Are data centres in space less crazy than we think?

I enjoyed your piece on why data centres in space are less crazy than we might think (“Orbital number-crunching”, March 7th). However, you underestimated the cost and supply constraints of space-grade solar generators. Satellite-design costs could fall to “less than $5 per watt”, you say. In reality, the baseline price for a space-grade photovoltaic triple-junction cell is closer to $1,000 per watt, and supply is highly constrained.

It is true that massive, speculative demand could eventually drive economies of scale for cheaper silicon-based cells ($20 to $100 per watt). But these cells operate at only two-thirds the efficiency of triple-junction cells, suffer from uncertain long-term radiation degradation, and lack the manufacturing base for high-quantity procurement.

Also, your assumptions regarding thermal management require careful scrutiny. The International Space Station already utilises extremely efficient, state-of-the-art radiator systems. Although it is true that an unmanned data-centre satellite could improve specific power by stripping away the heavy fail-safes required for human life support and for equipment to operate at high temperatures, bridging the remaining gap to a tenfold increase in heat dissipation per kilogram remains speculative.

Finally, the competitive model relies on launch costs falling dramatically to $500 per kg. Next-generation heavy-lift vehicles may eventually alter the economics, but SpaceX currently holds a quasi-monopoly in the market. In real terms, the actual costs for rideshare and dedicated Falcon 9 launches have remained flat for the past six years.

All these realities suggest a realistic timeline is closer to 15 years, rather than the two to three years being touted.

You are right that the mindset in Europe is changing with regard to AI (“Breeding Eunicorns”, March 7th). For European tech to stand a real chance, three things need to happen. First, AI startups must form partnerships with strong traditional industries, the backbone of many European economies. If they leverage industrial data meaningfully, that’s when the party starts. Second, governments across the EU need to bundle their activities instead of creating small AI clusters here and there. Third, law enforcers must ensure open and competitive markets.

Governments are still blindly relying on big tech, thereby perpetuating dependencies and lock-ins. Free competition drives innovation, not the puny hope of being integrated into the ecosystem of a big tech player.

Rupprecht Podszun
Co-chair
German government commission on competition and AI
Düsseldorf

“Rocks around the clock” (February 28th) reported on America’s race for rare earths, and briefly mentioned the Lobito Corridor in Africa. The Lobito Corridor rail system, connecting the coppery Congolese provinces of Katanga and Lualaba to Angola’s deepwater port at Lobito, ought to be an urgent investment priority for America. The basic infrastructure is already in place, a capable Western-owned company is operating the tracks, it terminates in Angola, which is increasingly America-friendly, and when fully operational it will slash the time and cost of exporting minerals from Congo and Zambia.

Those exports currently rely on gruelling multi-week truck journeys to ports on Africa’s east coast, most of which are funded or operated by Chinese firms. If America follows through on its big (and, so far, miraculously bipartisan) commitment to bring the colonial-era Lobito Corridor back to life, it could be its smartest move yet.

Sam Williams
Producer of “Lobito Bound”
London

You raise valid concerns about the Trump administration’s potentially “scattergun approach”. The rush to subsidise new mines risks repeating the same supply-chain vulnerabilities it seeks to solve. Expanding extraction without commensurate investment in domestic processing and recovery of rare-earth magnets from e-waste and defence infrastructure would still mean America has to ship materials to China for refining, undermining the point of the exercise.

Part of the American policy response is beginning to address this gap. Roughly $1bn is planned in funding to strengthen domestic supply-chains for critical minerals and materials, including up to $135m to demonstrate domestic refining and recovery of rare-earth elements from tailings and waste streams, alongside support through R&D programmes.

Recycling and recovery deserve far more attention because they represent the fastest pathway to new supply. New mines typically take a decade or more to permit, finance and build. By contrast, large quantities of rare-earth magnets already exist in end-of-life electronics, data-centre infrastructure, electric motors and defence systems, providing potential near-term supply while mining and refining capacity scale.

This is particularly important for dysprosium and terbium, which are scarce in most known American deposits but widely present in technologies already operating within the American economy. Historically less than 1% of rare-earth metals have been recycled, creating an untapped economy of rare earths for recovery. Recycling will not meet all demand, but alongside new mining and refining capacity it can provide a cleaner and faster bridge toward the resilient supply chain America seeks.

