A grid afraid of too much power, the sky's water highways, and good faith in the boardroom

The trigger was mundane. A lightning arrestor on a 230-kilovolt line failed, and the line's automatic reclosing controls, the switches that try to restore a faulted line by snapping it back into service, produced six successive faults in eighty-two seconds. [1] Each fault dragged local voltage down to between a quarter and two-fifths of normal for forty to sixty-six thousandths of a second, brief blinks that most equipment simply rides out. [1] Data centers do not. Their uninterruptible power supplies, the battery systems designed to shield servers from exactly this kind of disturbance, transferred the computing load off the grid; many run a counting scheme that treats three dips inside a minute as reason to stay on backup until a person manually reconnects. [1] About 1,500 megawatts of demand vanished at once, and roughly 1,260 megawatts of it stayed dark for hours. [1]

Losing a large block of load is the mirror image of losing a power plant, and the grid is engineered mainly for the latter. [3] The system holds a constant balance between what generators produce and what customers draw; when demand disappears in milliseconds, the turbines still spinning have too little to push against and speed up, nudging frequency above its narrow band around sixty hertz. [3] In the 2024 event frequency rose to 60.047 hertz and voltage climbed about seven percent above normal before operators removed capacitor banks and backed down generation to settle it. [1] Engineers call the behavior they want ride-through, the willingness of a load to stay connected through a brief sag rather than bailing out. [3] The very equipment that protects the servers is, from the grid's point of view, badly behaved.

For the first time, federal regulators are moving to write that behavior into binding rules. [2] On July 16, 2026, the Federal Energy Regulatory Commission directed the North American Electric Reliability Corporation, the body that sets enforceable grid standards across the continent, to draft mandatory standards for what it calls computational loads, meaning data centers, large computing campuses and cryptocurrency mines, and to require that such operators register as regulated entities for the first time. [2] The reliability corporation has until the final day of 2026 to file the standards, with a workplan for a second phase due by March 2027. [4] The order follows a rare Level 3 alert, the corporation's highest tier of urgency, issued in early May 2026, which told planners and grid operators to model these loads at sub-second timescales and to open direct communication lines with the facilities. [3]

The urgency tracks the arithmetic of growth. The reliability corporation's most recent ten-year outlook projects summer peak demand rising by 224 gigawatts, about 24 percent, a figure nearly 70 percent higher than the previous year's forecast, with new large computing loads accounting for most of the increase. [7] A single large campus can now draw roughly as much power as a nuclear unit, which means the accidental loss of one cluster lands on the grid the way the sudden loss of a major generator once did, except that nothing tripped on the utility's side of the meter. [1]

What remains unsettled is how fast the fix can be built and who pays for it. Rewriting the standards is the straightforward part; retrofitting ride-through behavior into thousands of installed backup systems, and deciding whether operators or ratepayers absorb the cost, is not resolved by the July order, which sets the mandate and hands the engineering to the corporation's stakeholder process. [2] The established facts are narrow and firm. The disturbances happened, the abrupt loss of load stressed the system, and these loads will now sit under enforceable reliability rules. [1] Whether those rules are in force before the next cluster blinks off is the open question. [3]

Atmospheric rivers are long, narrow corridors of concentrated water vapor that stream out of the tropics toward the poles. They are slender by planetary standards, covering less than a tenth of Earth's surface at any moment, yet they carry roughly 90 percent of the moisture moving from the tropics into the middle latitudes. [1] Where they make landfall they can be a blessing or a catastrophe, supplying much of the water that fills reservoirs in places like the American West while also delivering that region's most destructive floods. [4]

The team, led by physicist Tobias Braun of the Potsdam Institute for Climate Impact Research, took 84 years of atmospheric records, from 1940 through 2023, and treated each atmospheric river as a journey with a beginning and an end. [1] They divided the globe into hexagonal cells of about 87,000 square kilometers and borrowed the mathematics of network science, the same graph theory that navigation apps use to find routes, to test whether the tracks clustered along shared paths. [1] They did. Rather than a smear of independent storms, the analysis found a sparse skeleton of recurring corridors linking distant regions, built from two catalogs of storm tracks derived from the ERA5 reanalysis. [3]

