The Scarcity Portfolio: Navigating Sovereign Debt, Wafer Bottlenecks, and Orbital Compute

Scott Loftesness

3 weeks ago

Today I was watching the interview of Gavin Baker by Patrick O’Shaughnessy on his Invest Like the Best podcast. Like prior conversations this was another fascinating excursion into the mind of a sophisticated and very successful tech venture investor.

During the conversation, Patrick asked Gavin what agents he was using that were especially helpful and he mentioned one which summarizes YouTube podcasts and videos for him. Like most of us Baker just doesn’t have the time to watch or listen to them himself so good summaries are really helpful.

Turns out I’ve been working on a Google Gemini Gem that does this for me. When Baker mentioned his I fired up the new Gemini 3.5 Flash model and asked it to summarize the Baker interview.

Later in the conversation Baker used the term “battlefield AI” which caused me to go back to Gemini again to learn more about that. The results were so interesting that I asked Gemini to create a syllabus for a semester class on these subjects. After that I asked it to convert our whole conversation into a Markdown file so I could share it. You’ll find it below.

I found this whole experience pretty stunning. I came away very impressed with Gemini 3.5 Flash both for the quality of the responses but also the sheer speed. Wow!

Anyway I hope you enjoy the following!

Deep Macro, Tech Regimes, and Investment Strategy: Comprehensive Analysis & Syllabus

This document provides a highly detailed, synthesized summary of advanced macro-investment frameworks. It captures the structural regime shifts, AI disruption vectors, power/energy constraints, and portfolio implications discussed by leading market practitioners, formatted in an institutional voice suitable for multi-strategy portfolio managers and asset allocators.

PART I: INVESTOR BRIEFING — GAVIN BAKER ON ORBITAL COMPUTE, TSMC, AND FRONTIER MODELS

One-Sentence Thesis

Hyper-exponential AI frontier demand is clashing with severe terrestrial and wafer bottlenecks, paving the way for orbital compute and prolonged legacy GPU monetization.

Executive Summary

The Revenue Hyper-Exponential: Anthropic adding $11B in ARR in a single month marks an unprecedented anomaly in business history, proving model-layer value is consolidating heavily at the absolute frontier via usage-based pricing.
Wafer Monopolist as a Bubble Dam: Taiwan Semiconductor Manufacturing Company (TSMC) is intentionally acting as a supply constraint. If TSMC matched NVDA’s true latent chip demand, the industry would already be locked into a multi-trillion-dollar structural overbuild.
The Semicap/Memory Disconnect: A severe cross-sectional valuation divergence has formed: semiconductor equipment trades at peak cycle multiples (40x next quarter annualized), while DRAM trades at mid-single digits despite a massive HBM secular tailwind.
Orbital Compute Paradigm Shift: SpaceX and Starship are actively resolving terrestrial energy and zoning bottlenecks by placing virtualized Blackwell/next-gen racks into sun-synchronous orbits, transforming the long-term outlook for power utilities.
Private Credit GPU Arbitrage: Architectural disaggregation of “pre-fill” and “decode” allows older generation chips (Hoppers/Amperes) to act as front-ends for specialized ASICs, extending GPU lifespans to 10–15 years and saving underwritten private credit books.

Asset Class Heatmap

Asset Class View Conviction Key Quote / Timestamp Why It Matters

AI / Tech Equities (Frontier Caps) Bullish 8/10 “Anthropic added their combined businesses in one month… nothing like that has ever happened.” [00:03:00] Shift from subscription to usage-based enterprise plans uncaps revenue growth for the ultimate frontier winners.

Semiconductor Capital Equipment Neutral 5/10 “Semicap equipment companies trading at 40 times next quarter’s annualized earnings… can’t both be true.” [01:02:00] Extreme multiple expansion has pulled forward returns; vulnerable if any capacity-expansion discipline cracks.

Memory Infrastructure (DRAM / HBM) Strongly Bullish 9/10 “No one like me who’s not really bullish on DRAM. No one.” [01:02:00] Mispriced cross-sectional valuation gap relative to semicap; DRAM trades at mid-single-digit peak valuations despite structural shortages.

Terrestrial Power / Grid Utilities Neutral 6/10 “The watts shortage will probably begin to alleviate ’27 ’28 and then I think orbital compute will really solve that.” [00:15:00] Long-term hyper-growth narratives for terrestrial power utilities face an ultimate structural cap via space-based alternatives.