Luke Wray
Vice-president of critical minerals and defence
Paladin EnviroTech
Tampa, Florida

You wrote about how AI tools are being prepared for the physical world (“I can show you the world”, February 28th). The article treats grocery shopping as an optimisation problem awaiting automation. Retailers may tremble at the thought. Supermarkets are not just warehouses with lighting. They are carefully choreographed arenas of temptation. The smell of baking bread, the strategically placed confectionery and the serpentine layout are designed not design accidents but engines to maximise profits. Business schools teach sensory marketing, shelf placement and psychological merchandising precisely because humans are gloriously irrational. A perfectly rational humanoid dispatched with a shopping list would stride past promotional displays, ignore end-of-aisle seductions and return with exactly what was requested. Basket sizes would shrink, along with margins. “World models” may help robots navigate aisles. Whether they can replicate the human susceptibility that keeps those aisles profitable is another matter.

Your article correctly argues that higher income taxes are a poor instrument for extracting revenue from the ultra-wealthy, whose fortunes derive largely from appreciating assets rather than wages (“The Robin Hood state”, February 21st). We should indeed be sceptical about New York City’s proposed levy and California’s wealth tax.

Yet after acknowledging the “outrageous” loophole that allows the ultra-rich to avoid capital gains if assets are held until death, even as they borrow against the assets to fund spending, you dismiss reform because fixing this would raise only around 0.1% of GDP a year. This conflates two separate questions: whether such reform would solve budget pressures, and whether the current arrangement is defensible.

A tax system that permits the wealthiest Americans to accumulate and transmit dynastic wealth largely untaxed sits uneasily with your claim that redistribution has kept pace with inequality. Moreover, by pointing to increased income taxes on the rich as evidence for this redistribution you make the same error that New York and California have made—that taxing the incomes of the rich is relevant to the wealth gap.

The problem is not that politicians want to tax the rich but that they are reaching for the wrong tools. Closing gaps to address the structural asymmetries of the taxation of capital versus labour would strengthen the coherence, perceived legitimacy and fairness of our tax system.

I fear your merry band of economists are aiming at the wrong target. Although it is true that income inequality after taxes and benefits is falling in many countries, it is also the case that wealth inequality is rising in those same countries when land, property and pension wealth is assessed. You should take the lead from Lee Kuan Yew, Singapore’s founder, who changed the land-acquisition laws and cleared compensation for sea frontages in order to reclaim the land. “It was Robin Hood but I succeeded in giving everyone their own home,” he said. With 90% of the land in public ownership, about half of Singapore’s public revenue is derived from land and property values.

On your other assertion, although a few entrepreneurs and rentiers may leave Britain if income taxes become more progressive, the trouble with an increase to broad-based taxes such as pay-as-you-earn and national insurance is that the population at large suffers. As we can see since the rise in national-insurance contributions last year, unemployment is rising and many small businesses are pulling back on investment, putting off new hiring and suffering falling profits, tipping many of them into administration.

Dr Andrew Purves
Honorary research fellow
Bartlett School of Planning
University College London

Fairness looks different depending on whether one measures the share of income paid in taxes (with some high earners facing relatively low effective rates), or the share of total tax revenue contributed (the top 1% already shoulder a big chunk of the burden). Ronald Reagan liked quoting the famous aphorism that “any government that robs Peter to pay Paul is bound to have the support of Paul.”

Bagehot (February 28th) identified the paranoid style in British politics. The concept was introduced by Richard Hofstadter’s lecture, “The Paranoid Style in American Politics” at Oxford University on November 21st 1963, the day before President Kennedy’s assassination in Dallas.

The profile of China’s tech elite mentioned that Xi Jinping describes them as nongchaoer, those who “ride the tide” of great economic changes (“Among the lucky few”, February 28th). That revives a 1,000-year-old term that seems charmingly outdated today. Although it was a fitting label for those “plunging into the sea” (xia hai) in the 1980s, the new generation would much rather be dubbed China’s Elon Musks than “tide-players”. Innovation today is less about riding the waves and more about redefining the current.