CMBS office delinquency hit 12.34% in January 2026 — past the GFC peak of ~9.2%. February came in at 11.4%. That isn't recovery — it's extensions masking the real number. ~$25B in CMBS loans sitting past maturity without resolution, the highest since 2008. In 2010–12, delinquency peaked then fell because the underlying asset recovered. Office isn't recovering. Hybrid work is structural, not cyclical. There's no comparable recovery trade on the other side of this curve.

Trump relocating to Bedminster, which is his golf club. Two readings: (a) he's staying close to Washington because a strike decision window is open over the long weekend b) he's staying close because an MOU signing/announcement window is open. The same behaviour fits both the kinetic and the diplomatic resolution.

Strategic ambiguity just collapsed. Not via policy change — via the president stating publicly, in Beijing, that he wouldn't fight a war 9,500 miles away and naming Taiwan independence as the scenario he's disclaiming. Rubio holds the formal line, but the substance is gone. Taipei is more alone tonight than it was yesterday.

Pre-summit "we hope to convince them to play a more active role" → Post-summit "we're not asking for China's help."

Read on the summit so far: Beijing got: warm summit theatre with US CEOs as props, the Rubio handshake (a real diplomatic win given the sanctions history), Taiwan arms sales quietly deprioritised, and a "board of investment" framework opening Chinese capital flows into the US. Washington got: "common ground on Iran" — undefined — and Bessent's "do what they can" on Hormuz, which is not a commitment, it's a direction. Rubio's "we're not asking for China's help on Hormuz" is the tell — that's a face saving denial of the obvious. They absolutely came hoping China applies pressure on Iran and Beijing gave them the language without the action.

278 out of ~4,600 US commercial banks carry CRE loans above 300% of risk-based capital. Each red dot is one institution. Together they hold $1.8T in CRE debt — on the same balance sheets carrying rate-sensitive Treasury portfolios. When the maturity wall hits them, the Treasury holdings don't stay either.

Today's summit isn't "America's top CEOs" it's specifically: the CEOs whose companies need something from China: Huang needs export-license clarity (or carve-outs) to ship H20/H100 successors. Cook needs supply-chain stability in Foxconn-and-beyond Chinese manufacturing. Boeing needs the China orders book. The financiers need access to Chinese capital markets that have been progressively restricted since 2021. Musk needs Tesla Shanghai gigafactory to keep producing while US-China decoupling rhetoric escalates.

In 1980 the US had 33 primary aluminium smelters. It has 4 today. Over the same period China went from ~0% of global production to 59%. The crossover happened around 2003–05 and the gap has widened every year since. Aluminium is in aircraft, vehicles, defence systems, construction, and packaging. The US lost 88% of its smelting capacity over 45 years — not to a crisis, not to a war, but to incremental decisions that each made sense individually.

Been quiet for a couple of weeks. Still here. In 2023 China held High Monopoly Risk status in 14 critical technology categories. In 2025 that number is 41 — every single one across defence, AI, energy, materials, and biotech. 10 new categories were added this year. China leads 8 of the 10. The US leads 5 of 74 total tracked fields. themacroledger.com

The $369B cash position isn't Treasury demand, but equity rejection. Buffett confirmed it today, when asked if prices are just too high: "Yeah." x.com/TML_Macro/status/20496…

The $957B CRE maturity wall that hit in 2025 didn't resolve, it transmuted. Some refinanced — into rates 300–400bp higher, with coverage ratios that can't support the new debt service, some extended — which is what $25B in CMBS loans sitting past maturity without resolution actually is, and some became the 12.34% office delinquency rate that just passed the GFC peak. The 2025 bar is grey on this chart because it's elapsed, what it didn't do is clear. $1.09T still incoming — $539B this year, $550B in 2027 — into a system still digesting year one.

What the the February 2026 TIC headline of $147B "demand", actually is: Four categories of buyer, none of them buying because they're confident in US sovereign credit. When any one of them decides the alternatives are attractive again, nothing holds the others in place.