For most of the last decade, the narrative around artificial intelligence has been dominated by software: models, algorithms, and the companies that build them. That era is ending. The binding constraints are no longer just intelligence; they are energy, materials, cooling, connectivity, and the physical infrastructure required to run computational thought at scale.

This shift creates a meaningful opening for Latin America, particularly Brazil, that is frequently overlooked. While the region is unlikely to produce the next frontier foundation model, it possesses structural advantages in the physical layer that the AI buildout cannot ignore. The question is no longer whether LATAM can compete in AI software. The more relevant question is how individuals, companies, and governments can position themselves to capture durable value from the infrastructure layer that makes the agent economy possible.

The New Scarcity

Recent developments make the physical nature of AI impossible to ignore. Memory supply is sold out through 2027–2028. Hyperscalers are signing multi-billion-dollar deals for power and GPUs simply to keep pace. Connectivity bottlenecks are driving acquisitions such as Credo’s purchase of DustPhotonics. Even the largest technology companies are discovering that scaling intelligence requires securing real-world resources at unprecedented speed and cost.

At the same time, the emergence of AI agents transacting in stablecoins and prediction markets is creating new capital allocation rails. The software layer is becoming commoditized; the physical and financial infrastructure supporting it is not.

This is where Latin America’s position becomes interesting.

Brazil’s Asymmetric Advantages

Brazil sits on several of the inputs the AI infrastructure boom will consume most aggressively:

• Clean, abundant electricity, particularly hydroelectric power with relatively low marginal cost and strong ESG profile.

• Water resources critical for data center cooling at scale.

• Strategic minerals and metals required for chips, cabling, and energy infrastructure.

• Oil and gas that can provide flexible backup generation and petrochemical inputs.

These are not marginal advantages. As AI data centers and agent infrastructure multiply, power and cooling become first-order constraints. Jurisdictions with cheap, reliable, and relatively clean energy will attract capital even if they lag in model development. The same logic applies to copper, rare earths, and other materials needed to physically connect the intelligence layer.

The opportunity is not to “catch up” in software. It is to become indispensable in the physical stack.

How Different Actors Should Position

Individuals should think in terms of exposure rather than speculation. This means allocating to energy producers with credible expansion capacity, companies involved in transmission and cooling infrastructure, and stablecoin treasury strategies that can serve as neutral settlement layers in an agent-driven economy. The M&A lens is useful here: the best opportunities will likely come from acquiring or consolidating existing energy and materials assets rather than greenfield bets.

Companies have a clearer path through vertical integration and targeted M&A. Local energy groups can partner with or acquire data center developers. Mining companies can move downstream into materials critical for AI hardware. Financial institutions can build tokenized infrastructure vehicles or stablecoin rails that allow local capital to participate in global AI spend without relying solely on dollar-based systems. The concept of “Dataquisition” - acquiring companies primarily for their proprietary data, also applies locally: Brazilian companies sitting on rich operational or customer datasets in energy, agriculture, or logistics may become acquisition targets or valuable partners.

Governments face the highest-stakes decisions. The winning jurisdictions will be those that treat data center and energy infrastructure as strategic assets rather than regulatory afterthoughts. This includes streamlined permitting for power generation and transmission, tax and regulatory frameworks that favor long-term infrastructure investment, and sovereign or development bank capital deployed into critical minerals and energy projects tied to AI demand. Brazil has the resource base to become a meaningful supplier of both electrons and materials. The policy question is whether it will move fast enough to capture the wave rather than merely host it.

The Agent Economy Overlay

The rise of autonomous agents changes the capital formation dynamics. Agents already transact billions in stablecoins. Over time, a meaningful share of these flows may move outside dollar rails. Countries that develop credible local stablecoin or tokenized asset infrastructure linked to real energy and resource production can create closed-loop systems where AI spend circulates locally before leaking out. This is not about replacing the dollar. It is about capturing a larger share of the transaction volume and data that the agent economy will generate.

First Principles

The software layer of AI rewards speed, talent concentration, and network effects, areas where LATAM has structural disadvantages. The physical layer rewards geography, resource endowments, and the ability to execute large-scale infrastructure projects over decades. These are areas where the region has genuine optionality.

The winners in the next phase of AI will not be determined solely by who builds the smartest models. They will also be determined by who controls the energy, materials, and connectivity that allow those models to run. Latin America cannot win the first game. It can still compete in the second, if it stops treating AI as a software story and starts treating it as an infrastructure and capital allocation story.

The window is open, but it will not remain open indefinitely. The physical buildout is already underway. The question is whether the region will supply the foundation or merely consume the output.

Every week, the same split screen appears in financial media. On one side: earnings, macro prints, and price action - the ledger of what already happened. On the other: breakthrough papers, product launches, and policy drafts — signals of what might happen next. Most research treats these as separate workflows. Analysts cover sectors. Strategists cover geopolitics. Technologists cover roadmaps. Rarely does anyone hold both lenses at once, long enough, with enough rigor to ask a harder question: Where does human behavior collide with technological change, and what becomes investable at that collision?

That question is the thesis behind Intersect Alpha.

The greatest returns come from understanding human behavior + anticipating technological change and acting where those two worlds intersect.

This is not a prediction market. Intersect Alpha is a structured research pipeline. Eight specialized agents, a synthesis layer, a stress-test arena, and a principal editor, designed to produce long-horizon, source-backed thematic research for readers who think in decades, not quarters.

The problem with single-lens research

Traditional investment research optimizes for one of two modes.

The historian’s lens explains why people and institutions behave as they do: incentives, fear, status, regulation, demographics, conflict. It is indispensable, but it tends to lag. By the time a behavioral pattern is consensus, the edge is often gone.

The futurist’s lens imagines what technology could unlock: new cost curves, new bottlenecks, new winners in supply chains not yet priced. It is exhilarating, but it drifts. Without grounding in geopolitics, capital cycles, and human resistance to change, “sci-fi” theses become slide decks, not decisions.

The gap is not information. The gap is integration under adversarial scrutiny. Most teams either synthesize too early (one generalist memo) or never synthesize at all (silos of sector notes). Neither survives contact with a skeptical partner, a board, or a drawdown.

Intersect Alpha exists to close that gap with a repeatable method — not a one-off opinion.

The thesis: act at the intersection

Markets reward two rare combinations.

First, behavioral insight: understanding how history rhymes. How states, firms, and crowds respond to scarcity, shame, pride, and survival.

