August was a very good month.
The strategy returned +9.95% against +4.18% for the Nasdaq-100, producing +5.77 percentage points of excess return. Since inception in January 2025, the strategy is +79.2% against +40.2%.
There is no reason to talk around that result. We had built positions for this regime, the regime showed up, and they paid.
The bet worked
Roughly 8.8 points of August’s return came from the two macro positions built around the financing side of the AI buildout.
That is not an awkward footnote to the thesis. It is part of the thesis.
The largest capital-spending wave in technology history has to be financed somewhere. My standing view has been that if the buildout continues while fiscal pressure and long-term borrowing costs rise, eventually the financing regime itself becomes part of the trade.
August gave us the first clean version of that event.
On 19 August, the US Treasury announced that it would at least double the size of its liquidity-support buybacks in the 10-to-30-year part of the curve, from a maximum of $2bn to at least $4bn per operation through 4 November. The announcement followed a selloff that had taken the 30-year Treasury yield to its highest level since 2007.
Then, on 28 August, Fed Chair Kevin Warsh used his first Jackson Hole keynote to make clear that inflation was still unfinished business. Markets pushed the probability of a September rate increase toward 60%.
Put those two events next to each other. The central bank is leaning tighter while the fiscal authority is intervening to support liquidity at the long end of the same market financing a huge investment cycle.
That collision is what the positions were built for. We were there before it happened.
The buildout strengthened too
The other half of the thesis did not weaken while the macro positions were paying.
On 26 August, Nvidia reported $96.2bn of quarterly revenue, up 106% year on year. Data-centre revenue reached $89.0bn, up 117%, with a 75% gross margin. Jensen Huang’s description was unusually simple: demand is accelerating.
CoreWeave told the same story from another angle. It closed a $2.6bn financing facility in August, raised its 2026 capital-spending plan, and reported 1.5GW of active power with 4.2GW contracted. Demand is not the obvious constraint. Getting enough power, sites and physical infrastructure remains the harder problem.
So the phase read remains the same: Buildout, not Digestion.
What changed is the financing around it.
What changed in the book
August also moved the book closer to the structure the thesis calls for.
The first rung in Talen Energy filled. More importantly, Talen became the first generation position to clear the repricing test behind the durable side of the thesis. PJM capacity prices moved from $28.92 to $269.92 to $329.17 per MW-day across successive auctions, with Talen’s 2028/29 capacity clearing around $325. That is scarcity turning into realized pricing power rather than merely appearing in a narrative.
Moonshots went from empty in July to 2.2% of the book. Compute & Platforms moved from roughly 16.0% to 14.3%.
Physical scarcity is still not where I ultimately want it. The difference from July is that we are now actually buying it when the price and the evidence line up.
What August changed
July showed that the book could protect relative performance while AI hardware sold off.
August showed something more important: the regime framework can also make money when the event it anticipated actually arrives.
The macro positions were not lucky passengers in the book. They were there because the financing of the buildout is itself part of the economic map. August was the first month that view paid at full scale.
The next major test comes with the late-October cloud and capex prints. If spending starts rolling over, the phase changes. If revenue and spending keep accelerating together, Buildout runs longer.
For now, three things are true at once.
The AI businesses are getting better. The physical constraint is still binding. And the state has started responding to the cost of financing it.
August was the month the book was positioned for that combination.
The framework these positions come from: the research. The founding essay: Who Gets Rich When Intelligence Is Free? The same read, applied to the decision on your desk: advisory.
Concentrated, high-volatility strategy; capital may be lost. Past performance does not guarantee future results. Nothing here is investment advice or an offer.