The AI Supercycle is Now Reorganising Macro Policy. Here is What Happens Next.

I have spent years building a framework called the Universal Code. At its core is a simple idea: all complex systems optimise for intelligence produced per unit of energy. Capital markets are no exception. If that framework is correct, then macro becomes surprisingly predictable from here, and the next five months are the most important window in the cycle. This post is the first public test of the framework as a live operational instrument. Posting the call so I cannot move the goalposts when it resolves. The thesis. The global financial system is now being reorganised around preserving the AI capex supercycle. That is the actual subject. Weaker dollar, energy suppression, foreign duration demand, managed China détente, financial repression, long-end yield management, stablecoin legislation, central bank coordination. All of these are downstream tools serving one objective function: keep capital flowing to the largest substrate transition in recorded history until the build completes. Most people still think AI is a tech story. It is now the primary gravitational field reorganising macro policy itself. Once you see that, the rest of the picture stops looking chaotic. Why no austerity is possible. The US has $9.7 trillion of debt rolling in 2026, similar in 2027. Net interest is now 13.8% of federal outlays and projected at 18.8% by 2036. Every escape route is closed. The Fed cannot tighten because the math of the rollover will not let it. Treasury cannot term out because the demand for duration is not there at acceptable yields. Politics cannot austerity because cutting spending into a debt spiral makes the spiral worse and ends careers. The only path is financial repression. Run nominal GDP hot, let inflation burn the debt-to-GDP ratio down, and engineer the architecture that lets the rollover keep clearing. The historical analogue is 1946 to 1955. The Fed held the curve through 6 to 14% CPI prints while debt-to-GDP fell from 119% to 60%. Nominal GDP did the work the Fed could not. The same operation is running today in modern dress. Nobody can say it out loud. It has to happen anyway. Why the dollar has to go lower. The captive-buyer architecture (the eSLR rule that went live April 1, the Basel III re-proposal, the stablecoin legislation that turns USDC and USDT into structural bills buyers) handles the short end of Treasury issuance. None of this absorbs the long end. Tens and thirties need duration buyers and the only buyers of size are foreign central banks, foreign pension funds, and Asian life insurance companies. They price returns in their home currencies. They only buy at scale when the dollar is falling, because a strong dollar destroys the trade in yen, yuan, and euro terms. So the dollar has to weaken. Deliberately. Engineered. This is the actual mechanism behind everything you are about to see. Weaker dollar brings foreign duration buyers back to the long end, which holds yields down, which lets the rollover continue, which keeps the AI capex cycle funded, which keeps nominal GDP running 5 to 6 percent, which burns the debt stack down without the Fed having to cut into an inflation flare. Why oil has to go lower. Cheap energy is the denominator of I/E. Every dollar of higher oil price subtracts from disposable income, from corporate margins, from manufacturing competitiveness, from the AI build itself. Data centres are energy conversion systems for intelligence production. Iran sitting outside the system at three million barrels per day of suppressed supply, Venezuela at one million barrels of constrained supply, and Strait of Hormuz risk priced into the term structure are the three biggest drags. Resolve all three and Brent goes from $120 toward $70. That is the cheap energy regime the substrate transition requires. The administration has every incentive to make it happen because gasoline at the pump is the most visible household price input there is and the midterms are five months away. Why Taiwan and the tech stack matter. The intelligence numerator of I/E is now being built in fabs that sit ninety miles from the Chinese coast. The compute substrate (Nvidia GPUs, TSMC manufacturing, ASML lithography, Korean memory, Japanese materials) is the most concentrated supply chain in industrial history and the most fragile geopolitically. The grand bargain has to stabilise it without either side conceding on the core technological hierarchy. The shape. US keeps the leading edge inside the alliance perimeter (Taiwan, Korea, Japan, US fabs). China gets continued access to legacy nodes and managed mutual access on critical materials. Neither side fights over Taiwan because the fight collapses both numerators simultaneously. Xi gets the win at home through tariff carve-outs and yuan appreciation that solves China’s deflation problem. Trump gets the win at home through manufacturing reshoring announcements and tariff theatre. The substantive bargain happens quietly. The optics happen loudly. The actual subject is keeping the AI capex cycle running for another three years without the substrate fracturing. The grand bargain has four moving parts. One. Energy resolution. Iran de-escalates within ninety days. Hormuz reopens. Venezuela sanctions architecture eases. Brent toward $70 by August. Two. FX cooperation. Japan signs first because politically aligned. BOJ resumes rate normalisation with Treasury providing FX cover. USDJPY trajectory toward 