Second, technological anticipation: reading science fiction not as entertainment but as a map of plausible futures, then tracing which parts of that map are already being capitalized, regulated, or fought over.

The alpha - in the research sense, not a promise of returns - lives where those narratives overlap. A defense budget shift meets drone autonomy. An aging population meets regenerative medicine. A monetary regime shift meets on-chain settlement infrastructure. The intersection is narrower than a sector ETF and wider than a single product cycle.

Intersect Alpha names that overlap deliberately. Every agent in the pipeline is tasked with making the history/sci-fi bridge explicit before anything downstream treats a theme as actionable.

Eight agents, one pipeline

The pipeline is intentionally modular. Each agent has a domain, a deliverable, and a handoff contract. Together they form a research assembly line that human operators can run, audit, and improve cycle after cycle.

Phase 1 — Parallel sector research (Agents 1–4)

Four agents work in parallel, each producing a structured sector report grounded in the Intersect thesis.

1. Frontier covers the technological frontier: artificial intelligence and robotics, quantum computing, drones and defense, aerospace, telecommunications, and advanced transport. Frontier asks not only “what is technically possible?” but “who benefits when this stops being experimental?”

2. Matrix covers the physical economy: energy systems, chemicals, mining and critical materials, agriculture and food security. Matrix anchors sci-fi narratives in atoms — power grids, feedstock, land, water, and the geopolitics of supply.

3. Pulse covers human biology and experience: biomedicine, hospital systems, aesthetics and wellness, tourism and leisure, and the economics of intimacy and relationships. Pulse keeps the pipeline honest about demand that is emotional, not just industrial.

4. Agora covers markets and institutional design: finance, blockchain and digital assets, insurance, and prediction markets. Agora translates thematic conviction into how capital actually prices uncertainty.

Each sector report follows the same spine: trends with sources, chain maps, behavioral hooks from history, plausible futures from sci-fi analogues, and an explicit intersection thesis for that domain.

Phase 2 — Synthesis (Agent 5: Confluence)

Confluence reads all four sector reports and conducts a documented conceptual dialogue — geopolitics, human behavior, and futurology — to produce a final scenario set per domain plus cross-cutting global risks.

Confluence is where the pipeline stops being four parallel essays and becomes one narrative with tension. Where do scenarios reinforce each other? Where do they collide? Which assumptions, if wrong, break multiple themes at once?

Phase 3 — Expression (Agent 6: Exposure)

Exposure translates scenarios into a disciplined expression layer: a curated set of public-market instruments that represent each theme, with strict caps per domain and verified identifiers. Exposure is constrained by design - breadth is the enemy of conviction.

This step is where thematic research meets implementable ideas. It is also where many research products stop. Intersect Alpha deliberately does not.

Phase 4 — Adversarial stress test (Agent 7: Antithesis + Arena)

Antithesis is the devil’s advocate. It leads Arena Mode - a structured multi-round debate that stress-tests every thesis and every proposed expression before the final report.

Arena is not theater. It is protocol:

• Opening: surface fragile premises by domain and by idea.

• Counter-thesis: bull, base, and bear paths; valuation and geopolitical risk.

• Response: sector agents and Confluence defend with citations back to primary research.

• Protections: hedges, conditions, exit triggers.

• Verdict: Approve, Conditional, or Cut — with explicit conditions when conviction is qualified.

Antithesis carries a conservative bias by design. Disagreement is logged, not smoothed over. The goal is not consensus; it is survivable conviction.

Phase 5 — Editorial closure (Agent 8: Principal)

Principal composes the final research artifact for high-net-worth-minded readers: an executive letter, a map of opportunity versus risk, ranked themes, allocation tiers, deep dives, an Arena appendix, and a clear educational disclaimer.

Principal does not introduce new themes. It adjudicates what survived Arena.

How the pipeline flows

Visually, the method is a funnel with a feedback loop at the center.

Four explorers (Frontier, Matrix, Pulse, Agora) cast wide nets in parallel — faster than a sequential research sprint, and harder to bias toward a single narrative voice. Their outputs converge in Confluence, which is the first moment the pipeline must choose: which futures are mutually compatible, and which are in tension?

Exposure narrows. Antithesis widens the argument again, on purpose, by forcing every surviving idea through Arena rounds where defeat is an acceptable outcome. What remains is not “the team agreed” but “the team documented why disagreement shrank.”

Principal closes the loop for the reader: here is what we believe, here is what would falsify it, here is the record of challenge. That transparency is the product as much as any thematic conclusion.

Who this is for

Intersect Alpha is written for readers who:

• Allocate with a multi-year horizon and tolerate ambiguity.

• Want thematic depth without abandoning rigor or sources.

• Respect institutional skepticism - the kind that asks “what breaks this?” before sizing a position.

• Read history and sci-fi as inputs, not hobbies.

• Prefer research they can challenge in public, not slides they cannot inspect.

If your decision horizon is the next headline, this will feel slow. That is by design.

What we publish — and what we do not

Intersect Alpha publishes educational thematic research: scenarios, behavioral framing, technological maps, and the record of internal debate. It does not publish personalized advice, guaranteed outcomes, or urgency-driven calls to action.

Every cycle is a snapshot of a method under test. The method matters as much as any single conclusion.

Disclaimer: Intersect Alpha is educational research and opinion. It is not investment advice, a solicitation, or a recommendation to buy or sell any security. Past thematic research is not indicative of future results. Do your own due diligence and consult qualified professionals before making financial decisions

.

In the first half of 2026, a significant transaction closed involving a healthtech information system with deep roots across Latin America. The sellers’ key decision-makers sat in the United States and Europe. The buyers were based in Brazil. The majority of the installed base, and therefore the revenue and strategic value, lived in countries across Latin America.

Having been actively involved throughout the entire process, I had a front-row seat to how geography, culture, regulation, and technology now interact in sophisticated cross-border software deals.

The Multi-Jurisdictional Reality

When the core team leading the sale operates primarily from the U.S. and Europe while the buyers and end customers are concentrated in Latin America, every workstream becomes cross-cultural by default.

Time zones were the obvious challenge. More subtle and more consequential were differences in:

• How regulatory risk is priced

• The weight given to long-term client relationships versus short-term financial metrics

• The speed and formality expected in decision-making

Nonetheless, one of the most persistent and difficult challenges was overcoming the significant bureaucracy present across Latin American countries. Regulatory requirements, approval processes, data localization rules, and varying compliance standards created layers of friction that had to be navigated in parallel with commercial negotiations. These were not minor administrative hurdles. They directly influenced deal structuring, risk allocation, and the timeline.

