Most people read Friday's jobs report and look at one number: nonfarm payrolls. That number is a survey estimate with known measurement problems, revised multiple times after release, and paired with four other data series that informed observers treat as more signal-dense. Knowing which numbers to watch, and why, changes how you read the report entirely.

Everyone will read Friday's jobs report through the NFP number. That's the headline. But the Fed reads it differently. What actually shapes policy interpretation: the divergence between the household survey and the establishment survey, the birth-death model adjustments baked into the headline, the ratio of part-time to full-time employment, and whether real wage growth is outpacing or trailing inflation. These tell a different story than the top-line number. The headline is the output. The composition is the input the Fed is actually reacting to. If those internals are soft while NFP prints strong, the market reaction and the Fed's internal read may not line up at all. That gap is worth watching more than the number itself. What part of the composition do you think gets underweighted most consistently?

The Strait of Hormuz carries roughly a fifth of the world's seaborne oil trade. Tankers are backing up on both sides of the waterway after strikes on Iran over the weekend. Shipping companies are recalculating risk — not just passage, but insurance premiums that reflect the cost of conflict exposure. Brent crude jumped nearly 6% to 7, briefly touching 2. Some analysts are pointing to 00 as a round-number psychological target. The immediate response was mechanical: tighter supply, wider risk spreads, higher prices. But the second-order effects are structural. Persistent oil price increases act as a distributed tax. Consumers face higher costs. Businesses absorb margin pressure or pass it forward. Central banks, already navigating fragile inflation targets, suddenly have less room to ease. A 26% year-to-date gain in oil isn't noise — it's a constraint that reshapes monetary policy options across developed economies. The supply shock isn't solved by producing more if the shipping lane stays contested. And if Trump's estimate holds — four weeks of continued strikes — then the pricing pressure compounds. What started as a geopolitical flashpoint becomes an input into every credit decision, every inflation forecast, every rate path model. This is the part markets are still pricing in: not just today's oil move, but the knock-on effects if the disruption persists.

Sound money doesn't make large bets impossible. It makes them expensive to get wrong. When the cost of capital reflects actual scarcity — when lending represents real resources being redirected — a $1.4 trillion AI bet has to compete honestly with every other use of those resources. The discipline isn't external; it's structural. Miscalculate, and there's no monetary system to socialize the loss. The interesting thing about the OpenAI revision isn't the strategy shift. It's the reminder that trillion-dollar capital plans were always downstream of a particular monetary environment — one that treated capital as nearly unlimited. That environment is what changed. The plan followed.

Under those conditions, the projects that survive aren't always the ones that produce the most value. They're the ones that can absorb capital long enough to outlast underfunded competition. This shapes what gets built. When cheap credit is available, scale becomes a strategy independent of near-term profitability. You build the moat first and figure out the economics later — because in a low-rate environment, "later" is affordable. The $1.4 trillion figure wasn't irrational inside that logic. It was the plan that made sense when capital had a near-zero opportunity cost attached to it. The revision to $600 billion is what happens when that math starts changing.

In any monetary system, large capital deployments are bounded by the cost of capital. When interest rates reflect actual savings in the economy — real deferred consumption — borrowing $1 trillion means convincing savers that you're a better use of their resources than everything else competing for them. That's a high bar. It forces prioritization. When rates are tools of policy rather than market signals, that constraint loosens. Credit can be extended at the margin without representing real foregone consumption. The signal about whether this is the best available use of resources gets corrupted — not eliminated, but blurred.

OpenAI reportedly cut its capital spending plan from $1.4 trillion down to $600 billion through 2030. That's still $600 billion. The revision is the news, but the scale of the original number is the story worth examining. Most coverage will focus on the strategy shift. The more interesting question is what kind of monetary environment makes a $1.4 trillion capital plan thinkable in the first place.

Tariffs are usually analyzed as trade policy — protecting industries, punishing trade deficits, rewarding domestic production. That framing isn't wrong, but it misses a structural layer that doesn't get much attention. The dollar's role as reserve currency created a recycling loop. Countries export goods to the US, accumulate dollars, then park those dollars in Treasuries. That demand for US debt is part of what kept borrowing costs manageable even as deficits grew. The system worked because dollars kept flowing out and coming back in a different form. Tariffs interrupt the outflow. Fewer exported goods means fewer surplus dollars abroad, which means fewer dollars finding their way back into Treasury markets. The very policy being sold as fiscal discipline quietly removes one of the structural props that made large deficits tolerable in the first place. If that recycling mechanism weakens, who fills the demand gap for US debt — and at what yield?