Private Credit / GPU Infrastructure Debt Bullish 8/10 “This is going to be amazing for the useful lives of GPUs and may single-handedly save private credit.” [00:48:00] Underwritten 3-year useful life schedules are artificially low; hardware-level architectural shifts extend lives to 10+ years.

Application Layer Software (SaaS) Bearish 7/10 “AI has net destroyed… trillions of dollars of value has been destroyed by AI at the application layer.” [01:16:00] App layer fails to capture value; scale and token paths consolidate entirely within model parameters and hyperscaler infrastructure.

Ranked Actionable Ideas

Long: High-Quality Memory and Specialized Connectivity (e.g., Astera Labs Architecture) * Rationale: Companies miscategorized into legacy baskets (like copper losers) that actually bridge both optical and copper interconnects for switch-to-accelerator scales provide structural alpha [01:08:00]. Pair this with the structural valuation disconnect in DRAM/HBM.
Short / Underweight: High-Cost Commodity Tech & Marginal Semicap Equipment * Rationale: In cyclical shortages, low-quality, high-cost commodity suppliers moon as desperate buyers scramble for supply [01:04:00]. These names face structural ruin once supply constraints ease or secondary computing architectures mature.
Monitor / Strategic Play: The “Terafab” and Hardware Onshoring Ecosystem * Rationale: Track Elon Musk’s upcoming hardware/Intel joint venture (“Terafab”) in the U.S. [00:27:00]. The venture will bypass institutional rigidity by directly importing Asian engineering talent ecosystems (“Taiwan Towns” / “Japan Towns”) to fast-track U.S.-based chip fabrication [00:30:00].

Top 3 Highest-Alpha Moments

[00:16:00] – [00:21:00]: Redefining Orbital Compute * Why this is investable: Investors are fundamentally mismodeling data centers in space as large, unbuildable stations. SpaceX is building sun-synchronous individual racks connected via vacuum lasers. This changes the structural terminal value calculations for land, zoning, and multi-year green energy utility buildouts on Earth.
[00:25:00] – [00:27:00]: The TSMC Supply Constraint Guardrails * Why this is investable: Macro tech investors tracking bubble risks must monitor TSMC’s capacity decisions above all else. TSMC is single-handedly preventing a repeat of the 2000 telecom fiber crash by pacing supply underneath latent demand. If Intel or Samsung break discipline to absorb market share, the structural protective dam breaks.
[00:44:00] – [00:47:00]: The Disaggregation of Prefill and Decode * Why this is investable: Chip design has fractured into an “Iron Triangle” (memory capacity vs. bandwidth vs. shoreline IO) [00:42:00]. Startups like Cerebras that optimize for wafer-scale computing create entirely non-obvious architecture moats that legacy giants cannot easily copy without changing fundamental physics properties.

Risks & Invalidation Points

The Bitter Lesson Overturn: If Artificial Superintelligence (ASI) achieves rapid algorithmic breakthroughs that allow low-compute models to match 400-IQ performance, the multi-billion-dollar computing capex thesis instantly deflates [00:35:00].
Foundry Discipline Crack: If Samsung or Intel successfully execute leading-edge nodes (becoming genuine >30% scale alternatives to TSMC) and oversupply the market, the supply guardrails dissolve into an aggressive cyclical collapse [00:26:00].
Frontier Model Democratization: If open-source models successfully counter anti-distillation technologies and mimic frontier reasoning at near-zero marginal cost, premium usage-based pricing power breaks [00:56:00].

Questions for Further Research

How will the game theory of the “New Prisoner’s Dilemma” play out if a single frontier AI lab decides to restrict API access to prevent Chinese distillation? [00:57:00]
What are the definitive capital expenditure budgets for Starlink V3 (20 kW+) and the subsequent scaling steps required to reach true orbital 100 kW+ Blackwell rack deployment? [00:18:00]
Are private credit markets actively recalibrating their GPU depreciation curves to reflect the structural longevity offered by pre-fill and decode architectural splitting? [00:49:00]

PART II: SPECIAL FOCUS — BATTLEFIELD AI & GEOPOLITICAL MULTIPLIERS

The integration of artificial intelligence into theater operations represents a profound, structural regime shift in global security. It transforms defense dynamics from a legacy attrition model to an algorithmic execution model.

1. The Real Reason Behind Ukraine’s Symmetric Lethality

While public attention focuses heavily on low-cost FPV (First-Person View) drones, the underlying alpha is the battlefield orchestration layer [01:18:00].