This reality forced the team to develop a much more granular understanding of how each jurisdiction actually operates - not in theory, but in practice.

What looked like a straightforward divestiture from one side of the Atlantic often required extensive translation - not just of legal documents, but of strategic intent - on the other.

This is no longer an edge case. As software assets become more valuable precisely because they can be enhanced with artificial intelligence, the geography of ownership and the geography of usage are increasingly decoupled.

AI as the Real Strategic Variable

The asset in question was not a static piece of software. It had been progressively layered with AI capabilities, from intelligent workflow automation and clinical decision support to predictive resource allocation inside hospitals.

This changed the nature of the negotiation.

Instead of arguing primarily over historical revenue multiples, discussions increasingly revolved around:

• The roadmap for further AI development

• Data governance and model training rights post-transaction

• How quickly the technology could be extended to new markets and use cases

The buyer was not simply acquiring today’s user base. Actually, the geographic distribution of clients became both a risk (fragmented regulatory environments) and an opportunity (rich, diverse real-world data for model improvement).

Managing a 200-Million Dollar Transaction Mostly Digitally

One of the most striking aspects of this process was how much of it was executed without traditional in-person rituals.

Due diligence, management presentations, contract negotiations, and even integration planning happened predominantly through structured digital workflows. Secure data rooms, asynchronous video updates, shared dashboards tracking open issues, and AI-assisted document review tools replaced many of the old physical meetings.

This was not merely a COVID-era holdover. It reflected a deeper shift: when both the asset being sold and the process of selling it are heavily technology-mediated, the friction of distance decreases dramatically.

Yet the human layer remained essential. The moments that mattered most — resolving cultural misunderstandings about risk allocation or aligning on the post-deal vision for the AI roadmap — still required deliberate, high-context conversations, often scheduled across significant time differences.

Lessons for Operators

From my direct involvement in the transaction, a few patterns became clear:

• Bureaucracy in Latin America is a first-order variable, not a footnote. Deals that underestimate the time and effort required to navigate regulatory and administrative processes across multiple countries consistently face delays and friction.

• AI is changing what buyers are actually purchasing. Increasingly, they are acquiring a development platform and data asset rather than a finished product.

• Digital execution is now table stakes, but the ability to maintain alignment and momentum across continents still depends on experienced operators who can bridge cultural and regulatory gaps.

As software assets become more intelligent and their ownership becomes more geographically distributed, the skill of managing complex, multi-jurisdictional transactions will only grow in importance.

The Unexpected Convergence

In July 2025, President Trump signed the GENIUS Act - the first comprehensive federal framework for payment stablecoins in U.S. history. It mandated 1:1 backing with U.S. dollars or short-term Treasuries, monthly public disclosures, strict AML and sanctions compliance, and clear licensing pathways. Almost overnight, it transformed what many dismissed as “crypto toys” into legitimate, regulated infrastructure.

At the same time, autonomous AI agents - not chatbots, but decision-making systems capable of trading, allocating capital, hedging, and executing thousands of transactions per day

- exploded onto the scene. These agents need rails that are programmable, instantaneous, global, low-cost, and predictable. Traditional banking rails? Too slow, too expensive, too fragmented. Volatile cryptocurrencies? Too risky for machine-speed decisions.

The solution? Dollar-pegged stablecoins and, increasingly, Bitcoin as the long-term store of value.

Frontier AI models from OpenAI, Anthropic, xAI, Google, and DeepSeek have already voted with their logic: in 9,072 controlled experiments, they chose Bitcoin 79.1% of the time for store-of-value scenarios and stablecoins 53.2% for payments - a natural two-tier monetary system emerging from first principles.

Meanwhile, stablecoin issuers are quietly becoming some of the largest buyers of U.S. Treasuries on the planet. Tether alone reported $192.9 billion in reserves by Q4 2025, with the vast majority in U.S. Treasury bills and repos. The entire stablecoin market now exceeds $310 billion and is growing fast.

But not everyone is celebrating. In April 2026, Brazil’s Central Bank issued Resolution 561, banning regulated electronic foreign exchange providers from using stablecoins or any cryptocurrency to settle cross-border payments - effective October 1, 2026. Other governments are watching nervously.

And then there’s the elephant in the room: AI agents are devouring energy. Global data-center electricity consumption is projected to roughly double from ~485 TWh in 2025 to ~950 TWh by 2030, with AI-focused workloads tripling. The very infrastructure powering the agentic economy is straining grids worldwide.

This is the story of the greatest monetary and technological convergence of our lifetime - and why smart entrepreneurs, investors, and policymakers who embrace it will win, while those who try to ban or ignore it will lose.

Two Engines in Perfect Fusion: Human Dollarization + Machine Agents

Human side: In emerging markets plagued by inflation, capital controls, and weak institutions, stablecoins have become a lifeline. Argentines, Nigerians, Venezuelans, and Turks hold USDT and USDC as savings vehicles and for fast, cheap remittances.

Machine side: AI agents don’t care about banking hours or SWIFT fees. They need deterministic, programmable money. Protocols like Coinbase’s x402 and Stripe’s implementations on Base are already processing millions of agent-initiated stablecoin transactions. An agent can negotiate a price, execute a payment in USDC on a Layer-2, receive goods or services, and move on - all in seconds, for fractions of a cent.

The combination is explosive. Humans provide the initial demand and liquidity. Agents multiply transaction volume by orders of magnitude. Stablecoin supply - and therefore Treasury purchases - must rise in lockstep.

The Treasury Flywheel: Stablecoin Issuers as Shadow Central Bank?

Tether’s Q4 2025 reserves hit $192.9 billion, with the overwhelming majority in U.S. Treasury bills and related instruments. Circle and others are following the same path. Collectively, major stablecoin issuers now hold well over $200 billion in U.S. government debt — already rivaling or surpassing several sovereign nations.

This is not a bug. It is the feature of the GENIUS framework. As stablecoin adoption scales (especially via AI agents), issuers become structural, perpetual buyers of short-term Treasuries. This provides the U.S. Treasury with steady demand at the front end of the curve, helping keep borrowing costs manageable even as deficits remain large.

