“Mother All Crises” | Luke Gromen on America’s Choice Between AI Dominance and Real Value of Treasury Market

Episode Summary

Executive Summary: Luke Gromen argues the U.S. faces a fundamental tradeoff: reshore industry/compete with China in AI, or preserve the real value of the dollar and Treasury market. He says real-world constraints—grid, labor, rare earths, capital—will force de facto yield curve control, financial repression, and likely a weaker dollar, while making gold the key reserve asset and Bitcoin vulnerable in the near term.

Main Topics: Reshoring vs. Treasury-market stability (Priority: 5/5): The core thesis is that the U.S. cannot simultaneously rebuild manufacturing, compete in AI, and preserve the real value of long-duration Treasuries. Reshoring requires higher real rates, more capital spending, and likely bond-market suppression. AI, industrial policy, and real-resource constraints (Priority: 5/5): AI is treated less as a pure software story and more as a physical-capital bottleneck story: data centers need grid power, rare earths, copper, labor, and engineering capacity. These constraints drive inflation in real inputs. Yield curve control and financial repression (Priority: 5/5): To keep borrowing costs manageable amid rising deficits and capital needs, Luke expects some version of yield curve control, SLR relief, repo support, or QE-by-another-name. That would cap yields while debasing long bonds in real terms. Gold as the neutral reserve asset (Priority: 5/5): Gold is presented as the only large, politically neutral reserve asset that can absorb global surplus capital if Treasuries stop serving that role. Luke sees central bank buying and reserve reallocation as a major tailwind. Bitcoin weakness and ‘last functioning smoke alarm’ (Priority: 4/5): Luke turns short-term bearish on Bitcoin because it is still trading like high-beta tech, failing to outperform gold, and may be signaling broader liquidity stress, quantum risk, and forced selling. Grid shortages and the coming bottleneck (Priority: 4/5): The transcript stresses that the U.S. grid has stagnated while China expanded aggressively. This makes infrastructure buildout slow, capital-intensive, and politically fraught, even if demand is rising fast. Inflation vs. deflation from AI (Priority: 4/5): AI is expected to be inflationary on the production side (capex, labor, power) but deflationary on the consumer side (job loss, lower credit demand). Both pathways ultimately threaten debt markets and provoke policy response.

Key Arguments: The U.S. cannot reshore production and simultaneously preserve the real value of the Treasury market; doing both would violate basic balance-sheet arithmetic. Competing with China in AI requires massive real-economy spending on grid, power, chips, labor, and materials—things the Fed cannot print. If capital shifts from Treasuries into factories and infrastructure, yields must rise unless the Fed/MoF caps them through yield curve control or similar repression. Tariffs can help redirect consumption toward domestic production, but they must be stable and much larger than current levels to change investment incentives. Gold is the only reserve asset large and neutral enough to absorb global reserve flows if Treasuries are debased or capped. Bitcoin is weakening because it has not decoupled from tech, is facing rising real capital costs, and may be responding to broader systemic stress and quantum-computing concerns. AI likely produces a deflationary shock in consumer credit and employment first, which then forces government and banking-system intervention, making the end result inflationary again. The bond market is not a safe haven in a systemic stress event because hedge funds, banks, and foreigners may all become sellers at once. China’s long-term advantage comes from choosing grid and industrial capacity over asset-price inflation, even if that meant weaker local equity performance for years. Short-term policy inconsistency—tariffs, immigration, AI subsidies, and swap lines—reflects an attempt to delay the unavoidable choice rather than solve it.

