
Artificial intelligence arrives on our screens almost without weight. A sentence materialises. An image appears. A model reasons through a problem in seconds. The interface encourages a seductive fiction: intelligence has escaped matter. It lives somewhere called the cloud. The economics now reveal the opposite. AI is becoming one of the most physically demanding capital projects of the modern era. In April, the International Energy Agency reported that capital expenditure among five large technology companies exceeded $400 billion in 2025 and was expected to increase by another 75 per cent in 2026. This month, Nvidia announced arrangements with major financial institutions intended to mobilise more than $500 billion of third-party capital for AI infrastructure. Alphabet, meanwhile, has returned repeatedly to debt markets as technology companies finance an AI investment cycle that Reuters says could push sector spending beyond $730 billion this year. Money is only the beginning. Intelligence at industrial scale requires semiconductors, servers, transformers, substations, transmission networks, cooling equipment, water, land, concrete, skilled labour and — above everything — electricity. Data-centre electricity demand rose 17 per cent in 2025, according to the IEA, while AI-focused facilities grew faster still. The agency now expects data-centre electricity consumption to double by 2030, with electricity use at AI-focused centres potentially tripling. Then comes the environmental contradiction. A Financial Times analysis of 60 large planned American data-centre projects estimates potential annual emissions of approximately 101.5 million tonnes of carbon dioxide if their projected electricity requirements are supplied under anticipated generation conditions. Utilities are adding gas capacity, and some coal retirements are being delayed as electricity demand accelerates. We called it artificial intelligence. The infrastructure required to produce it is brutally physical. The consequential AI story is therefore no longer merely which model can reason fastest, generate the best video or dominate the next benchmark. The deeper story is the emergence of an industrial system capable of reorganising capital, electricity, land, supply chains and geopolitical power around the production of machine intelligence. The cloud has touched the ground. And what it is building there may prove considerably more important than the chatbot.

Two sisters living with an exceptionally rare growth disorder have become part of a much larger scientific question: what if one of the most useful ways to understand cancer is to study human bodies in which some of the biological conditions cancer exploits have been altered from birth? The condition is Laron syndrome, a form of growth-hormone insensitivity caused by dysfunction of the growth-hormone receptor. The resulting biology includes extremely low levels of insulin-like growth factor 1, or IGF-1, an important regulator of growth, metabolism, and cellular activity. Researchers studying people with Laron syndrome, particularly a distinctive cohort in Ecuador, have repeatedly reported unusually low incidences of cancer and diabetes compared with relatives and wider populations. The finding has attracted scientific attention for decades. But “The Cancer Twins” must resist the temptation that makes the story irresistible online. Laron syndrome does not establish human immunity from cancer. Cancer has occurred within the Ecuadorian population, and the evidence concerns substantially diminished incidence, not biological invulnerability. A 2023 review of the Ecuadorian cohort explicitly documented cancer cases while maintaining the larger finding of unusually low cancer incidence. The more consequential proposition lies underneath the headline. Cancer requires more than mutations. Malignant cells must obtain signals, energy, metabolic support, opportunities to proliferate, and mechanisms for avoiding destruction. By studying people whose growth-signalling architecture differs markedly from the norm, researchers may gain an unusual natural experiment into the conditions that make malignant growth easier — or harder. The twins, then, are not scientific curiosities. Nor are they miracle patients. They represent something more intellectually valuable: a human biological exception capable of revealing the rules governing the rest of us.

A flying umbrella sounds like the sort of invention engineered primarily for social-media virality. A canopy hovers above its user, follows as they walk, and removes the small inconvenience of holding something over one’s own head. Charming, certainly. Civilisational breakthrough? Not quite. But the machine is interesting for precisely the reason the umbrella itself is not. The experimental system associated with Canadian engineer John Tse points towards a much larger design transition: robots are beginning to leave the category of objects we deliberately operate and enter the environment around us as responsive infrastructure. The umbrella is merely an unusually legible prototype of that future. Robotics has traditionally announced itself. Industrial robots occupy cages. Domestic robots are recognisable appliances. Drones require pilots, applications, controllers, or predefined missions. The next generation will increasingly sense context, understand intent, maintain spatial relationships, and act with less explicit instruction. A machine that knows where you are, understands that its purpose is to remain above you, and continuously adjusts itself as you move represents a rudimentary version of ambient robotics. That transition is being enabled by the convergence of computer vision, depth sensing, artificial intelligence, lightweight propulsion, localisation, batteries, edge computing, and increasingly capable autonomous-control systems. None is revolutionary in isolation. Their combination changes the relationship between people and machines. The profound question, therefore, is not whether anybody needs a flying umbrella. It is what happens when the physical world begins to follow, anticipate, reposition, and respond to us.