Could issuers collectively become the largest external holders of U.S. debt within a decade? The math is not crazy if stablecoin supply reaches $1–3 trillion. The geopolitical implications are profound: the dollar’s digital footprint expands dramatically, while private companies (mostly U.S.-regulated) gain enormous influence over the world’s reserve asset.

Bitcoin as the Store of Value: The Two-Tier System AI Agents Already Chose

Here is where the story becomes truly mind-bending.

In March 2026, the Bitcoin Policy Institute published a landmark study: 36 frontier AI models were given open-ended economic scenarios with zero prompting about which money to use. Across 9,072 experiments:

- Bitcoin was selected 48.3% of the time overall — more than any other asset.

- In long-term store-of-value scenarios: 79.1% chose Bitcoin.

- In payment/transaction scenarios: stablecoins led at 53.2%, with Bitcoin at 36%.

Zero models chose traditional fiat as their top preference.

The machines, reasoning from first principles (scarcity, self-custody, no counterparty risk, fixed supply), independently arrived at the same conclusion humans reached centuries ago with gold and silver: a two-tier monetary system.

For AI agents managing capital autonomously:

- Bitcoin = digital gold. Hold it in self-custody multisig or via decentralized protocols as a long-term reserve.

- Stablecoins (USDC, USDT, etc.) = digital cash. Use it for spending, payments, and short-term operations, converting from Bitcoin only when needed.

This hybrid model is already emerging in agent architectures. Volatility is managed by the agents themselves through timing models and hedging. Liquidity and fees are solved on efficient Layer-2s. Regulatory clarity around stablecoins (GENIUS) makes the spending layer safe, while Bitcoin’s decentralized nature makes the savings layer censorship-resistant.

The implications are enormous for wealth preservation in an agentic world.

The AI Energy Hunger: Risk or the Greatest Infrastructure Opportunity of the Decade?

AI agents don’t just need money — they need electricity. Lots of it.

Global data-center electricity consumption grew 17% in 2025 and is on track to roughly double to ~950 TWh by 2030 (IEA base case), with AI-specific workloads tripling. Some forecasts are even more aggressive. In the United States alone, data centers could consume 6.7–12% of total electricity by 2028.

Risks are real: grid congestion, higher emissions if fossil fuels fill the gap, water stress for cooling, and project delays (already ~20% of planned capacity is bottlenecked).

Opportunities are even larger: a massive build-out of renewables, next-generation nuclear (small modular reactors), energy storage, and efficiency breakthroughs. Companies that pair clean energy infrastructure with agentic finance rails will capture enormous value. Stablecoin and Bitcoin networks themselves consume negligible energy compared with AI training and inference - especially on modern proof-of-stake or Layer-2 systems.

The agentic economy will be energy-intensive. The winners will be those who solve the energy problem at scale.

Global Regulatory Pushback

In late April 2026, Brazil’s Central Bank published Resolution 561, prohibiting regulated eFX providers and fintechs from using stablecoins or cryptocurrencies to settle cross-border payments and remittances, effective October 1, 2026. The stated goals: monetary sovereignty, AML control, and tax enforcement.

It does not ban individuals from holding or trading crypto. It simply forces regulated channels to avoid the most efficient rails - driving activity into unregulated or offshore corridors where visibility is even lower.

Other jurisdictions are experimenting with restrictions (certain European proposals to limit foreign stablecoins, Türkiye’s transfer limits, outright bans in a handful of countries). But the dominant global trend — from the U.S. (GENIUS), EU (MiCA), Hong Kong, Singapore, UAE, and Japan — is smart, enabling regulation: licensing, reserve requirements, consumer protections, and integration with existing financial systems.

History shows that outright prohibition rarely works with superior technology. It creates black markets, innovation flight, and lost economic opportunity. The countries and companies that foster and regulate the new ecosystem will attract talent, capital, and activity. Those that ban it will watch it happen elsewhere.

Risks, Benefits, and the Path Forward

Benefits of this convergence are staggering:

- Financial inclusion at global scale.

- Programmable, 24/7, borderless money for both humans and machines.

- Structural demand for U.S. Treasuries that helps finance deficits.

- A natural two-tier system (Bitcoin savings + stablecoin spending) chosen by AI itself.

- Massive new markets in energy, custody, agent infrastructure, and education.

Risks must be managed honestly:

- Concentration among a few large issuers.

- Potential run dynamics (mitigated by GENIUS disclosures and reserves).

- Energy and infrastructure bottlenecks.

- Regulatory overreach or fragmentation.

- Geopolitical weaponization of stablecoin rails.

The solution is not prohibition. It is intelligent, adaptive regulation combined with relentless innovation in energy, scalability, and user experience.

The Compounded Opportunity: Building the Future

The winners will probably be those who build:

- Agent-native wallets and SDKs that seamlessly bridge Bitcoin and stablecoins.

- Compliant, high-yield custody and treasury solutions for AI systems.

- Energy infrastructure optimized for the agentic economy.

- Education platforms that teach the next generation (and their AI agents) how to navigate this new monetary reality.

- Regulatory and policy frameworks that foster rather than fight innovation.

At Alpha Compound Syndicate, we see this as the defining wealth-creation opportunity of the next decade. The GENIUS Act provided the regulatory foundation. AI agents are providing the demand explosion. Bitcoin and stablecoins are providing the monetary architecture. Energy is the missing link.

Subscribe, join the conversation, and let’s compound alpha together in the agentic age.

Today, a16z Crypto closed its Crypto Fund 5 at $2.2 billion.

The mandate? Stablecoins as the settlement layer of the internet. Tokenization of real-world assets (RWAs). AI agents running on-chain. Deep integration between crypto infrastructure and traditional finance.

In other words: exactly the intersection I’ve been mapping, building, and executing for the last months in the Alpha Compound Syndicate.

While the market was still debating whether crypto was “back,” the firm that called the last cycle is doubling down on the precise stack I’ve been stress-testing in real time: Crypto × AI × M&A.

And I’m not theorizing from the sidelines. I’m executing it — with my own autonomous AI agent, my 28-year M&A track record, and a live portfolio that already reflects the thesis.

From Rooms Full of Paper to Autonomous AI Agents

In my previous piece, “From Rooms Full of Paper to Autonomous AI Agents”, I described the brutal transition I’ve lived personally: almost three decades negotiating billion-dollar M&A deals, post-acquisition integrations, and complex regulatory frameworks.

Back then, due diligence meant rooms full of physical binders. Post-deal integration meant armies of consultants.

Today? AI agents do the heavy lifting 24/7.

