Landmark Finance · Geopolitics · AI SK10 min read

The Last Window

Five stories everyone is watching separately. One transition no one is naming. And a window that closes in 18 months.

RESEARCH LAB THE LAST WINDOW · 2026

Gold ran from $2,600 to above $5,000 in twelve months. It's correcting now. Central banks are still buying 60 tonnes a month. The CBO just called U.S. debt "not sustainable." Minneapolis is burning. Greenland is in play. AI just rewrote the GDP numbers.

Most people are watching these separately. They're not separate.

They're one transition, the kind that happens every 75 years and reorganizes everything. The last one ran from 1929 to 1945. What came out of it built the architecture we've been living inside since.

We're in the stage that precedes it. Every marker is live.

The Big Cycle, Where We Are Now
Rise Consolidation Excess Tension 5 Pre-breakdown YOU ARE HERE Disorder 1929–1945
Each cycle ~75 years · Last Stage 6: 1929–1945

The pattern is complete. But this time the pattern has a variable it's never had before.

· · ·

Every prior Stage 5 unfolded while the dominant new technology was still nascent. The steam engine, electrification, computing, all took 20 to 50 years to deploy at scale. The debt crises played out before the technology could offset them.

AI is already in the GDP numbers.

3.7% U.S. GDP growth, Q4 2025
−403K Payroll revision, same period
2.7% Productivity growth, 2025 (nearly double pre-AI baseline)

Q4 2025: GDP at 3.7%. Payrolls revised down 403,000. Productivity up 2.7%, nearly double the pre-AI baseline.

Output rising. Employment falling.

That's not a paradox. That's what a productivity revolution looks like in the data, arriving precisely as the debt crisis peaks.

For the first time in 500 years of Big Cycle history, some nations might actually grow their way out of the breakdown. Not through austerity. Not through war. Through AI-driven productivity that compresses decades of deployment into years. For those nations, the exit ramp exists. For the rest, Stage 6 plays out on schedule.

This split, who finds the exit ramp and who doesn't, is the new geopolitical fault line. It's widening faster than any prior Stage 5 could produce.

· · ·

Here's the other thing no one is saying: the institutions that were supposed to manage this transition can't.

Not because they're corrupt. Because they're obsolete.

The IMF, the WTO, the UN Security Council, NATO, all built in 1944-1948 for slow-moving information, nation-state actors with clear borders, and human-speed decision-making. AI propaganda now spreads in microseconds. Financial contagion propagates sub-second. Compute is stateless. Autonomous weapons don't need authorization chains.

They're running 2026 on 1948 architecture.

1st Order, Already Visible

The legitimacy vacuum is real and accelerating. The WTO cannot adjudicate AI trade disputes because its rules predate AI. NATO's Article 5 doesn't clearly cover AI-driven influence operations. The UN Security Council is paralyzed by veto. The Fed's tools are blunted when productive capacity is growing faster than credit can adjust.

2nd Order, 18 to 36 Months

The vacuum doesn't get filled by new international institutions, those take decades. It gets filled by platforms. When nation-state institutions fail, the entities with global reach, real-time information processing, and the ability to coordinate at scale are not governments. They're AI platforms. OpenAI's usage policies affect 200 countries. Anthropic's safety decisions shape what AI can do globally. These companies haven't chosen to become governance actors. But they are, because they're the only entities with the technical capacity to manage the complexity of 2026 at the speed 2026 requires.

3rd Order, 5 to 10 Years

A sovereignty gap opens that no legal framework addresses. OpenAI is accountable to its board, its investors, and U.S. law. Its decisions affect 200 sovereign nations. Carter's framework is optimistic that networks self-organize into new governance structures. History suggests they don't, they concentrate. The entity that controls the AI layer in a Stage 5 world is the new hegemonic power, regardless of what country it's headquartered in.

· · ·

So if the old institutions can't manage the transition, who builds the new order?

The answer is sitting inside the energy-compute-capital triangle.

The Triangle That Determines AI Leadership
AI CAPABILITY ENERGY ACCESS CAPITAL STACK AI LEADERSHIP = energy access requires compute drives investment geopolitically contested
Not algorithms. Not capital. The bottleneck is always energy.

AI needs compute. Compute needs energy. Energy is geopolitically contested. Which means the AI race is ultimately an energy race, and the U.S. has the algorithms, China has the manufacturing, and neither has enough clean, cheap power to run the AI stack at civilizational scale.

Nuclear permitting is AI policy

The DOE approvals for small modular reactors. The Ex-Im Bank's $4.2B backing for nuclear fuel chains. The bipartisan push on licensing. These are not energy policies.

They are AI capacity policies. Every megawatt approved in the next five years is a direct investment in running frontier AI cheaper than competitors. AI labs are already signing 20-year power purchase agreements with nuclear operators. The energy question is the AI question.