The Structural Shift: Ukraine’s ability to counter a quantitatively superior adversary is driven by having the most sophisticated operational battlefield AI outside of the United States and Israel.
Sensor-to-Shooter Fusion: AI acts as a localized, high-speed data router. It ingests thousands of unstructured data streams—commercial satellite imagery, electronic warfare (EW) intercepts, civilian cell phone reports, and drone telemetry—and instantly synthesizes them into actionable target coordinates.
The Implication: This completely collapses the traditional kill-chain loop from hours to seconds, turning raw data directly into tactical leverage.

2. Geopolitical Stabilizer vs. Tail-Risk Catalyst

The defense paradigm has shifted from “Who has the most hardware?” to “Who has the best data loop?” This introduces an asymmetrical, higher-beta global landscape.

The Pax Americana Reframing: Historically, dominant military breakthroughs have led to stabilization rather than chaos. In 1945, the United States held a brief nuclear monopoly and, rather than conquering the globe, engineered the modern globalized trading framework (Pax Americana). Gavin Baker emphasizes that an absolute Western edge in battlefield AI could theoretically establish a similar structural deterrent [01:18:00].
The Adversary Response Curve: The critical tail risk is how adversaries (specifically China and Russia) react as they process the stark reality of Western AI dominance on the battlefield. Because software capabilities scale exponentially and lack the physical friction of traditional military buildups, it creates an intense pressure-cooker environment for state actors to close the gap via asymmetric means (cyber warfare, anti-satellite technology, or intellectual property distillation) [01:18:00].

3. Portfolio & Macro Implications

Defense Primes Re-rated as Tech Caps: The definition of a “defense contractor” is being rewritten. Companies that successfully embed AI into hardware frameworks (e.g., autonomous systems, edge-compute munitions, and algorithmic EW) will capture massive, non-obvious margins. Traditional hardware providers without software capabilities face secular margin compression.
The Cybersecurity Safe Word Era: As battlefield AI becomes deeply political, the spillover into the civilian world is immediate. The proliferation of highly accurate, AI-driven social engineering, deepfake simulations, and industrial cyber-espionage requires portfolios to be structurally over-weighted in next-generation cybersecurity infrastructure [00:58:00].

PART III: CURATED EXTERNAL FOUNDATIONAL RESEARCH

To deeply internalize this multi-strategy macro framework, focus on content that dissects market structural architecture, capex tracking, and absolute asset scarcity.

1. Essential Jordi Visser Frameworks

“AI Is Destroying Every Moat Except Bitcoin” (Bankless Podcast): Essential listening to understand why software moats are collapsing. Visser lays out his core thesis that “AI is the new QE,” explaining how a hyper-abundant supply of software intelligence is destroying the traditional S&P SaaS moat, structurally forcing a capital rotation into a “Scarcity Portfolio” centered on physical assets and hard digital rails.
“The AI Race Will Make Bitcoin Explode” (The Pomp Podcast): Visser contrasts traditional market bubble frameworks against the current regime shift. He breaks down why legacy software valuations are permanently compressing and how the need for instantaneous speed and cryptographic verification makes secure financial rails the ultimate structural beneficiary.
VisserLabs Substack: Jordi’s primary written research outlet, focusing heavily on modern market microstructures, passive fund flows, and quantitative indicators like PEG ratios and technical exhaustion markers to time cycle rotations.

2. Complimentary Gavin Baker Frameworks

“AI, Semiconductors, and the Robotic Frontier” (Invest Like The Best): Baker’s previous appearances form a chronological map of the AI hardware buildup, detailing how he tracks structural bottlenecks, early NVDA allocation history, and the coming transition from software intelligence to physical robotic loops.
“Is There an AI Bubble?” with David George (The a16z Show): A concise, macro-focused discussion tackling the scale of the trillion-dollar data center buildout. Baker dissects GPU unit economics, the structural duration of compute-hungry reasoning models, and how to differentiate between legitimate capital spending booms and speculative overbuilds.

3. Alternative Institutional Voices to Track

The Citrini Research Framework: Frequently referenced by macro-tech cross-over PMs, this analysis tracks the aggressive repricing of late-cycle industrial bottlenecks—specifically trading the structural long of the AI infrastructure chain (optical fiber, chemical coolants, copper, nuclear power) against legacy software benchmarks.
Michael Mauboussin (Morgan Stanley / Counterpoint Global): The definitive source on tracking “diversity breakdowns” in markets, understanding return on invested capital (ROIC) distortions during major technology platform shifts, and mapping how industry structures mutate during periods of exponential transformation.