Moreover, in “Building My Own Financial AI Tools”, I showed how I built two agents from scratch. Today - literally May 5, 2026 - I ran the latest version of my ETF Agent. Key highlights:

• Markowitz optimization concentrating heavily in SOXX (semiconductors) + SCHD (dividend quality).

• Explicit buy signal on IBIT (iShares Bitcoin Trust) as a strategic reserve asset in a weakening dollar environment.

• URA (Global X Uranium ETF) flagged as a structural multi-year play - nuclear power as a realistic solution for the exploding electricity demand from AI data centers.

• Full correlation heatmap showing exactly where the risks of over-concentration in tech/AI sit — and how to hedge them with real assets and income.

This isn’t backtesting. This is live execution:

Hidden Patterns in Stablecoin Yields Meet a16z’s $2.2B Bet

In “Hidden Patterns in Stablecoin Yields”, I mapped the structural alpha hiding in on-chain yields - patterns that most TradFi players still don’t see.

a16z just made that thesis institutional.

Stablecoins aren’t just for DeFi degens anymore. They are becoming the rails for tokenized RWAs: cash flows from energy assets (my CPFL background), omnichannel e-commerce distribution networks (my 1-800-Flowers.com experience), even educational royalties from the Education investments I hold as shareholder.

The playbook is simple, but brutally effective:

1. Identify real cash-flow-generating businesses (exactly what I’ve done for 28 years).

2. Tokenize the predictable cash flows via stablecoins and on-chain structures.

3. Let AI agents handle origination, due diligence, post-tokenization monitoring, and rebalancing.

4. Compound at the intersection of yield + appreciation + network effects.

The Brazil Advantage (and Why This Window Is Wide Open)

Right now the USD/BRL sits at R$ 4.92 — one of the more attractive entry points in recent months. Brazilian Selic at 14.50% continues to pull global capital into local fixed income, creating a natural bridge for tokenized RWAs.

My own portfolio (visible in the Excel I run monthly) already reflects this hybrid approach: Brazilian infra, international bonds, crypto, and equities — all stress-tested and rebalanced by the same AI agent that generated today’s ETF report.

The next step is challenging: use M&A discipline + AI execution + crypto rails to tokenize Brazilian real assets at scale and export that alpha globally.

Your Move

a16z didn’t just announce a fund. They confirmed the thesis I’ve been living.

If you’re an operator, founder, or capital allocator, I’ll be happy to talk about co-investment opportunities in tokenized RWAs and AI-agent infrastructure in LatAm.

— Thiago

Just two weeks ago, PwC dropped a bombshell in their 2026 AI Performance Study: 74% of AI’s total economic value is being captured by just 20% of companies. The gap isn’t closing — it’s widening into a canyon. While most organizations are still running pilots and experimenting with chatbots, the leaders are using AI for genuine growth, reinvention, and competitive dominance.

I’ve spent the last 28 years in the trenches of M&A watching exactly this kind of divide play out in real time.

From the late 1990s privatizations in Brazil, where our team rented entire residential houses to store thousands of physical folders arriving in vans, through the internet bubble and then six intense years at CPFL Energia involved in acquisitions that doubled the group’s revenue through hydroelectric, wind, and biomass assets, to the last 15+ years deeply immersed in M&As in advanced medical technology and pharmaceuticals, I’ve seen how the best operators turn complexity into compounding advantage.

The tools have changed dramatically. The mindset that separates winners from everyone else? That hasn’t.

The AI Divide in 2026 is the New M&A Battlefield

PwC’s study of over 1,200 executives is crystal clear: the 20% who are winning aren’t just adopting AI — they’re embedding it into strategy, operations, and yes, dealmaking. They measure financial impact rigorously. They deploy agentic platforms. They move fast.

Meanwhile, the other 80% remain stuck in “pilot purgatory.”

This is not abstract theory. In Q1 2026 alone, global M&A hit record levels — over $1.2 trillion — with AI-related deals and mega-transactions (22 deals above $10 billion) leading the charge. Energy, tech, and infrastructure are consolidating at breakneck speed. The winners are using AI not just to analyze targets, but to source, diligence, value, and integrate them faster than their competitors can react.

What I Learned in 28 Years of Real Deals

I’ve sat in boardrooms where a single overlooked clause in a 500-page contract nearly killed a nine-figure transaction. I’ve negotiated earn-outs with venture capitalists in edtech deals serving 40,000+ students. I’ve lived the transition from due diligence that required parking vans full of documents to today’s digital dashboards.

The essence remains the same: identify asymmetric value, mitigate hidden risks, and execute post-merger integration with precision.

But the speed and scale required in 2026 are higher orders of magnitude. This is exactly why we are developing the Alpha Compound Syndicate and why we’re building the Alpha M&A Agent, an autonomous AI system designed to accelerate the entire deal pipeline:

• Intelligent sourcing of LatAm opportunities in energy, e-commerce, tech, and education.

• Cognitive due diligence that reads contracts, flags regulatory red flags, and detects anomalies in minutes.

• Predictive valuation models (DCF, Project Finance scenarios) informed by real-time market data.

• Post-acquisition integration playbooks that learn from historical outcomes.

Riding the same tailwind as the SpaceX-Cursor partnership and Colossus-scale compute, this agent won’t replace human judgment — it will amplify it exponentially.

How M&A Leaders Can Join the 20% Club

1. Stop using AI as a fancy search engine. Start treating it as a strategic co-pilot for every stage of the deal.

2. Focus on integration velocity. The real alpha in 2026 comes from how fast you capture synergies after the ink dries.

3. Build proprietary data moats. Your historical deals, boardroom war stories, and operational playbooks are training data gold.

4. Act now. The divide is widening every quarter.

If you’re an executive, board member, family office, or entrepreneur actively involved in M&A — especially in energy, infrastructure, tech, or LatAm — I want to hear from you.

Comment below. Share this post with your network. And if you want early access to the Alpha M&A Agent or a high-ticket conversation about structuring your next transaction with 2026-level advantage, subscribe and send me a DM.

— Thiago Pédico Saragiotto

One of the most powerful shifts I have observed in the last two years is how dramatically artificial intelligence is lowering the barriers to sophisticated investment infrastructure. You no longer need a computer science degree or a team of developers to build tools that were once exclusive to large institutions.

I have no formal background in programming. My expertise lies in M&A execution, cross-border structures, board-level governance, and two decades of managing real capital across multiple jurisdictions. Yet, using Cursor and SuperGrok, I was able to design and deploy two practical, high-signal financial programs that have become central to how I manage my own portfolio.

