Being myself from beyond the wall, I am intrigued by Andy Burnham’s plan to establish a “No 10 North”. Companies, of course, are no strangers to relocating power centres for strategic reasons. Elon Musk’s Tesla and SpaceX moved from California to Texas; Shell and Unilever upped their roots in the Netherlands in favour of the UK. Lex this week notes other ways in which Burnham’s devolution plan fits the corporate zeitgeist. Tell us if his plans are owt or nowt at lexfeedback@ft.com.

Space is definitely “in”. There’s the SpaceX IPO, of course, but also the voyage of the Artemis II, a rare moment of humans doing something nice in 2026. The Trump administration has proposed boosting its “Space Force” budget by almost 80 per cent. And, um, the top-grossing movie worldwide this year is “The Super Mario Galaxy Movie”.

This week, there was also a space-related merger, undoubtedly one of many to come. Rocket Lab, a US company that puts objects into space, bought Iridium, a satellite telecoms operator, for $8bn. Rocket Lab, run by a New Zealander who once ate his hat to make a point, wants to be an “end to end” company that makes space-bound objects, puts them in orbit and then uses them to sell services.

In the pre-space age, this would just be called “vertical integration”, but, of course, in the tech world there is no point using an old term when a new one will do. Read the Lex note here.

Iridium’s satellite network might be the future, but it’s also the past — and anyone concerned about investment bubbles could do worse than revisit what happened to this company in a past life.

Once part of Motorola, Iridium came up with what in 1998 seemed like an incredibly good idea: a satellite network that would enable mobile phone calls anywhere in the world. Never again would a Motorola engineer’s spouse be unable to make a call from her Bahamas holiday (this is literally the company’s origin, per this excellent account).

Money was poured in to fund a $5bn build-out, and more than $4bn of that was funded by debt. The stock soared. The catch was that the company’s brick-like mobile phones were clunky, expensive and didn’t work indoors or in moving vehicles.

Meanwhile, regular telecoms companies built out cellular networks, thus mitigating the pernicious Bahamas problem. Having hoped for 500,000 subscribers by 1999, but actually ending up with nowhere close, Iridium went bust and was later bought by a former airline executive for $25mn.

Spending hand over fist to build products that users may not want to pay for is a habit that never dies. Right now, it’s almost certainly happening in the world of AI, which is why Iridium’s miserable first act — or at least, the executive biases and grand claims that made it so — is worth studying.

Some things are different: Iridium borrowed heavily, and it is debt that really causes businesses to blow up when things don’t go to plan. Despite a few big bond issues, today’s high-spending Silicon Valley giants are, so far, more cautious with their balance sheets.

But some things are undoubtedly the same. In writing about Iridium, Sydney Finkelstein and Shade Sanford talk about the “escalating commitment” problem: executives “making decisions based on the size of previous investments rather than on the size of the expected return”. It’s hard to imagine the bosses of OpenAI, Meta Platforms and Anthropic aren’t at risk of that condition.

But look: Iridium still exists and is useful in ways its architects couldn’t have guessed. The company’s network really does cover the world. It enables — and recently acquired — Aireon, the company that made it possible to track planes’ locations in real time when they’re over the ocean. You won’t be streaming Netflix from its satellites any time soon, but you might be glad of it if you get lost while hiking in the Andes.

Lex isn’t the first to observe that AI — and notably the $9tn or more being spent on data centres — might take a similar path. Railways, shale wells, dotcom infrastructure and the 1920s aviation boom all torched fortunes but created things both useful and highly valuable today. Iridium, after the Rocket Lab deal, is worth a handsome $8bn. But that’s not much use to investors who ascribed it that very same value back in 1998.

• Britain’s defence budget is getting a £15bn boost by 2030, one-third of it on drones. Louise Lucas doubts, given the shallow pool of providers, that the UK government will get a good deal.

• Dangerous heatwaves in Europe have reignited the debate over making air conditioning as commonplace as it is in the US. Camilla Palladino explains why doing so is more affordable than many people think.

• Comcast is spinning off NBCUniversal, which the cable giant bought from General Electric in 2011. Sujeet Indap wonders why it took Comcast so long to learn what GE already did: sprawl is rarely great for investors.

• In the world of earth-based telecoms, BT’s joint venture with US peer Verizon is a sign that its long turnaround is bearing fruit. Gaia Freydefont sketched out how the UK company’s cash flows could soon almost double.

1. Here’s what we now know about Donald Trump’s recent riches (FT)

2. The UN is sounding the alarm over AI inequality (UN)

3. Five-hour airport queues? It must be summer in Europe (FT)

4. A market oddity called the “overnight drift” is fizzling out (NY Fed)

5. More legal gambling means more food insufficiency, a study finds (NBER)

Have a great week — and happy Fourth of July to those who celebrate.

John Foley
Head of Lex