130 over six to twelve months. Japanese life insurance complex returns to the US long end. Three. China appreciation. Yuan strengthens deliberately. USDCNY trajectory toward 6.80 over the same window. Solves China’s deflation problem, gives Bessent the duration bid he needs, lets Xi claim victory at home through a managed framework on AI access and tariffs. Four. Bond market response. Term premium on the 10Y compresses as foreign duration demand returns. The 10Y nominal can drift modestly higher while term premium falls. That is the architecture working. The operator. Bessent is the only Treasury Secretary in modern history who reads the long end at the muscle-memory level. Career macro PM, ran one of the great currency trades of the 1990s, understands the bond-market-as-discipline-mechanism the way the rest of us understand gravity. He is not running fiscal policy. He is running a bond sales operation with a global book. Japan and China are the duration counterparties he has to bring back. The dollar weakening is the price he pays. He will execute because the alternative is the long end pricing the regime as incoherent and the architecture fracturing from the outside in. The political forcing function. Republicans need to win the November midterms because divided government breaks the policy mix that funds the substrate transition. Bessent has roughly five months. Not five years. Not five quarters. Five months to make the bargain visible before Labor Day, because the campaign period that runs from September through November needs the asset-price expression and the cheap-energy regime already locked in to argue the case to voters. The funder coalitions. The deepest layer underneath all of this and the part nobody else in macro is yet pricing properly. The administration’s continuity is funded by two coalitions that delivered 2024. The crypto industry which spent over $250 million through Fairshake and won an unusually high share of its targeted races. And the tech principal defection which moved Musk, Andreessen, Sacks, Thiel, Armstrong, the Founders Fund and a16z networks, and the broader Silicon Valley funding centre of gravity from left to right. These funders are betting their political spending against the asset-price rally the policy mix is supposed to produce. If the rally fails to express, they defect, the funding architecture breaks, and the operator’s continuity becomes impossible. The bargain has to work in markets, not just in the rollover arithmetic. The fragile link. Worth naming honestly. The framework assumes geopolitical actors will subordinate domestic ideological incentives to system-level stability. The Universal Code says they should because the I/E logic is overwhelming. History says they usually do when the stakes are this large. But history also contains examples of states acting against obvious self-interest when prestige, nationalism, or internal legitimacy pressures dominate. Taiwan is the obvious case. The economic logic for non-conflict is overwhelming. The political logic is fragile. The framework can be right about the system-level incentives and wrong about whether the actors actually execute on them. This is the part of the architecture worth watching most carefully over the next five months. The call. Bessent runs Tokyo, Seoul, Beijing this week as one coordinated operation. Multiple smaller deals that add up to the bargain in substance even if the readouts are muted. Iran resolves on US-Saudi terms with Chinese acquiescence within ninety days. Venezuela supply returns. Brent toward $70 by August. BOJ resumes normalisation with Treasury FX cover. USDJPY trajectory toward 130, USDCNY trajectory toward 6.80. Term premium on the 10Y compresses as Japanese and Chinese duration buyers return. Equities print new highs into the midterms. Crypto and Bitcoin catch the deliberate dollar weakening cleanly because the Universal Code says capital routes to non-sovereign monetary assets when sovereign currencies are being deliberately debased. Wrong if either of these fires. DXY closing above its recent four-month high for three consecutive weekly closes before Labor Day (September 7). The dollar has to weaken for the bargain to work. Three weeks above recent resistance says it is not weakening and the foreign duration bid is not returning. 10Y yield closing above its recent four-month high for three consecutive weekly closes before Labor Day. The bargain has to clear the long end. Three weeks above recent resistance says the bond market is rejecting the architecture and the rollover is fracturing. Two falsifiers. The mechanism (dollar) and the result (long end). Everything else in macro cascades from these two. If both hold below their recent four-month highs through Labor Day, the bargain is working. If either breaks through and stays through, the framework was wrong. The political leg resolves separately on November 3. Republicans losing both chambers tests whether the bargain compounds into 2027 rather than whether the bargain itself worked this summer. The Universal Code says this is what has to happen because the system is now reorganising itself around preserving the AI substrate transition, and every actor in the architecture has incentives aligned with the outcome. The math says this is what has to happen because no other path is mathematically available. The operator running it knows what he is doing. The political clock is running. Holding myself to it. Let’s see if the framework earns its keep.