The archive
NVIDIA has found a way to turn knowledge of its own roadmap into a lower cost of capital for its customers. The result could move AI infrastructure from venture equity into private credit, insurance balance sheets and securitized debt.
The buyer did not need a better AI forecast. Citadel was paid to provide liquidity when Situational Awareness could no longer finance the path to its own long-term thesis.
A leveraged $16 billion portfolio changed hands at a reported discount above 10%. The AI forecast was not the decisive edge.
Four hyperscalers invested almost four dollars in first-quarter infrastructure for each dollar of depreciation. The spending can still earn an excellent return, but the next margin test is already visible.
Where the next AI infrastructure dollar goes, which bottlenecks can retain pricing power, how the boom is being financed, and what could break first.
The first half of 2026 was not a rally. It was one macro shock and three rotations underneath it: down the AI stack into memory, out of the mega-caps into small caps and Asia, and into the war trade and back out. Plus a Fed that changed regimes while almost nobody was looking. This is the full map.
The thesis behind Yield Theory, and why understanding capital flows beats chasing headlines.
this research stays open. Members get the monthly flagship, the weekly checkpoint, and the evidence that would break the call. $15/month or $150/year.
$15/mo or $150/yr · cancel future renewals anytime · sources included