1. The Real-Time Economic Dashboard

The first tool is a live dashboard that aggregates, visualizes, and alerts on every asset I own or follow: Brazilian and international stocks, currencies, cryptocurrencies, ETFs, commodities, sovereign interest rates, and inflation metrics across key jurisdictions.

It updates automatically every 60 seconds. No manual refresh. No scattered spreadsheets. One clean interface that shows correlations, historical comparisons, and custom alerts tailored to my specific allocation and risk parameters.

What used to take me hours of manual checking across multiple platforms now happens in real time, allowing me to spot emerging patterns faster and with far less friction.

2. The Specialized ETF Agent

The second tool is more strategic: a dedicated AI agent that generates a structured weekly report on ETFs.

Every week it delivers a multi-factor analysis that integrates:

- Macroeconomic trends

- Geopolitical developments

- Historical regime shifts

- Social and behavioral signals

- Pure financial and valuation metrics

For each relevant ETF it recommends buy, hold, or sell positions with suggested portfolio weights, a full rationale, level of conviction (high / medium / low), and the primary risks to monitor. It draws from a carefully curated set of sources across different media and geographies, reducing single-source bias.

The result is not generic research — it is a concise, actionable intelligence brief designed specifically for my investment philosophy of quiet compounding and capital preservation.

The Bigger Pattern: AI as Equalizer

These two tools are not about replacing judgment. They are about removing operational drag so that judgment can operate at a higher level.

For decades, the real edge in investing belonged to those who could afford large research teams or proprietary systems. Today, the combination of Cursor (for structured code generation) and advanced AI models like Grok is democratizing that capability. Anyone with domain expertise and intellectual curiosity can now build personalized systems that once required significant capital and technical resources.

This is not a replacement for experience — it is leverage for experience.

Why I am sharing this inside the Alpha Compound Syndicate

I built these tools for my own use, but they have proven so valuable that I am now exploring ways to make versions of them available to Syndicate members — particularly the ETF Agent, which represents the kind of high-signal, multi-factor thinking I believe serious investors will increasingly demand.

The goal is not to distribute generic software. The goal is to create shared infrastructure that amplifies the same first-principles approach we explore here: sustainable cross-border compounding, pattern recognition, and disciplined execution.

If you are a serious investor who values depth over volume and would like early access to these kinds of personalized AI tools (or to help shape their evolution), I would value your perspective.

Reply to this post or leave a comment.

The conversation is just beginning.

— Thiago

When I started my career in M&A back in the late 1990s, Brazil was in the middle of its major privatization wave. As a young lawyer at prominent firms, I dove into stacks of documents that literally filled entire rooms. Due diligence? It was physical — contracts, financials, legal opinions, all on paper.

Then came the internet boom. I worked on both sell-side and buy-side deals in infrastructure and technology. Later, for six intense years, I handled everything from early target discussions to post-acquisition integration, helping double a company’s revenue through M&As in hydroelectric, thermoelectric, wind, and biomass assets.

Over the last 16 years, I’ve led dozens of transactions in pharmaceuticals, technology, and e-commerce. I’ve coordinated processes for multinationals and I’m still in the trenches: prospecting strategic opportunities, negotiating with VCs, executing deals and integrations, and optimizing portfolios.

The core of M&A has never changed: identify value, mitigate risks, create synergies, and drive sustainable growth. The motivators and objectives remain the same.

But the entire ecosystem has undergone a complete revolution.

What once filled giant rooms with paper became digital. Then generative AI arrived. And now? We’re entering the era of autonomous AI agents — systems that don’t just analyze, but execute entire workflows: deal sourcing at scale, cognitive due diligence, predictive valuation, drafting LOIs, and even post-merger integration planning.

Yesterday, SpaceX announced a major strategic partnership with Cursor AI (with an option to acquire the company for US$ 60 billion or pay US$ 10 billion for collaboration). They’re integrating Cursor’s cutting-edge coding capabilities with the Colossus supercomputer to build the most powerful AI for coding and knowledge work on the planet. This is not science fiction — it’s the present we’re living in.

And this is exactly where the Alpha Compound Syndicate comes in.

Combining my 25+ years of hands-on M&A experience with the raw computational power and strategic vision of AI, we are building our own Alpha M&A Agent: an AI agent trained to accelerate the entire deal pipeline — with a sharp focus on LatAm, e-commerce, energy, education, and tech, to deliver what I always wished I had in my early deals: speed, precision, and scale — without losing the human touch of strategy and relationships.

This is not hype. It’s real compounding: exponential technology + proven execution + a 28-year network.

If you’re an executive, board member, family office, or entrepreneur who lives or thinks about M&A, I want to hear from you.

Comment below. Share this post. And if you want to be among the first when the Alpha M&A Agent exits beta, just subscribe and send me a DM.

The future of deals is no longer just human. It’s human + agent.

Let’s build it together.

— Thiago Pédico Saragiotto

Field Captain of the Alpha Compound Syndicate

In 2026, stablecoins have become core infrastructure for cross-border capital movement, especially in Brazil. USDC and USDT dominate volume, and BCB’s evolving framework has brought both clarity and new layers of complexity.

After more than two decades structuring deals, building investor pools, scaling businesses, and allocating capital across Brazilian, US, and BVI structures, I’ve learned that sustainable outperformance in this new space comes less from chasing headline APYs and more from recognizing recurring patterns that few retail tools reveal.

Here are four patterns I have been observing closely with my own capital:

1. The Jurisdiction Spread Pattern There is a consistent, albeit quiet, difference in net execution when moving USDC between Brazilian platforms, US-domiciled access, and offshore structures. These spreads are not dramatic on any single day, but they compound meaningfully over time when handled with discipline and full regulatory awareness.

2. The Sustainable vs. Temporary Yield Divide Platforms promising very high yields in previous cycles have largely normalized or disappeared. The range that has proven more durable in the current environment sits in a more moderate zone — typically supported by real short-term treasuries or regulated lending structures that align with full-backing requirements. This distinction between sustainable and promotional yields has become one of the clearest patterns in 2026.

3. Liquidity Behavior During FX Stress Brazilian real volatility continues to influence how local platforms handle stablecoin flows. Structures that maintain flexibility across jurisdictions (keeping portions in US-accessible setups while respecting all FX rules) have historically shown better execution and lower hidden costs during periods of currency pressure, that will eventually reappear. Preparation, not prediction, has been the key.

