Four things on my mind today. They seem unrelated until you look at them together, and then they start to feel like facets of the same underlying question: who controls the game, and how did they get there?
Let's start with AI, because it always starts with AI these days.
The observation that landed cleanest for me this week is this: the race at the top of the AI stack is a high-stakes poker game, and compute is the buy-in. You don't get a seat at the table without it. Not a meaningful seat. The capital requirements to train frontier models have become so large that they function as a natural filter — not just for who wins, but for who is even allowed to play. OpenAI, Anthropic, Google, Meta, xAI — they are not competing on ideas at this point. They are all smart. The ideas overlap more than anyone publicly admits. What they are competing on is the size of their stack.
What troubles me about this is the divergence it creates downstream. Every major player is building their own infrastructure — their own data centres, their own chips, their own power deals, their own cooling solutions. The duplication is staggering. The same physical problems being solved in parallel by teams that will never share a single kilowatt-hour of the result. And for what? Competitive advantage. Proprietary moats. The race logic demands it even when everyone involved can see the waste.
That is the divergence we are being asked to endure. And the frustrating part is that the alternative is not naive. Shared infrastructure doesn't mean shared models, shared data, shared IP. It means the power plants and the cooling and the undersea cables don't need to be duplicated six times over. The failure modes at that layer — an outage, a sabotaged facility, a regulatory seizure — become shared risk too. Which historically is exactly the kind of thing that motivates cooperation between competitors. Airlines share safety data. Banks share fraud signals. The infrastructure layer often becomes a commons because the downside of not sharing it is too large for everyone. We may get there eventually. The question is how much gets burned first.
Now: Israel.
Ten million people. Surrounded by adversaries for its entire existence. Operating under a level of existential pressure that most nations never experience and can barely imagine. And yet — punching so far above its weight in technology, intelligence, medicine, agriculture, and now AI that the gap between its size and its output has become one of the more quietly remarkable facts of the modern world.
The blindspot observation is sharp: it's right in front of our eyes. Israel is not a hidden story. It is covered constantly, analysed obsessively, argued about endlessly. And yet the throughline — the specific question of how a nation of ten million sustains this level of output under this level of pressure — gets lost in the noise of the political argument around it. The mechanism disappears into the controversy. Which may, as someone put it to me, be Israel's actual secret. The debate about Israel consumes so much energy that the lesson inside it goes unexamined.
I don't have a clean answer to what the mechanism is. Probably several things compounding: mandatory military service creating a talent pipeline that blends technical and strategic thinking from age eighteen; the necessity of genuine innovation because buying security off the shelf was never an option; a culture that treats failure as data rather than disgrace. But I keep returning to the necessity factor. Innovation under existential constraint is different in character from innovation under competitive pressure. The former tends to be more resourceful, more integrated, less tolerant of elegant-but-undeployable ideas. Survival has a way of clarifying requirements.
Then there is the Strait of Hormuz, and Iran's surprisingly patient hand.
The strait is one of those geographic facts that sounds almost too neat to be real: a narrow channel through which roughly a fifth of the world's oil passes, controlled — or at least threatenable — by a nation that has been under sanctions and international pressure for decades. Iran has never needed to close Hormuz to benefit from its existence. The threat alone is the asset. And the way that threat has been managed — drones to regional neighbours, calibrated pressure on shipping lanes, keeping the media positioned as witness rather than adversary — is a more sophisticated piece of statecraft than it usually gets credit for.
The deeper point is about stability. The region is closer to a durable equilibrium than the daily news cycle suggests. Not close in an absolute sense — seventy-five years is a genuine estimate for the long tail of tensions to fully resolve, if they do. But the trajectory is different from what it was twenty years ago. The Abraham Accords changed the geometry. The drone economy changed the cost calculation. Even the major powers are more cautious about large-scale commitments than they were in the 2000s. Iran, for all its bluster, seems to understand that the current arrangement — threatening but not triggering, pressuring but not provoking — is the optimal position. The noise around Trump and Hormuz is real. The underlying stability is also real. Both things can be true.
And finally: Formula 1.
The Canadian Grand Prix was this past weekend, and next Sunday brings Austria — and reportedly a new Ferrari power unit. I find myself genuinely curious about this in a way that I didn't expect. F1 sits at an unusual intersection: it is a sport, but it is also an engineering competition, a geopolitical soft-power exercise, a marketing platform, and — perhaps most importantly — one of the last remaining cultural spaces that reliably bridges generations.
That last point matters more than it sounds. Most cultural phenomena now are segmented by age. The algorithm optimises for what each cohort wants to see and rarely surfaces content across the divide. F1 resists this. It has always been a sport that fathers and children watch together, argue about together, care about together. The technical complexity gives the older viewer something to explain. The spectacle gives the younger viewer something to feel. And in an era when shared cultural reference points are rarer than they used to be, that is not a small thing.
Ferrari with a new engine at Austria. That thread deserves more pulling.
Four things. One question underneath all of them: who controls the game, and how did they get there?
The answer is never simple. It is always worth asking.