There are 3 types of people in the AI era. Those who are 16. Those who are 35. Those who are 65. Each one faces a completely different set of challenges — and opportunities. Most people are getting the wrong advice because they’re reading the wrong playbook. Here’s what I mean:
The divergence here is uncomfortable.
332,000 tech jobs gone since ChatGPT. 3.16 million healthcare and education jobs added.
But these aren't equivalent substitutions. Healthcare and education are lower productivity, largely government dependent, and can't absorb displaced knowledge workers at equivalent salaries.
A senior engineer doesn't become a care worker at the same income.
The headline numbers look balanced. The lived experience for the 35 year old knowledge worker in the middle is not.
I wrote a book about exactly this divergence. It's called 16·35·65.
The demand/consumption distinction is interesting.
Demand falls on the spreadsheet. Consumption stays stickier, strategic reserves get drawn down, essential industrial use continues regardless of price signals.
The 6.2 mb/d demand cut looks like a buffer. But if it's masking sustained consumption rather than genuine reduction, the supply gap doesn't disappear, it just gets deferred.
Deferred gaps tend to resolve suddenly, not gradually.
The status risk is more dangerous than the default risk precisely because there's no clean alternative.
A country that can't pay can restructure. A reserve asset that loses confidence has nowhere to go, and takes the system with it.
The $8.3 trillion rolling over this year is really one to watch.
The lesson is not that governments always fabricate intelligence. It is that the process by which uncertain intelligence becomes public certainty is vulnerable to institutional pressure in ways that are predictable and repeatable. The WMD failure was not unique. It was a pattern.
The Butler Review in the UK and the Senate Intelligence Committee report in the US both found that intelligence had been stretched, caveats removed, and uncertainties flattened in the journey from analyst to policymaker to public.
The 1970s found one answer: more oil, different places.
The 2020s are producing five answers at once, and each one creates new dependencies to replace the old.
US: drill and export LNG. Europe: renewables plus nuclear revival. China: all of the above simultaneously. India:, Russian discounts plus solar. Global South: whatever's cheapest.
Same crisis. Completely different bets.
Interesting idea, but a tough call to make.
Reimbursing ship owners from frozen assets neutralises the revenue but adds a precedent.
The moment a reimbursement mechanism exists, Iran has established that it can charge toll on an international strait. Every future adversary knows that.
A face saving solution works financially. But will it be worth the real strategic concession?
Exactly! Slimmer batteries aren't just about efficiency, they're about reducing exposure to mineral supply chains China doesn't fully control.
Cobalt from the DRC. Lithium increasingly contested. The push for lighter EVs is partly an engineering story, but it's also Beijing hedging its own strategic dependencies.
The irony is the West worries about depending on China for clean tech. China worries about depending on unstable states for the minerals that make clean tech possible.
Dependency all the way down.
'The most dangerous man to any government is the man who is able to think things out for himself.' H.L. Mencken wrote that in 1919. It has not aged badly. The mechanisms for managing what populations know and
believe have become significantly more sophisticated since 1919. The underlying dynamic has not changed at all.
Hard to argue with the miracle framing, the cost collapse has been extraordinary.
But Pakistan shows what happens when the miracle outpaces the system around it. 55GW imported in three years, no plan, no grid reform. The panels work. The institutions didn't keep up.
@janrosenow wrote the best account of this I've read, the utility death spiral is already playing out in real time there.
https://t.co/JOiJF0sVIW
The question isn't whether solar is transformative. It clearly is. The question is who carries the cost when the transition moves faster than policy can follow.
@AssaadRazzouk genuinely curious your take on the death spiral dynamic, solvable design problem or inherent tension in distributed generation at scale?
Fascinating. The utility death spiral is the part most people miss.
When wealthier households exit the grid, fixed costs land on those who can't afford to leave. Their bills rise. More leave. Repeat.
Pakistan didn't just build the world's fastest distributed solar rollout. It accidentally ran the world's most brutal real world test of what happens when deployment outruns institutions.
The panels are the easy part. The system around them is where it breaks.
7% of layoffs cited AI in January.
40% in May.
That's a cliff edge.
The 35-year-old knowledge worker who spent a decade building expertise in a specific domain is now the most exposed person in the labour market. Not the 16-year-old who grew up with these tools. Not the 65-year-old who's already out.
The middle is being hollowed out fastest.
I wrote a book about exactly this moment. It's called 16·35·65.
Good numbers forPhilippines. But a really interesting story in the graphic.
The Netherlands leads by a huge margin, it's a re export hub for northwest Europe.
Europe is decarbonising on Chinese hardware while simultaneously legislating strategic autonomy from China.
Energy transition but is it also dependency substitution. Trading Russian gas for Chinese solar panels but with better PR?
This isn't new, it's a pattern.
The US and Israel have tolerated mutual surveillance for decades. What's changed is the stakes. Israel eavesdropping on Iran nuclear negotiations directly threatens its ability to shape the outcome.
When an ally's survival calculus diverges from your own, the intelligence relationship follows. Perhaps not betrayal, more it's alignment breaking down.
The surprise isn't that it's happening. It's that it's being made public.
The political timing matters as much as the number.
$8.3 trillion rolling over into midterms. Foreign central banks stepping back. Private investors now the marginal buyer.
That's not just refinancing risk, it's leverage. Whoever holds that paper holds the pressure point.
The Fed's 'independence' has never faced a structural test quite like this.
One company on one island manufactures 70% of the world's advanced semiconductors.
Every Nvidia AI chip. Every Apple processor. Every AMD data centre chip. All of them depend on TSMC. All of them ship through the Taiwan Strait or South China Sea.
Japan and the Philippines just agreed to delimit their EEZs in waters that overlap Taiwan's, without Taipei at the table. Beijing objected. Taipei's response was to direct its anger at Beijing rather than defend its own maritime rights.
The geopolitical manoeuvring around Taiwan is routinely framed as a sovereignty and democracy argument. It is also, inescapably, a semiconductor supply chain argument.
Whoever controls those waters controls the arteries of the global AI economy. That's not a side issue. It is The issue.