Four great highways emerged, aligned with the mid-latitude storm tracks: one crossing the North Pacific from East Asia to the American West, one spanning the North Atlantic toward Europe, and two threading the Southern Hemisphere. [1] Along them sit what the researchers describe as fueling stations, places where a passing river draws in fresh moisture and intensifies, among them the east coast of Australia, parts of central Asia, and the southern tip of South Africa. [2] Other stretches do the opposite, sapping a river's strength, and the polar regions act as collecting points where persistent flows pile up. [1]

The practical promise is in forecasting. If the highways and their seasonal shifts are known, Braun says, a flood in Europe becomes the downstream end of a pathway that can be followed back across the Atlantic, offering earlier evidence of where a given river is heading. [2] Kimberley Reid, an atmospheric scientist at the University of Melbourne who was not involved in the work, said the map captures where these systems typically grow and decay, a step toward better preparing for the extremes they bring. [2] She cautioned that individual rivers still stray from the mapped routes, and that the network itself will likely redraw as the planet warms. [2]

Atmospheric rivers already govern floods, droughts, and water supply across much of the world, and their reach extends to polar ice melt. [4] A structural map of how they connect does not by itself sharpen any single forecast, but it reframes the phenomenon as a system with discernible plumbing, which is usually where useful prediction begins. [1]

The company was Spring Media Investments, a creative-services group, and the chairman was Francesco Costa. A shareholders' agreement signed in 2016 bound the company and its investors to work together in good faith toward a sale, called an Exit, no later than the end of 2019, with an investment bank to be engaged if the deadline arrived unmet. Costa alone controlled the sale process, believed a later sale would fetch more, and steered around the timetable while misleading the board about whether the agreement was being honoured. The deadline passed, the pandemic then destroyed the prospect of a profitable sale, and a shareholder holding just over 22 per cent, Saxon Woods, sued.[2][5]

Section 172 of the Companies Act 2006 tells a director to act in the way he considers, in good faith, would be most likely to promote the success of the company. The trial judge read those words narrowly. He found that Costa genuinely believed delay served the company, and so held there was no breach of the good-faith duty, even while finding that the shareholders had been unfairly prejudiced. The Court of Appeal reversed, reasoning through the objective test for dishonesty drawn from Ivey v Genting Casinos. That left a clean question for the Supreme Court. Does good faith under section 172 ask only whether a director honestly believed in his goal, or also whether the means he used to reach it were loyal.[1][3]

Good faith, the Court held, extends not merely to the director's thinking but also to his conduct.

The five justices dismissed Costa's appeal unanimously, with Lord Briggs writing. Sincere conviction, the Court held, is no shield for covert disloyalty. A director who privately disagrees with the board must bring that view into the boardroom, not run his own strategy behind it, because allowing individual directors to subvert collective decisions would invite chaos and paralysis in corporate governance. The reasoning rested less on the words of the 2006 statute than on the equity beneath them. Because the Act directs that its codified duties be read the same way as the older common-law and equitable rules, the Court traced the point to a line of authority, reaching back to an 1878 ruling by Sir George Jessel, holding that a fiduciary cannot answer a charge of disloyal conduct simply by asserting an honest state of mind, alongside the modern cases requiring directors to be candid with their own boards.[1][4][6]

Having found a breach, the Court turned to the remedy under the unfair-prejudice provisions, sections 994 to 996, which let a court order one shareholder to buy out another when a company's affairs have been conducted unfairly. The trial judge, believing there was no fiduciary breach, had made only a conditional buy-out, requiring Saxon Woods to prove later that a sale worth more than 75 million US dollars would have closed by the end of 2019. The Court of Appeal removed that condition, and the Supreme Court, dismissing Costa's appeal, left the unconditional order in place. Costa must buy Saxon Woods' stake at its undiscounted pro rata value as of the end of 2019, the point at which, had he followed the agreed plan, a sale would likely have completed before the pandemic erased the company's worth.[2]

The mechanism worth watching is how a bare statutory phrase acquires its content. Parliament wrote good faith into the Companies Act in 2006 without defining it, and the Court has now confirmed that the phrase carries the full weight of the equitable loyalty that preceded it, reaching conduct and not only belief. For minority shareholders, that closes a gap a controlling director might otherwise have exploited, treating private sincerity as cover for steering a company off a course its owners had agreed by contract. For boards, the ruling restates a plainer rule of collective governance, that a director who loses an argument inside the room may not then go on to win it outside.[3][4]