For most of aviation history, human flight has required an aircraft: a machine large enough to generate lift, carry fuel, accommodate passengers, and surround its occupants with an engineered structure. Emerging personal-flight technologies are beginning to loosen that relationship. Jet suits, powered wings, compact electric vertical-lift systems, autonomous drones, and increasingly sophisticated flight-control technologies suggest that aviation may eventually encompass machines worn, mounted, or summoned rather than conventionally boarded. The viral spectacle is irresistible. A person rises from the ground, accelerates over water, and appears to have acquired a superpower. Yet spectacle obscures the engineering. Human-scale powered flight confronts brutal constraints involving energy density, heat, noise, stability, endurance, payload, weather, redundancy, training, regulation, and the consequences of mechanical failure. A technology can fly successfully and still be unsuitable for mass transportation. That distinction is central to understanding personal aviation. The most plausible near-term applications are unlikely to involve commuters casually flying between homes and offices. Specialist environments — emergency response, defence, offshore infrastructure, inaccessible terrain, inspection, rescue, and certain industrial operations — provide a more credible pathway because the economic value of reaching somewhere quickly can outweigh the technology’s considerable limitations. The deeper development, however, extends beyond jet suits. Aviation is becoming computational. Sensors can stabilise machines faster than human reflexes. Software can continuously adjust thrust. Lightweight materials reduce mass. Autonomous navigation increasingly separates piloting from constant manual control. Electric propulsion enables aircraft configurations that would have been impractical under traditional mechanical architectures. The result is not necessarily the death of the aeroplane. Commercial aircraft remain extraordinarily efficient at moving large numbers of people over long distances. Instead, aviation may be fragmenting into a richer ecosystem: aircraft for distance, drones for autonomous logistics, eVTOL systems for specialised regional movement, and wearable or highly compact systems for particular human-scale missions. The important question is therefore no longer simply, “Can a person fly without an aeroplane?” We already know that certain machines can make that possible. The better question is: when does removing the aircraft make flight more useful?

On 9 July 2026, the United States Federal Communications Commission granted Reflect Orbital conditional authority to deploy and operate Eärendil-1, a single experimental satellite designed to test whether a steerable, 18-metre reflector can redirect sunlight towards a targeted area on Earth after sunset. Supporters see a new form of infrastructure: controllable natural light that could extend solar-energy production, support emergency operations, and illuminate remote sites without installing poles, cables, or generators. Astronomers, dark-sky advocates, environmental organisations, and public-health specialists see a different possibility: a commercial precedent for altering a planetary condition that no company created, no nation owns, and countless species require. The immediate experiment is small. The question beneath it is not. Once darkness can be scheduled, directed, sold, and delivered from orbit, night ceases to be merely the absence of daylight. It becomes a governed resource. This is therefore not simply a story about an inventive satellite. It is a test of whether regulation can keep pace when commercial technology begins redesigning the natural environment itself.

For nearly four billion years, evolution operated according to a single principle: biological organisms adapted to changing environments through natural selection. Humanity may now be approaching the end of that era. Artificial intelligence, robotics, quantum computing, neural interfaces, biotechnology, and human augmentation are converging into a technological ecosystem unlike anything previously observed in evolutionary history. The significance extends beyond innovation. For the first time, a species has acquired the capacity to redesign itself. The next evolutionary transition may not emerge through genetics alone but through integration—human cognition enhanced by machine intelligence, biological systems connected to digital networks, and autonomous technologies capable of learning, adapting, and collaborating alongside their creators. The result may not be an improved version of Homo sapiens. It may be something fundamentally different. The question is no longer whether humanity will change. The question is whether humanity recognises that a new species may already be emerging.