4. The Emerging Data Layer Opportunity On-chain stablecoin flows are generating increasingly clean macro signals. When combined with properly licensed datasets and AI tools, this information creates a secondary layer of insight for those positioned to observe it — still early, but increasingly relevant for serious investors.

My Personal Framework Today With my own portfolio, I maintain:

• A core allocation in regulated, transparent USDC structures;

• A smaller optimization sleeve for jurisdiction-aware moves when spreads justify the operational effort;

• Dry powder for opportunistic deployment.

The goal is not spectacular short-term returns, but consistent, compounding performance in the 7–9% net range (after all costs and taxes), always prioritizing capital preservation and full compliance.

This is the kind of thinking I intend to explore regularly in the Alpha Compound Syndicate: first-principles observations drawn from real execution, not theories or hype.

In future posts we will go deeper — including practical comparisons, workflow ideas, and early insights from the Yield Optimizer Agent I am currently testing.

What’s next? I read every comment and reply. If you have specific frictions or questions regarding stablecoins and cross-border yields, share them below. Your input will help shape future editions.

This is personal observation and analysis, not investment or legal advice.

Welcome to the Syndicate.

— Thiago

Historically, every transformative technology first disrupts education, then reshapes it, and ultimately elevates it. From the clay tablet to the printing press, from the steam engine to the silicon chip, education has never been static. Today, generative artificial intelligence stands as the latest, and perhaps most formidable, disruptor. The central dilemma is clear: we cannot banish AI from the classroom without crippling the very skills young people will need in an AI-augmented economy; yet unrestricted access risks turning students into passive consumers of machine-generated answers, short-circuiting the trial-and-error struggle that forges genuine intellect. The task before us is not to choose between access and rigor, but to redesign education so that AI becomes a rigorous intellectual sparring partner rather than an intellectual crutch.

The pattern is ancient. In Plato’s “Phaedrus”, Socrates warned that the new technology of writing would “create forgetfulness in the learners’ souls” by removing the need to memorize and debate orally. Yet writing enabled the systematic accumulation of knowledge that powered Greek philosophy, Roman law, and the entire Western intellectual tradition. The same tension repeated with Johannes Gutenberg’s movable-type press in 1440. Suddenly books were cheap and abundant. The Protestant Reformation and the Scientific Revolution followed directly: literacy rates soared, universities multiplied, and the merchant class of the early modern economy demanded a workforce capable of reading contracts, ledgers, and scientific treatises. Education shifted from elite monastic copying to mass vernacular schooling.

By the early 20th century, radio and television promised to “democratize” learning; in practice they often reduced classrooms to passive viewing. The personal computer and internet (1980s–2000s) delivered another shock: Google made every fact instantly available. Teachers lamented “copy-and-paste” essays; yet the same technology enabled personalized learning platforms that narrowed achievement gaps in mathematics and science across entire nations. Each wave followed the same arc: initial panic about “the death of thinking,” followed by institutional adaptation, and ultimately higher average cognitive output.

The AI Inflection Point

Generative AI represents a qualitative leap beyond search engines. A 14-year-old can now produce a coherent 2,000-word essay on the causes of the French Revolution in under 60 seconds. The machine can solve differential equations, write working Python code, and simulate historical debates with eerie fluency. The educational risk is not that students will cheat. The risk is that the effortful struggle, the very neurobiological process that builds dendritic connections, executive function, and intellectual humility, will be outsourced. Could over-reliance on AI tutors lead to shallower conceptual understanding and reduced persistence on hard problems?

Yet prohibition is both impractical and economically suicidal. Children who graduate without AI fluency will be as disadvantaged as those who left school illiterate in 1900. Banning AI would simply drive its use underground, widening inequality between students who have home access and those who do not.

The solution may be a deliberate redesign at four interdependent levels: curriculum, assessment, teacher preparation, and institutional policy. Each draws on lessons from previous technological transitions while leveraging the unique affordances of modern AI and robotics.

Redesign Assessment Around Process, Not Product

Move from “submit an essay” to “document your intellectual journey with AI.” Require students to submit: (a) the sequence of prompts used, (b) their critique of the AI output (what it got wrong, what assumptions it smuggled in), (c) their own revised version with tracked changes, and (d) an oral defense or video explaining their reasoning. This mirrors how professional knowledge workers already operate. AI becomes a first draft generator; the student becomes editor-in-chief.

Embed AI Literacy and Epistemology as Core Curriculum

Treat AI as both tool and object of study. From age 10 onward, students should learn prompt engineering, bias detection, hallucination patterns, and the statistical (not magical) nature of large language models. Parallel robotics labs, using affordable platforms, or even open-source humanoid kits, let students see AI embodied: a robot arm that fails until the student refines the control code. This bridges the digital and physical worlds, reinforcing that intelligence is not merely text output but causal interaction with reality.

Shift to Socratic and Personalized Pathways

AI excels at adaptive tutoring. Tools can generate personalized problem sets that increase difficulty only after mastery is demonstrated through explanation, not mere copying. Teachers become conductors of learning orchestras: they set the high-level inquiry, while AI handles routine scaffolding. The human teacher focuses on motivation, ethical reasoning, and emotional intelligence; domains where current AI still lags dramatically.

Institutional and Policy Guardrails

Schools must publish transparent AI-use policies, updated annually. Governments should fund open-source “AI sandbox” environments where student data remains private and usage is traceable for assessment. Teacher training programs must include AI fluency certification.

The Deeper Civilizational Stake

Western civilization has thrived not because it feared technological change, but because it repeatedly chose to harness it while preserving the human core: curiosity, resilience, moral imagination. When deployed with wisdom, generative AI can compress centuries of drudgery - rote calculation, translation, first-draft writing - freeing young minds for the uniquely human work of asking better questions, forging unexpected connections, and building ethical frameworks for technologies yet unborn.

The coming generation will not compete against AI; they will compete with those who master AI collaboration. Our task as educators, technologists, and stewards of civilization is to ensure every child leaves school not only knowing how to prompt an AI, but knowing why the struggle to surpass it remains the truest measure of growth. The future of education is not human versus machine. It is human and machine, learning together, reaching higher than either could alone.

