There’s long been a disconnect between concerns about the massive impact of AI data centres on electricity demand and claims by Sam Altman and others that the impact is really modest. Ed Zitron recently posted a summary of OpenAI’s 2025 accounts which helps to clarify things a bit.

In short, if you look at actual electricity demand needed for current AI use, it’s small. And that doesn’t change if demand grows at high but plausible rates. On the other hand, if you look at what is needed to justify the current valuations of AI and its competitors, the implied growth is staggering.

Read More »

The SpaceX IPO, valuing a motley collection of dubious business at over a trillion dollars, marks the abandonment of the Efficient (financial) Markets Hypothesis, one of the zombie ideas I criticised in the wake of the Global Financial Crisis. Not only do financial markets fail in the task of valuing assets accurately, but the institutional structures that are supposed to make them work have given up trying.

This was prefigured by the rise of Bitcoin and other forms of crypto. Revealingly, no one any longer uses the term “cryptocurrency” – these assets are never used as currency in ordinary transactions, and even their illicit uses seem to have faded. Rather, Bitcoin is valuable solely because it is valued. As I pointed out back in 2018, (free from paywall here) once this logic is accepted, it can be applied to financial assets more generally, and particularly to stock markets.

For quite a while, financial institutions like Goldman Sachs held out against crypto. But the rewards became too great to ignore and the crypto industry too politically powerful. Crypto moved from the netherworld it once inhabited to the respectability of exchange traded funds.

The extension to stock markets has been happening for a while, with meme stocks like GameStop, and then with massive valuations of AI stocks. But meme stocks are a sideshow, and the future of AI is unclear enough that it’s possible to make a case.

It’s only with SpaceX that we can see the complete abandonment of any pretence at rationality. In the case of SpaceX, I was struck by Dave Karpf’s observation that Musk’s wealth in 2020 was “only” $24 billion. Everything of value in his career (Tesla cars, batteries, Starlink) had been achieved by then, and everything he has touched since then has been a disaster (Xitter, Cybertruck, robotaxis, Starship). Yet his wealth has multiplied 50 times over).

In support of the IPO, Goldman Sachs has put its name to the claim that the company will grow 100 times over by 2030. This is patently absurd. Nothing in Musk’s ragbag of assets has this kind of potential.

Starlink, the pre-2020 bit of the business, has been successful enough, but satellite-based Internet is never going to be huge. And there is plenty of downside risk, as Europe and China try to develop alternatives. Given that Starlink’s satellites have a life of only five years, the business could well be in sharp decline by 2030. Starship, by contrast, seems unlikely to succeed in getting humans to the Moon, and won’t make much money even if it does. Talk of Mars, data centres etc is a joke.

The AI part of the business is barely even a joke. Not only is Grok an also-ran in the LLM stakes, but Musks’ promotion of racism, terrorist riots and sex crimes leaves him, and the company open to huge liability, at least in Europe, where Trump can’t protect him.

The comprehensive corruption of the financial system confirms me in the view that the USA is one big grift. Not just the financial system, but politically and militarily as well. But that’s a topic for another day.

One my betes noires has been in the news lately. Jonathan Haidt has been annoying me since at least 2012, when I was critical of his bothsidesism on the culture wars.

At the time, he was a concern troll, posing as a liberal worried about other liberals who were, he claimed, misunderstanding Republicans. Whereas liberals thought of Republicans as bigots and misogynists, concerned with preserving their own position in racial and gender hierarchies, Haidt explained that they actually had their own set of values, based on order, purity, honour and loyalty. The wheels started falling off that one when Donald Trump, the antithesis of all of these things, came along and received the unqualified support of the supposed believers in purity.

Haidt responded by reinventing himself a free speech advocate, concerned about cancel culture and the coddling of young minds. He hung out on the “Intellectual Dark Web” with Bari Weiss and Stephen Pinker. Now that Weiss is busy suppressing reporting of Trump’s crimes, the IDW appears to have shut up shop.


Haidt’s next reinvention was an almost complete backflip. Despite having no relevant research background (as discussed his previous focus was on adult voters, followed by a shift to scolding college students) he suddenly became an authority on the effects of smartphone use on teenagers. His concerns about freedom of speech suddenly went out the window, replaced by a fear that the speech teens encountered on their phones was making them depressed as miserable.

Haidt’s work writings on this topic inspired the Australian government to pass legislation aimed at banning access to social media platforms for people under 16. It’s been mostly ineffective, but for the minority of kids who have left social media, a notable impact has been reduced access to news.

Possibly because recognition of this failure is spreading, Haidt has gone back to the “coddling” theme in a commencement address at NYU, for which he was roundly booed. Haidt appears not to have noticed that, far from protecting students from views that might upset them, NYU is busy suppressing speech by students that offends the administrators and of course the Trump Administration.

The kind of concerned punditry of which Haidt is an exponent never goes out of style, even if the topics of concern change from time to time. Given his ability to leap from one topic to the next with a fine disregard for consistency, I expect he will be around to annoy me for a long time to come.