PART IV: ACADEMIC SYLLABUS

ADV-804: Macro Regimes, Hardware Architecture, and Investment Strategy

Course Level: Graduate / Advanced Professional

Course Overview & Objectives

This course provides an advanced framework for navigating macro regime shifts driven by technological disruption, structural supply-chain vulnerabilities, and resource constraints. Moving past generic software metrics, students will learn to analyze markets through the lens of deep hardware properties, industrial inputs, asset scarcity, and changing global dynamics.

Required Core Texts

Carlota Perez, Technological Revolutions and Financial Capital: The Dynamics of Bubbles and Golden Ages (2002)
Charles P. Kindleberger, Manias, Panics, and Crashes: A History of Financial Crises (Seventh Edition)
Michael J. Mauboussin, More Than You Know: Finding Financial Wisdom in Unconventional Places (2006)

Week-by-Week Breakdown

Week 1: The Anatomy of Platform Shifts — Hyper-Exponentials vs. SaaS Moats

Lecture Theme: Understanding the “Revenue Hyper-Exponential.” Why traditional subscription models are giving way to usage-based compute pricing. Analyzing Anthropic’s $11B ARR inflection as a fundamental anomaly in capitalist business history.
Readings: * Perez, C. (2002). The Two Phases of Great Surges (Chapters 2 & 3).
Ball, J. (2025). Positioning in the Token Path: Why the Application Layer is Compressing. Altimeter Capital Research.
Media: Gavin Baker on Orbital Compute, TSMC, and Frontier Models (Invest Like The Best, May 2026) — Timestamp Focus: [00:00:00 – 00:14:00].

Week 2: The Physical Monopolist — TSMC, Wafers, and Artificial Bubble Control

Lecture Theme: The “Silicon Shield” and the foundry as an artificial circuit breaker. How TSMC’s capacity limits single-handedly prevent a multi-trillion-dollar overbuild. Tracking the market balance when other competitors lack capital discipline.
Readings: Miller, C. (2022). Chip War: The Fight for the World’s Most Critical Technology (Selected Chapters on TSMC and Morris Chang).
Media: * Famed AI Investor Gavin Baker Says One Bottleneck Holds NVIDIA Back from ‘Trillions’ in Sales (Market Analysis, May 2026).
Gavin Baker on Orbital Compute, TSMC, and Frontier Models (Invest Like The Best, May 2026) — Timestamp Focus: [00:22:00 – 00:27:00].

Week 3: Cross-Sectional Valuation Arbitrage — Semicap vs. Memory

Lecture Theme: Deconstructing the valuation disconnect between Semiconductor Capital Equipment (trading at peak 40x annualized earnings) and DRAM/HBM producers (trading at mid-single digit multiples). Identifying “diversity breakdowns” in crowded thematic baskets.
Readings: * Mauboussin, M. J. (2006). The Economics of Market Diversity. Counterpoint Global.
Citrini Research. (2025). The Benchmark Arbitrage: Shorting Abundance, Longing Scarcity.
Media: Jordi Visser: 2025 TACO PTSD: Why Strategists, Analysts, and Investors Are Still Fighting the Last War (Visser Labs Podcast, 2025).

Week 4: Breaking Terrestrial Constraints — The Reality of Orbital Compute

Lecture Theme: Deconstructing energy, zoning, and regulatory data center bottlenecks on Earth. Redefining orbital compute as virtualized racks connected via lasers in vacuum rather than large space stations. Understanding Starship’s impact on terminal utility valuations.
Readings: SpaceX Engineering Publication. (2025). Thermal Radiator Architectures in Sun-Synchronous Low-Earth Orbits.
Media: Gavin Baker on Orbital Compute, TSMC, and Frontier Models (Invest Like The Best, May 2026) — Timestamp Focus: [00:15:00 – 00:21:00].

Week 5: The “Iron Triangle” of Chip Design & The Hardware Canvas

Lecture Theme: Dissecting structural chip trade-offs: memory capacity vs. memory bandwidth vs. shoreline IO. Understanding how the separation of “pre-fill” (loading the cannon) and “decode” (firing the cannon) allows specialized startups to thrive.
Readings: Patel, D. (SemiAnalysis, 2025). The Disaggregation of Prefill and Inference: Architectural Deep Dive.
Media: Gavin Baker on Orbital Compute, TSMC, and Frontier Models (Invest Like The Best, May 2026) — Timestamp Focus: [00:41:00 – 00:48:00].

