By finely tuning currents and electromagnetic stimuli, they were able to force a system to switch between zero-voltage and finite voltage states, essentially demonstrating that the circuit could ‘tunnel’ from one state to another.

In addition, they demonstrated energy quantisation in these systems, that the circuits absorb and emit energy in discrete packets, consistent with quantum theory.

This work is widely viewed as a foundational bridge between theoretical quantum mechanics and practical quantum technology. Superconducting circuits (such as qubits in quantum computers) rely on precisely these kinds of effects, and the laureates’ results helped validate the notion that quantum engineering is possible in engineered devices.

As the Nobel announcement puts it, their experiments ‘revealed quantum physics in action’ in a device small enough to hold in hand.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot! 

Study shows how AI can uncover hidden biological mechanisms

Researchers in China have used AI to reveal how different species independently develop similar traits when adapting to shared environments. The study focuses on echolocation in bats and toothed whales, two distant groups that created this ability separately despite their evolutionary differences.

The Institute of Zoology, Chinese Academy of Sciences team found that high-order protein features are crucial to adaptive convergence. Convergent evolution is the independent emergence of similar traits across species, often under similar ecological pressures.

Led by Zou Zhengting, the researchers developed a framework called ACEP, which utilises a pre-trained protein language model to analyse amino acid sequences. This method reveals hidden structural and functional information in proteins, shedding light on how traits are formed at the molecular level.

The findings, published in the Proceedings of the National Academy of Sciences, reveal how AI can detect deep biological patterns behind convergent evolution. The study demonstrates how combining AI with protein analysis provides powerful tools for understanding complex evolutionary mechanisms.

Zou said the work deepens the understanding of life’s evolutionary laws and highlights the growing role of AI in biology. The team in China hopes this approach can be applied to other evolutionary questions, broadening the use of AI in decoding life’s hidden patterns.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

Study explores AI’s role in future-proofing buildings

AI could help design buildings that are resilient to both climate extremes and infectious disease threats, according to new research. The study, conducted in collaboration with Charles Darwin University, examines the application of AI in smart buildings, with a focus on energy efficiency and management.

Buildings account for over two-thirds of global carbon emissions and energy consumption, but reducing consumption remains challenging and costly. The study highlights how AI can enhance ventilation and thermal comfort, overcoming the limitations of static HVAC systems that impact sustainability and health.

Researchers propose adaptive thermal control systems that respond in real-time to occupancy, outdoor conditions, and internal heat. Machine learning can optimise temperature and airflow to balance comfort, energy efficiency, and infection control.

A new framework enables designers and facility managers to simulate thermal scenarios and assess their impact on the risk of airborne transmission. It is modular and adaptable to different building types, offering a quantitative basis for future regulatory standards.

The study was conducted with lead author Mohammadreza Haghighat from the University of Tehran and CDU’s Ehsan Mohammadi Savadkoohi. Future work will integrate real-time sensor data to strengthen building resilience against future climate and health threats.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!

Could AI win a Nobel Prize? Experts debate the possibility

AI is starting to make inroads into scientific discovery. In recent years, AI systems have analysed data, designed experiments, and even proposed hypotheses and behaviours once thought to be uniquely human.

Some researchers now argue that AI could compete with leading scientists, conceivably worthy of a Nobel Prize in a few decades. The ambition invites provocative questions: Can a machine be an author or laureate? What criteria would apply? Would human oversight remain essential?

Sceptics argue that AI lacks consciousness, intentionality or moral agency, all hallmarks of great scientific insight. They caution that the machine’s contributions are derivative, built on human data, models and frameworks. Others contend that denying AI recognition blocks a future where hybrid human-machine teams deliver breakthroughs.

Meanwhile, mechanisms for attributing credit are also under scrutiny. Would the institution or the engineers who built the AI deserve the credit, or the AI itself? The article notes existing examples: AIs have already co-authored papers and databases in genetics or materials science. However, instituting them as Nobel candidates demands shifting philosophical and institutional norms.

As AI systems achieve deeper autonomy, the debate over their role in science and whether they merit high honours will only intensify. The Nobel Prize, a symbolic instrument in the science ecosystem, may evolve to include nonhuman actors if the community permits it.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot! 

Anthropic launches Bengaluru office to drive responsible AI in India

AI firm Anthropic, the company behind the Claude AI chatbot, is opening its first office in India, choosing Bengaluru as its base.

A move that follows OpenAI’s recent expansion into New Delhi, underlining India’s growing importance as a hub for AI development and adoption.

CEO Dario Amodei said India’s combination of vast technical talent and the government’s commitment to equitable AI progress makes it an ideal location.

The Bengaluru office will focus on developing AI solutions tailored to India’s needs in education, healthcare, and agriculture sectors.

Amodei is visiting India to strengthen ties with enterprises, nonprofits, and startups and promote responsible AI use that is aligned with India’s digital growth strategy.

Anthropic plans further expansion in the Indo-Pacific region, following its Tokyo launch, later in the year.

Chief Commercial Officer Paul Smith noted the rising demand among Indian companies for trustworthy, scalable AI systems. Anthropic’s Claude models are already accessible in India through its API, Amazon Bedrock, and Google Cloud Vertex AI.

The company serves more than 300,000 businesses worldwide, with nearly 80 percent of usage outside the US.

India has become the second-largest market for Claude, with developers using it for tasks such as mobile UI design and web app debugging.

Anthropic also enhances Claude’s multilingual capabilities in major Indic languages, including Hindi, Bengali, and Tamil, to support education and public sector projects.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

AI maps over 1,300 mouse brain subregions with unprecedented precision

Researchers at UCSF and the Allen Institute have created one of the most detailed mouse brain maps. Their AI model, CellTransformer, identified over 1,300 brain regions and subregions, including previously uncharted areas. The findings were published in Nature Communications.

CellTransformer utilises spatial transcriptomics to define brain regions based on shared cellular patterns, rather than relying on expert annotation. Drawing city borders from building types reveals finer brain structures. This data-driven method provides unprecedented precision.

The model replicated known regions, such as the hippocampus, and revealed previously unknown subdivisions in the midbrain reticular nucleus. Researchers compared the leap from mapping continents to mapping states and cities. The tool provides a foundation for more targeted neuroscience studies.

Validation against the Allen Institute’s Common Coordinate Framework strongly aligned with expert-defined anatomy. The results gave researchers confidence in the biological relevance of the new subregions. Further studies will investigate their functions.

The model’s potential goes beyond neuroscience. Its methods can map other tissues, including cancers, by analysing large spatial transcriptomics datasets. However, this could support new medical research, helping uncover disease mechanisms and accelerate treatment development.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!