Prospectus / Core Sample 001
Since 1958, atmospheric carbon has returned a positive result every single year on record. We remain constructive on scarcity, long on consequence, and hedged against everything except hope.
Never a down year in 67 years of records. Past performance, in this instance, does guarantee future results.
Paris target: breached. Adaptation TAM: unlocked. Constructive on cooling, desalination, and haven real estate.
89 to 85 in a single year. Momentum favourable. Downside is, definitionally, limited.
The index we are all trying to beat. Note the correlation with every holding above. That is not a bug. That is the thesis.
Three books, rendered in relief. Each bar a decade; each decade, in our view, a position.
The flywheel is the whole business: capital builds compute, compute burns power, power prints emissions — and the returns bring the next round of capital. We are invested at every turn of the wheel.
Capital: Epoch AI / SEC filings; Stanford AI Index 2025–2026. Power: IEA (2025). Emissions: Google & Microsoft Environmental Sustainability Reports (2024–2026). Figures are each company's own reported values; the loop between them is the argument.
A brief reconciliation. Clean energy finally out-raises fossil two to one — and it has scarcely touched the ledger. The capital that compounds still finds the other column; we hold both sides of the book.
Bank finance to fossil fuels
$8.7 trillion committed since the Paris Agreement. JPMorgan Chase remains the largest financier, at $58 billion in 2025 — a decade after the sector's net-zero pledges.
Fossil-fuel subsidies · 2024
On top of $725 billion in explicit subsidies governments pay directly. Together, fossil fuels are under-priced by roughly $7 trillion a year.
We take no side in the ledger. We take both.
Energy investment: IEA World Energy Investment 2025. Venture capital: Sightline Climate (climate tech, 2025) & PitchBook (OpenAI, xAI, Anthropic combined, 2025). Bank finance: Banking on Climate Chaos 2026 (RAN et al.). Subsidies: IMF 2025 update (2024). Anthropic — one of the three AI labs shown — makes the assistant used to build this page.
A note on distribution. The returns on everything above accrue narrowly — and that concentration is not a side-effect of the system. It is the system's finest product.
US CEO-to-worker pay
Since 1979, US productivity rose about 90% while typical worker pay rose 33%. Had pay kept pace, the median worker would earn roughly $16 more an hour.
Who causes it vs. who pays
They cause roughly a tenth of emissions. The richest 10% — responsible for about half — face around 3% of the losses.
We are, naturally, positioned at the top of every chart on this page.
Income & wealth shares: World Inequality Report 2026 (World Inequality Lab — Chancel, Piketty et al.). Pay: Economic Policy Institute, 2025. Climate burden: World Inequality Lab & Oxfam.
Our benchmark and the atmosphere have rarely tracked one another so closely. As the parts-per-million climb, so does the portfolio. We did not cause the correlation. We simply chose which of the two lines to own.
Q3 Positioning Note · Strictly Confidential
We continue to see durable, multi-decade tailwinds across the book. Scarcity remains structurally bid. Consequence compounds. The capital chasing compute has never been deeper, and the firms emitting most are, conveniently, the firms being funded most. We are long carbon, long heat, long midnight, and long the flywheel. We remain, as ever, constructive.
— The Investment Committee
Every position in this book assumes the trend lines keep climbing. They don't have to. Threshold is the research behind the joke: an interactive atlas of the points where the climate system changes state, and the underfunded technologies that are the real high-leverage trade — before they're priced in.
Open Threshold — the real position