1. Strategic Actions and Decisions * Investigate Counterparty Risk in AI-Driven Lending: The failure of Community Bank and Trust (GA) due to an AI underwriting algorithm suggests a new class of operational risk. Audit any third-party AI vendors or automated loan origination systems for concentration risk and fraud vulnerability. [00:04:06] * Prepare for a Liquidity Squeeze in Hyperscalers: Free cash flow is evaporating at Meta and Amazon, forcing debt issuance while dollar shortages emerge globally. Reduce exposure to Mag7 equities that rely on continuous CapEx spending to sustain valuations, as the market may soon penalize cash incineration. [00:11:28] * Reallocate from Hyperscalers to Physical Enablers: Capital expenditure is flowing to engineering, construction, and electrical component stocks (PWR, ETN, WCC), which are outperforming the Mag7 by over 70% year-to-date. Rotate portfolio weight into industrial picks-and-shovels plays that benefit from AI buildout without balance sheet risk. [00:20:39] * Monitor Dollar Shortages for Gold Entry Point: The dollar cannot break 101, and central banks (Japan, UAE) are intervening, signaling potential dollar weakness. Initiate or add to gold and silver positions if the DXY breaks below 96, as sentiment in miners is at extreme lows (11.54 bullish percentile). [00:43:30] * Focus on Energy Service Companies Pivoting to Data Centers: Natural gas demand from AI is creating structural tailwinds. Focus on energy service companies transforming from pressure pumping to power providers for “behind-the-meter” data center electricity, rather than traditional oil tankers. [01:17:13] 2. Executive Summary The discussion centers on the first “AI-induced bank failure” in Georgia, where executives used an algorithm to mass-produce SBA loans, resulting in a 50% capital loss with taxpayers on the hook for 75% of losses. Concurrently, Meta and Amazon are burning cash on AI CapEx so aggressively that free cash flow is nearly gone, forcing debt issuance. The panel recommends rotating out of hyperscalers and into the physical economy: engineering and construction stocks (up 70% year-to-date), memory chips (SanDisk signing five-year prepaid contracts), and energy infrastructure. A liquidity crisis is looming due to dollar shortages and rising bond yields (ten-year at one-year highs). Gold sentiment is in the “toilet” at the 11th percentile, presenting a favorable risk-reward entry, while energy remains critically underweighted at only 3-4% of the S&P 500 despite massive structural demand. 3. Key Takeaways and Practical Lessons 1. AI is a Double-Edged Sword for Financial Risk: The Georgia bank failure proves that AI underwriting without human oversight created catastrophic losses (50% of capital lost). Executives bragged on a podcast about replacing “banking relationships” with algorithms, leading to presumed fraud where fake companies got automatically approved for government-backed loans. * Practical Lesson: Require manual review of government-guaranteed loans (SBA and USDA) issued via AI. Ensure the 25% unguaranteed portion is not securitized into “SOUP” and sold to yield-chasing pensions, as happened with this bank. 2. Free Cash Flow is the Only Truth in the AI Bubble: Hyperscalers are hiding debt via special purpose vehicles and shrinking free cash flow to service CapEx. Google inflated earnings via a change in the valuation of their Anthropic stake, and Meta issued another $25 billion in debt after reporting. * Practical Lesson: Ignore adjusted earnings. Screen for companies where operating cash flow is declining while capital expenditures are rising more than 30% year-over-year, and avoid those with negative tangible free cash flow. 3. The Bottleneck is Physics, Not Chips: Data centers are being canceled due to grid transmission limits and public opposition (Virginia gigawatt project pulled by Brookfield). You cannot code your way around turbine blade production or water availability. * Practical Lesson: Invest in companies solving physical constraints: transformer manufacturers (Eaton), electrical parts distributors (Wesco), and natural gas turbine servicers, not the data center operators themselves. 4. Local Inference is a Threat to Cloud AI: Running AI models locally on high-RAM hardware (Apple Mac Studios with 128-256GB) solves privacy and legal issues for finance, healthcare, and legal sectors, bypassing expensive cloud tokens where costs are rising. * Practical Lesson: Be cautious on cloud AI compute providers and consider hardware enablers that benefit from the shift to edge computing, as inference represents 80 to 85% of data center demand today. 5. Commodity Currencies Signal a Shift to Hard Assets: The Australian Dollar and Brazilian Real are breaking out versus the USD, which is the first time since 2002 that every commodity-rich country’s currency is accelerating against the dollar. Gold sentiment has pulled back to the 11th percentile. * Practical Lesson: If the US dollar breaks below 96, increase allocation to gold miners and aluminum (which is blowing away S&P performance), as the “inflate or die” policy will struggle against inelastic demand for food and energy. Follow Nobody Special on X here - @JG_Nuke Follow David Nicoski on X here - @davevermilion Follow Bob Coleman on X here - @profitsplusid Follow Zach Marx on X here - @zmarx_the_spot Watch on youtube Below - This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit georgenoble.substack.com/subscribe