Gold is right for 5 years, wrong for 10

The correct long-term store of value in a world where AI determines productive capacity is not gold. It's energy infrastructure and compute access.

The next Berkshire Hathaway owns nuclear plants, GPU clusters, and the applications running on top of them.

Gold preserves value. Compute generates it.

The oil price floor is a weapon

Iran's Gulf escalation. OPEC+ discipline. Energy market volatility. These are not separate from the AI race.

Every $10 rise in oil is a direct tax on data center operating costs. The actors who benefit from elevated energy prices are precisely the actors who cannot compete on AI. Energy supply disruption is becoming a geopolitical weapon aimed specifically at AI-leading nations. It's not about oil. It's about slowing down the compute stack.

Greenland and Canada are chip moves

Greenland: rare earths critical for semiconductor manufacturing. Canada: hydroelectric capacity at scale. These aren't imperial land grabs.

Dalio reads them as classic Stage 5 great power moves. He's right about the pattern. Wrong about the object. The competition isn't over territory. It's over the inputs needed to run the next century's productive capacity.

"The dollar isn't backed by the Fed anymore. It's backed by the U.S. ability to run frontier AI at lower cost than anyone else. Countries accumulating gold are hedging the old order. Countries building nuclear plants are positioning for the new one."

· · ·
· · ·

Put it together and what you have isn't a debt crisis. Debt crises have happened 30 times in the last 500 years. They always resolve the same way: devaluation, default, renegotiation, new monetary order.

This is a sovereignty transition. The world is moving from being organized around nation-state capital to being organized around AI compute capacity. And the new order is already being built, by private actors, in 36-month cycles, without coordination, faster than any institution can respond.

The 1944 Bretton Woods settlement took years of wartime cooperation to produce. This one is happening in the background while everyone argues about tariffs.

The Divergence Within the Big Cycle
Stage 5 · Now AI-early nations productivity exit ramp AI-laggard nations Stage 6 on schedule
The Big Cycle is not universal anymore. The divergence is the story.

The winner of Stage 5 is not the nation-state that manages its debt ratio best. It's the actor (nation, company, or platform) that controls the energy-compute stack when the old monetary order breaks. Whoever owns that stack writes the next Bretton Woods.

· · ·
· · ·

What to do with this

If you have capital

Hold gold through the transition period, that part is correct. But the 10-year bet is compute infrastructure and energy access. Nuclear operators, grid modernization, semiconductor supply chains. These are underpriced because most investors are treating a structural transition as a cyclical trade.

That gap is the opportunity.

If you're building

The intelligence gap is temporary. In three to five years, everyone has access to what exists at the frontier today.

The moat isn't the technology. It's the context accumulated, the audience built, the trust earned during the transition. Every piece of intelligence published before the mainstream catches up is capital in the new order. The window is 18 to 36 months. After that, differentiation requires something models can't replicate: perspective earned through time.

For everyone else

The anxiety is not irrational. It's the predictable response to watching the frameworks that organized life, institutional trust, career identity, national narrative, dissolve without replacements ready.

The people who navigate Stage 5 best aren't the ones who predict outcomes correctly. They're the ones who build frameworks that hold across multiple scenarios. Orientation systems. Mental models. Community. Things that don't depend on any particular outcome to remain valuable.

That's the most leveraged position available right now.

· · ·

The Window

Some of this is already decided. Gold running from $2,600 to $5,000 in twelve months isn't a forecast: it already happened. The current correction doesn't change the direction. The CBO debt warning already landed. Minneapolis already happened.

Some of it resolves this summer. One AI platform decision that conflicts with a nation-state's sovereign interests. One AI-driven financial event that moves faster than any institution can respond. These don't need 18 months. They need one incident.

Some plays out over the next year. Nuclear permitting locking in who runs cheap AI for the next decade. Chip export controls tightening. Energy infrastructure deals being signed right now.

The window to position, in capital, in content, in systems, in self, is not one date. It's a cascade. And the early parts of it are already closed.

Sources

Ray Dalio, "We Are Now in Stage 5" (Fortune, March 14, 2026) · Ray Dalio, "Investing in Light of the Big Cycle" (raydalio.substack.com, Feb 24, 2026) · Ray Dalio, "Stage 5 and 6 Developments" (LinkedIn, March 2026) · Radigan Carter, "A Framework for What Comes Next" (X, March 2026, 1.1M impressions) · Erik Brynjolfsson, "AI Productivity Is Now Visible in GDP Data" (Financial Times, March 2026) · CBO Fiscal Outlook (March 2026) · J.P. Morgan Global Research 2026 Gold Outlook (target: $6,300) · Reuters, "OpenAI to nearly double workforce to 8,000 by end-2026" (March 2026) · Atlantic Council, "Eight Ways AI Will Shape Geopolitics in 2026" · FP Analytics AI-Energy-Geopolitics simulation · Research Lab Weekly Synthesis W12 (March 2026)