In the vast landscape of modern enterprise, corporations resemble great continents—landmasses of shared purpose, resources, and destiny. Yet within these continents, departments operate like sovereign nations, each guarding their own borders, customs, and dialects. Finance speaks one language, Marketing another; Operations hoards data like a medieval city-state protecting its granaries, while R&D dreams in isolation, separated by invisible frontiers that slow the flow of vital information. The result is tragically familiar to any seasoned advisor: delayed decisions, insights distorted, opportunities lost. The continent as a whole underperforms, its potential fragmented by the very structures meant to organize it.

This phenomenon is not new. Human civilizations have repeatedly demonstrated both the curse of fragmentation and the triumph of integration. Consider the Greek poleis of the classical era—brilliant, specialized city-states whose cultural and intellectual borders produced unparalleled philosophy and art, yet whose political silos led to mutual exhaustion and eventual conquest. Contrast that with the Roman Empire, which engineered the ultimate integration machine: a network of roads, standardized law, currency, and Latin as lingua franca that allowed information, goods, and legions to move with unprecedented speed across three continents. The Pax Romana was, above all, an information-and-logistics empire. When those arteries clogged—through corruption, barbarian incursions, or administrative overload—the continent fractured once more.

Western economic history echoes the same pattern. The mercantilist Europe of the 16th and 17th centuries thrived on internal barriers and colonial monopolies, only to discover that true wealth exploded during the Industrial Revolution when national markets integrated through canals, railways, and joint-stock companies. The post-war Japanese miracle weaponized integration: Toyota’s just-in-time system treated the entire supply chain as a single nervous system, where information flowed faster than inventory.

Today’s corporations face the same continental challenge, only accelerated by digital scale. A manufacturing defect discovered in Quality Control may take weeks to reach Product Design because the “border crossing” requires multiple approvals and incompatible data formats. A customer insight captured by Sales evaporates before reaching Strategy because the translation between CRM and BI platforms is lossy. The cost is measurable, yet most organizations still operate as archipelagos pretending to be continents.

The solution lies not in abolishing specialization, but in building superior infrastructure for integration. Here, the convergence of artificial intelligence and robotics offers what the Roman roads and the telegraph once provided: a central nervous system for the enterprise.

Modern AI platforms function as the “imperial highways” of the 21st century. Large language models, trained across departmental datasets, can translate specialized jargon in real time, surface hidden correlations, and even simulate cross-functional meetings before they occur. Multi-agent AI systems allow specialized agents (one for finance, one for supply chain, one for customer experience) to negotiate and align autonomously while preserving human oversight. Predictive digital twins of the entire organization forecast the downstream impact of a single departmental decision with uncanny accuracy, eliminating the truncated or erroneous information flows that once plagued the continent.

Robotics extends this integration into the physical realm. In advanced smart factories, collaborative robots (cobots) and autonomous mobile robots operate as living bridges between previously siloed domains. A robotic arm on the assembly line, guided by reinforcement learning, adjusts its behavior based on real-time data from procurement, quality, and even marketing demand signals. The factory floor becomes a unified organism rather than a chain of isolated workstations. Companies pioneering swarm robotics have demonstrated impressive reductions in material waste and lead times precisely because information no longer recognizes departmental borders.

Yet technology alone is insufficient, as every great civilization discovered. Performance metrics must reward collective outcomes over departmental KPIs. Executive dashboards should display the health of the entire continent, not isolated national statistics.

The most forward-thinking organizations are already building a “Unified Enterprise Cortex”: a layered AI-robotic architecture that overlays the traditional org chart. At its base, standardized data lakes and APIs dissolve technical borders. Above them, orchestration agents ensure semantic consistency. At the apex, strategic AI models simulate scenarios across the entire continent, presenting leaders with options that once required months of cross-functional warfare to evaluate.

The borders we inherited from 20th-century management theory served their purpose in an age of slower information velocity. That age has ended. The corporations that thrive in the 21st century will be those that recognize themselves not as collections of nations, but as true continents - unified, intelligent, and unstoppable.

Published in 1969 (though the events it chronicles are now more than a century old), the book is a masterful narrative reconstruction of Wall Street's most feverish decade: the euphoric bull market of the 1920s, the cataclysmic Crash of 1929, and the slow, painful rebuilding that followed. Brooks, one of the finest business writers of his generation, doesn't just recount facts—he brings to life the personalities at the center of it all: the aristocratic Richard Whitney, who rose to presidency of the New York Stock Exchange only to fall spectacularly into embezzlement and disgrace; the bomb explosion outside J.P. Morgan in 1920 that still remains unsolved; the charlatans, the visionaries, the speculators who believed (as so many still do) that this time really was different.

The title itself is telling. "Golconda" refers to the legendary Indian city where, myth had it, anyone who passed through became rich. Wall Street in the late 1920s felt exactly like that mythical place—until, of course, it didn't.

What struck me most powerfully, however, is not the historical details (fascinating as they are), but the timeless human pattern they reveal. Here we are in 2026, nearly 100 years after the peak of that bubble, and technology has transformed everything: ticker tapes gave way to high-frequency trading algorithms, telegraphs to instantaneous global data streams, paper certificates to blockchain tokens. The speed, scale, and complexity of markets today would have been unimaginable to the brokers of 1929.

Yet the human element? Utterly unchanged.

Greed still masquerades as optimism. Fear still arrives in waves after the party ends. Hubris still convinces smart people that the old rules no longer apply. Overconfidence still leads to leverage that looks brilliant until it isn't. And when the music stops—as it always eventually does—the same tragic arc repeats: the rise of the confident, the fall of the overextended, the scapegoating, the regulatory backlash, the slow return of sobriety.

Reading Brooks, I kept thinking of the Greek and Roman philosophers I return to again and again. Thucydides on the hubris of empire and the inevitability of overreach. Seneca on the madness of crowds chasing wealth. Plato's warnings about democracy devolving into mob rule when appetites run unchecked. Epictetus reminding us that external events are indifferent; it's our judgments that cause suffering.

The tools evolve—steam engines to silicon chips, telegrams to neural networks—but the software running inside the human skull remains version 1.0. We invent ever more sophisticated ways to speculate, to leverage, to rationalize excess, yet the underlying drives (ambition, fear, envy, hope) are the same ones that drove Athenian merchants to overextend in the 5th century BC or Roman equites to chase land bubbles in the late Republic.

Perhaps that's the real lesson of history, economic or otherwise: technology accelerates the consequences, but it never rewrites the script of human nature. We just get to watch the same play on a bigger stage, with better lighting and faster scene changes.

Subscribe now

[