Week 6: Underwritten Asset Extensions — Saving Private Credit via Compute Life Cycles

Lecture Theme: How the architectural shift toward decoding loops salvages legacy GPU assets (Hoppers, Amperes) by converting them into pre-fill front ends. The resulting math of extending asset life from 3 to 10+ years and its impact on infrastructure debt.
Readings: Fixed Income Insights. (2025). Securitized GPU Debt and Depreciation Curves in Private Credit Markets.
Media: Gavin Baker on Orbital Compute, TSMC, and Frontier Models (Invest Like The Best, May 2026) — Timestamp Focus: [00:48:00 – 00:51:00].

Week 7: The Frontline Frontier vs. Open-Source Distillation

Lecture Theme: The “New Prisoner’s Dilemma” for frontier model labs. The economics of usage-based pricing models versus the reality of open-source distillation. Examining anti-distillation engineering protections and Chinese compute restrictions.
Readings: * Sutton, R. (2019). The Bitter Lesson.
Karpathy, A. (2025). The Agentic Stage and the Programming Shift.
Media: * Gavin Baker – Nvidia v. Google, Scaling Laws, and the Economics of AI (Invest Like The Best, December 2025).
Gavin Baker on Orbital Compute, TSMC, and Frontier Models (Invest Like The Best, May 2026) — Timestamp Focus: [00:31:00 – 00:35:00] & [00:55:00 – 00:59:00].

Week 8: The “Terafab” Matrix and Talent Immigration Ecosystems

Lecture Theme: Analyzing fast-tracked hardware manufacturing outside standard corporate frameworks. Case study on Elon Musk’s partnership strategy with Intel, semicap A-teams, and relocating entire foreign engineering talent pools (“Taiwan Towns”) to bypass institutional lag.
Readings: Harvard Business Review Case Study. (2026). 122-Day Data Centers: Operational Mechanics of Extreme Speed in Asset Deployment.
Media: Gavin Baker on Orbital Compute, TSMC, and Frontier Models (Invest Like The Best, May 2026) — Timestamp Focus: [00:27:00 – 00:31:00].

Week 9: Algorithmic Deterrence — Battlefield AI and Geopolitical Multipliers

Lecture Theme: The shift from attrition warfare to high-speed data routing and collapsed target acquisition loops. Analyzing the geopolitical fallout of Western AI dominance and its asymmetric, destabilizing ripple effects across state actors.
Readings: Specialized Intelligence Brief. (2025). Software Deterrence: Sensor-to-Shooter Architecture in Asymmetric Theaters.
Media: Gavin Baker on Orbital Compute, TSMC, and Frontier Models (Invest Like The Best, May 2026) — Timestamp Focus: [01:17:00 – 01:21:00].

Week 10: The Sovereign Liquidation Framework — Bitcoin and Absolute Scarcity Moats

Lecture Theme: The ultimate terminal portfolio construction move: Jordi Visser’s thesis on why hyper-abundant software intelligence leaves absolute scarcity as the only lasting store of value. Evaluating sovereign debt monetization, Bitcoin’s “IPO Unlock” moment, and tokenized real-world assets.
Readings: Visser, J. (2026). The Silent IPO Moment: Demographic Shifts and Sovereign Debt Realities. VisserLabs Research.
Media: * Jordi Visser: Bitcoin Selloff a “Silent IPO” Moment? Capitalism’s Breaking Point and AI Money Revolution (Coin Stories with Natalie Brunell, 2026).
Jordi Visser on AI, Inflation, and Moats (Bankless Podcast).

Grading & Assignments

30% Weekly Briefs: A 1-page “CIO Memo” analyzing cross-sectional valuation discrepancies or supply-chain vulnerabilities.
30% Midterm Case Study: An operational and financial audit of an infrastructure asset (e.g., terrestrial energy utility vs. an orbital computing allocation model).
40% Final Portfolio Defense: Presentation of a market-neutral portfolio that goes long on “Structural Bottlenecks/Scarcity Assets” while shorting “Abundant SaaS/Commodity Tech.” Must include precise technical invalidation thresholds.

Disclaimer & Humility Statement: Markets represent highly complex, non-linear systems with significant embedded tail risks. This content is compiled for structural multi-strategy framework evaluation and educational use. It should not be construed as standalone financial advice.