Nothing Stops This Train

The week ended on September 11, 2026, interesting news items to look at for the past fortnight are:

Nothing Stops This Train: In my previous blog, I provided a technical indicator as to why I boldly forecast that an unwinding of the dollar hegemony will occur in October this year. In this blog, I will try to flag the fundamentals involved.  (Fig. 1)  

1. The Triffin Dilemma - The Triffin dilemma is an economic conflict that occurs when a national currency also serves as the global reserve currency. The dilemma comes in two parts: (a) To supply global liquidity: The world needs a steady, abundant supply of the reserve currency to fund international trade and foreign exchange reserves. This forces the issuer to run persistent trade deficits (buying more from other countries than it sells) and (b) To maintain currency confidence: Flooding the global market with its currency and running high trade deficits weakens the perceived value and stability of the currency, eventually eroding the confidence required for it to remain the world's reserve standard. (Fig. 2).

To the credit of Henry Kissinger, his design of the petrodollar has extended the shelf life of the USD as a global reserve currency by 40 years (1974 to 2014). Witness blow graphically, how the Triffin Dilemma worked out in U.S. current account deficits from 1971. (Fig. 3)

2014 was a pivotal year when China stopped adding her current account surpluses in USD to her national reserves. 

2. Ponzi Accounting - Once China stopped recycling annual trade surpluses into US Treasuries, alarming gaps began to develop in US sovereign funds flow. First, U.S. domestic purchases of treasuries need to step up to compensate for foreigners dragging their feet to recycle their trade surpluses. (Fig. 4).

Then there was the endless wars and COVID which has to be funded. The sad truth is that US citizens as a whole have very little savings. (Fig. 5)

To maintain confidence in the USD, the U.S. Treasury and the Federal Reserve began Ponzi accounting to whitewash the funding gaps. All the accounting tricks that caused the collapse of Enron in 2001 were brought in to make the U.S. National Accounts do not look as sick as they should. Where there were insufficient foreign buyers, hedge funds were brought in to warehouse bonds in offshore financial centres (London, Tokyo, BLICS and Liechtenstein) using leverage, swaps, and derivatives. (Fig. 6).

Where US domestic demand was insufficient, the same hedge funds were brought in to do domestic basis trade. (Fig. 7).

The sums involved in these synthetic demands are humongous and are estimated by Wall Street analysts to be around US$8 trillion. Every night, these positions are funded by the Fed in collaboration with crony central banks using swap lines of up to $5T with the rest via interbank SOFR of up to $3T. The top five hedge funds—Citadel, Millennium, Capula, ExodusPoint and Balyasny—accounted for  $3.5 trillion of these synthetic demands, and their combined equity is only $200 billion. A 5% swing in bond prices would trigger an immediate $7 trillion in liquidation. To put this in context, the so-called risk-free US10YR Notes already had a YTD price drop of 6.26%. Just pause and think about why Bessent is screaming at the top of his lungs. “I am the House”. (Fig. 8).

3. False AI Narrative - The frontier US AI models will go the way of the first movers in railways, dot-coms and optical-fiber telecoms. By that, I mean, despite being a disruptive technology, the first movers will likely go bankrupt, but those who came later will reap the reward.  ChatGPT is running out of free web data to train (Reddit, X and free web sites). They are taking advantage of a change in US law to buy thousands of second books outside the U.S. to scan and train. Proprietary data farms like Thomson Reuters or People's would use open-source platforms to develop their data into income streams. 1% of customers of OpenAI and Anthropic provide 80% of their revenue. These customers are only in the hundreds.

So the reverse pyramid of US AI is like a few hundred customers ( a lot are AI startups with no revenue but funded by Venture Capitalist) spending altogether $23 billion (actual 2025). With that, OpenAI and Anthropic are running losses at least twice that amount. This, in turn, has to support a computing commitment of $1.2 trillion to the hyperscalers (Amazon, Google, Microsoft, Meta, and Oracle), which in turn commit $750 billion to buy NVIDIA chips. NVidia in turn provide circular vendor financing. The hyperscalers are borrowing billions at rates of 6% to 7% in the capital market, and this squeezes Bessent in the bond market. Bessent in turn, squeezes Takaichi.  All the US big tech cash flows have or about to turned negative shortly. Banks that have lent big bucks to Private Credit to on lend to the special purpose vehicles of Hyperscalers to build data centers are getting nervous. In reaction, banks are buying CDS (credit default swaps) against the big techs to protect their lending exposure. So share price of Mag 7 naturally lag the general market for a change. Guess who is writing the CDS and are also foundational lenders to Private Credit which provided financing to data centers. If you have learned your lesson at the sub-prime fiasco in 2008, you would know the answer—the insurance companies. Just pause and think about why Jensen Huang is screaming at the top of his lungs. “AGI has arrived” and congratulates OpenAI’s latest model, Astra. (Fig. 9).

 

The following is an excerpt from a technology analyst of Astra - The model can draft complex 3D models (such as turbofan engines), but it omits crucial engineering parts (bypass ducts, fuel injectors, bearings) because it mimics visual form without understanding physical principles, mechanics, or causality. Despite the marketing buzz around GPT-6 / Astra operating CAD/KeyCAD, models still fail on fundamental physical-world tests (e.g., scoring ~31.7%–32.3% on the CreatePT physics benchmark). 

When one truly understands the human neural network, one should be totally humbled that no machine in our lifetime could come close to the potency of the human mind. Yes, there are narrow aspects of computing that would excel any human being but in the total context and comparing apple to apple only a fool or deceiver will make a claim of AGI or Super AGI.

4. Depletion of Strategic Petroleum Reserves and Munitions — If one comes right down to basics, one can relate the physical economy or warfare is an exercise in energy transformation. Military analysts estimate that for every barrel of oil used at the front line of battle, 5 barrels of oil directly and indirectly another 35 are needed for logistical support and munition manufacturing. (Fig. 10).

For every calorie of food on our plate, on average, man needs to put in 15 calories of energy. (Fig. 11).

How sad the last superpower nation is run by a bunch of amateurs, and the world is coming fast to a day of reckoning. There will be grave consequences for allowing fools to run amok the global supply chain of energy and fertilizers. The U.S. Strategic Petroleum Reserve will run dry in October 2026, and the price of oil will spike. Equally, there will also be a fertilizer shortage and by extension food.

Be warned and be prepared, my friends.

Isaiah 26:20  Come, my people, enter your chambers, And shut your doors behind you; Hide yourself, as it were, for a little moment, Until the indignation is past. 21  For behold, the LORD comes out of His place To punish the inhabitants of the earth for their iniquity; The earth will also disclose her blood, And will no more cover her slain.