Gavin McCracken (Pt. 2): Robotics, AI, and A Commodities Supercycle

Episode Summary

Executive Summary: The conversation links AI, math, robotics, and commodities into a single thesis: AI will mostly improve search and pattern-finding, not replace human invention, and its biggest real-world effect will be accelerating demand for compute, memory, robotics, and raw materials. The speakers argue this sets up a commodity super cycle, especially in silver, tungsten, oil, and North American miners, while geopolitical fragmentation increases supply risk and volatility.

Main Topics: AI as advanced search, not true intelligence: The guest argues LLMs are excellent at finding counterexamples, gaps, and patterns, but do not create conjectures or true understanding. He frames current AI as powerful but fundamentally limited, better described as virtual intelligence than superintelligence. End of the information era and rise of a commodity super cycle: He claims the tech tree of the information age is nearing exhaustion, with future growth shifting toward robotics, analog computation, memory, and physical build-out. That shift should increase demand for raw materials and industrial inputs. Robotics as the next major demand driver: Robotics is presented as the bridge between AI and commodity demand. The hosts discuss how widespread deployment of robots would require metals, materials, and manufacturing capacity, making miners and materials suppliers attractive. North American energy and geopolitical supply shocks: The discussion emphasizes oil-market fragility from Strait of Hormuz risk, possible U.S./Canada export restrictions, and war-related disruptions. The guest argues these factors justify hedging with Brent and North American E&Ps/miners. Silver, tungsten, and critical minerals trade: Silver is highlighted as a beneficiary of AI, solar, EVs, and robotics demand. Tungsten is framed as a defense-linked material with strong wartime demand, while resource nationalism and bifurcated markets may create pricing distortions and alpha. Portfolio positioning, margin, and optionality: The guest discusses his own positioning in miners, Brent calls, oil, and selective exposure to memory/compute names. He prefers raw materials and options on geopolitical shocks over crowded tech equities. Market regime, debasement, and hidden stress: Both speakers note that markets feel detached from reality, with many assets rising together amid monetary debasement and high leverage. They worry an oil spike could trigger a liquidity crunch and force a broader repricing.

Key Arguments: AI is best viewed as intelligent search: it excels at finding counterexamples, loopholes, and patterns, but it does not independently generate worthwhile conjectures or true understanding. Mathematics will be one of the first fields materially changed by AI because AI can handle high-dimensional search and bookkeeping that humans struggle to manage. The world is nearing the end of the information era; the next phase will be more physical, centered on robotics, analog-style computation, and industrial build-out. A robotics boom would create major demand for industrial commodities such as silver, aluminum, tungsten, and memory chips. Memory and compute are effectively becoming commodities because AI infrastructure requires massive, repeated inputs of these resources. North American oil and gas assets are attractive because global supply is vulnerable to geopolitical shocks, export bans, and regional retaliation. Brent is the cleaner hedge versus WTI risk because an export ban or North American disruption could cause Brent to reprice violently. Tungsten is structurally attractive due to defense rearmament and missile production demand. Gold remains supported by debasement and capital inflows, but a severe liquidity crunch could still create a sharp drawdown. The market is increasingly driven by evolutionary dynamics and feedback loops, so investors should think in systems, not linear narratives.

Data Points: Conference: Topology, Algebra, and Geometry, and Data Science - AI conference attended by the guest in Boston Call on AI era: "end of the information era" - Guest’s framework for the current macro/tech cycle Compute bottleneck: Matrix multiplication - Described as core operation in deep neural networks Historical algorithm: Fourier transform in n log n time - Used as an example of wartime-driven innovation Oil drawdown: 1 to 2 million barrels a day - Guest claims this is the current global stock draw speed China SPR/player estimate: 5 million barrels a day - Guest claims China had been the largest SPR participant Brent-WTI spread: $6 - Discussed as higher than normal, though smaller than expected Typical Brent premium: $3 - Guest said Brent usually trades at about this premium Saskatchewan royalty holiday: 2.5% royalty on first 38,000 barrels - For first production from southeast Saskatchewan wells Typical royalty rate: 25% - Compared against Saskatchewan’s reduced royalty regime Gold correction: 50% cut in gold miners - Guest reduced exposure anticipating liquidity stress Silver target/floor: $100 - Guest said he sees $100 as a likely floor price for silver Net worth move: 50% move up from oil breakout; 60% drawdown on peace deal narrative - Guest described portfolio volatility and recovery AI solution type: Counterexamples - Guest said AI is mainly proving conjectures false rather than true Tanker hits: About 50 year-to-date - Guest claims product tankers have been hit, supporting his oil-risk thesis

Pivotal Quotes: ""I actually think that we're at the end of the information era."" — Guest: Defines the macro thesis connecting AI, robotics, and commodities ""AI is really just been proving things false. I haven't seen it do anything as a true that I was, you know, impressed by."" — Guest: Summarizes his view that current AI is strong at search but weak at original creation ""I think silver's inflection is like, I think actually it's front-running robotics."" — Guest: Explains why he sees silver and miners as early beneficiaries of the robotics cycle

Implications: Listeners should expect continued strength in physical assets tied to AI, robotics, defense, and energy security. The key investable idea is to favor miners, memory, and North American resource exposure over crowded software narratives, while hedging geopolitical and liquidity shocks.