Data Points: U.S. debt-to-GDP: 120% - Used to argue that the U.S. cannot tolerate much higher long-term yields without market dysfunction. 10-year Treasury yield: 4.8% referenced as problematic - Luke says the economy and equity markets do not handle 10-year yields around this level well. Fed balance sheet growth in WWII: 10x in three years - Cited as the scale of action needed if the U.S. wants to pursue a wartime industrial mobilization. WWII-era deficits: 27% of GDP - Used as a benchmark for the magnitude of deficits required for large-scale industrial reordering. U.S. net international investment position: -85% of GDP - Shows the U.S. has gone from net creditor to major net debtor over four decades. Gross foreign holdings in U.S.: $63T - Luke cites foreigners as holding a very large gross claim on U.S. assets. Net foreign claims on U.S.: $26T - Referenced as the scale of capital that cannot simultaneously fund Treasuries and new domestic factories. Fed share of long-term Treasury issuance: 37% since Jan. 2022 - Used to illustrate the importance of leveraged basis trades and official support in Treasury absorption. Hedge fund Treasury holdings: ~$1.8T - Luke says levered hedge funds hold a huge amount of Treasuries funded in repo. Banks’ Treasury/agency holdings: ~$4T - Used to show banks are also key holders who may have to sell under stress. Foreign-held U.S. Treasury stock: ~$8.5T to $9T - Cited as the Treasury stock foreigners could sell if the dollar strengthens or funding needs rise. Dollar-denominated debt outside the U.S.: ~$13T - Used to explain why a stronger dollar can trigger foreign selling of dollar assets. Gold in global FX reserves: 28% - Luke says gold already represents a large share when marked to market. U.S. dollar in global FX reserves: 46% - Used to show gold could plausibly overtake the dollar if current trends continue. China electricity grid vs. U.S.: China surpassed the U.S. around 2008-2009 - Part of the argument that China chose physical infrastructure over financialization. Chinese grid buildout: Installed capacity equal to the entire U.S. grid in the last 10 years - Illustrates the scale of China’s infrastructure advantage. AI demand by 2030: Could equal all current global electricity generation - Used to show how severe the power bottleneck could become. Cost of nuclear power per gigawatt: China ~1/6 of U.S. cost - A testimony example used to argue U.S. input costs are structurally uncompetitive. Implied dollar devaluation vs. yuan per gigawatt: ~87% - Luke’s rough calculation based on the gigawatt cost gap. Tariff level needed (illustrative): Possibly 300%+ - Used to suggest current tariffs are far too low to offset cost differentials. U.S. ISM manufacturing: 47-ish / new orders ~44 - Described as recessionary and inconsistent with an AI-led boom alone. U.S. unemployment for under-25 bachelor’s holders: 7% - Cited as an early warning sign of AI-driven labor displacement. OpenAI reported losses: Half a trillion dollars over five years - Used to frame AI as too-big-to-fail and potentially bailout-dependent. AI capex in the U.S.: $3T from 2025 to 2028 - A forecast used to highlight the scale of capital demand competing with Treasury issuance. Gold’s rise vs. oil: 6 barrels/oz in 2007 to 72 barrels/oz now - Used to argue gold’s real value has re-rated sharply against commodities. Silver performance: Up ~80% over the past year; roughly tripled over 3 years - Presented as evidence of broad precious-metals demand. Gold performance: Up ~60% over the past year - Attributed to central-bank buying and reserve-asset substitution. U.S. true interest expense: 96% of trailing 12-month receipts - Shows fiscal room is very limited even before a recession. Private sector AI-related layoffs: Amazon planned 160,000 hires but may not hire them - Example of expected labor displacement from AI adoption. Gold reserve ratio examples: Official U.S. gold at market value = 12% of foreign-held U.S. debt; long-run average 40%; 1989 20%; 1980 130-135% - Used to justify a potential 3x-5x higher gold price.

Pivotal Quotes: "The U.S. can either, if it wants to reshore... It cannot have reshore and have the capital here in your capital markets. You can't do the two opposite sides of the same balance sheet. You got to choose one or the other." — Luke Gromen: Core tradeoff between industrial policy and preserving Treasury-market value. "What do you want? We want to reshore and we want to have 10-year yields... You can have 10-year yields at four, you can have them at three, and you can reshore. Absolutely. Sure... The trade-off is the dollar is no longer reserve currency, gold is." — Luke Gromen: The episode’s central framing: reshoring requires sacrificing monetary/financial dominance. "Gold wins in inflation or deflation. And it's the only asset on the board that wins in inflation or deflation." — Luke Gromen: Why gold is his preferred reserve and crisis asset across both macro regimes.

Implications: Listeners should expect more policy volatility, higher real-resource inflation, and persistent pressure on bonds/dollars as the U.S. pushes AI and reshoring. Gold looks structurally favored; Bitcoin and long-duration Treasuries look vulnerable until the market and policymakers choose a clear path.