Late-Cycle Investment Theory: Winter 2025–2026 in 10 Snapshots
From stagflation signals to the electrostate, ten snapshots of a world order under strain
Late-cycle investment theory holds that we are living through the 1970s of tech. The age of semiconductors, computing, and networks is reaching maturity, and the symptoms are now harder to ignore: slowing growth, sticky inflation, institutional breakdown, and a concentration of capital in the hands of a shrinking number of incumbents.
When I introduced this framework in June last year, it was still possible to treat it as a hypothesis. That is less true today. The data arriving from Q4 2025 make the case on their own terms. US GDP growth came in well below expectations. Inflation is running above target. Consumer spending is softening. The stagflation I flagged in August is no longer a prediction—it is now a reading on the instrument panel.
But the framework is not only about what is breaking down. It is also about what is being built in the gaps. The 1970s were not just a decade of crisis; they were the decade in which lean production, financial innovation, and the early personal computer laid the ground for everything that followed. Something similar is happening now, though the geography of that construction has shifted in ways that most Western analysis is still not equipped to see.
China, in particular, demands a different kind of attention. Standard economic comparisons—growth rates, debt ratios, trade balances—typically apply between economies of a certain scale. China, however, has long exceeded that scale, and the frameworks have not kept up. What is taking shape there, in electricity, batteries, robotics, and the industrial base that underpins them, looks less like a conventional emerging market story and more like the early US viewed from mid-19th-century Europe: a mass so large that it is generating its own rules.
Meanwhile, the financial system is undergoing its own restructuring. Tokenisation is moving from experiment to infrastructure. The boundary between venture capital and credit is blurring as the economy shifts back toward hard assets. And the monetary order is fracturing along a clear fault line: dollar-denominated digital networks on one side, physical gold and parallel settlement systems on the other.
As with the previous progress report, I have organised this edition around ten snapshots. Some are data points, some are structural arguments, and some are uncomfortable signals—including one that Karl Polanyi would have recognised immediately. I welcome your feedback.
1/ Economic data points to emerging stagflation in the US
The latest economic data from the US suggest that the stagflationary pressures I predicted in August last year are beginning to manifest. US Q4 2025 GDP growth came in at a disappointing 1.4%, well below the 3% expectation, with final sales at 1.2% versus 2.6%. Consumer spending remains modest at 2.4%, signalling that households are already feeling the squeeze.
Meanwhile, inflation remains elevated: the official PCE Price Index rose 2.9% year-on-year, with core PCE hitting 3%—both above expectations. Month-over-month, December PCE and core PCE climbed 0.4%, the highest since February 2025, suggesting renewed upward pressure on prices even as broader macroeconomic growth slows—and Trump seems determined to double down on tariffs, even after the landmark decision issued by the Supreme Court two days ago.
Real-time price tracking from Truflation paints an even sharper picture. Its PCE index reads 1.55%, while core PCE is 1.92%, highlighting that consumers are already paying more at the point of purchase than conventional monthly statistics capture. Some note that these figures precede anticipated price shocks in the second half of 2026, particularly as supply-chain disruptions—such as the memory chip shortage—propagate through consumer electronics and other sectors.
Taken together, these developments indicate the emergence of the stagflation I warned of: growth faltering while prices remain sticky. Unlike the 1970s, however, wage growth remains weak, offering little cushion for households, while energy and technology costs continue to rise. If the current trajectory persists, we may soon witness broad-based inflation without compensating wage gains, confirming that our entering the 1970s of tech is moving from theory toward reality.
2/ China is so large that it defies every analytical framework we have
Most Western analysis of China still looks at growth rates, debt ratios, trade balances. And when it comes to China, the numbers are impressive: a trade surplus heading toward $1.2 trillion in 2025, manufacturing output at nearly a third of the global total, 160 cities with over a million inhabitants. But these tools were built to compare economies of a certain scale, and China has long since exceeded that scale. It should make us wonder: how should we analyse China?
Adam Tooze published an essay last year that made me realise how much China stands in its own category. Look at the coal production curve across all of recorded human history, he argues, and you see three phases: the pre-industrial world, the industrial age from 1750 to 2000, and then something else entirely—a vertical break, within living memory… driven entirely by China’s urbanisation and industrialisation! Adam’s conclusion is as follows: China’s growth is not an instance of a general rule, nor an exception to one. It is, as he puts it, “the ball game”.
Balaji S. Srinivasan recently drew similar conclusions on X: “China is the exception to every rule.” At a billion-person civilisational scale, it can sustain nationalism, near-autarky, and centralised control in ways no ordinary nation-state can. As a result, standard categories do not apply.
The right comparison, to me, is the US viewed from Europe in the mid-19th century. I wonder how European elites looked at America back then. They probably kept measuring America against frameworks built for older, smaller states, and missed the paradigm shift that sheer mass was making inevitable—that is, America taking the lead in the age of electricity, steel and heavy engineering, then completely reshaping the world in its image in the subsequent age of oil, automobiles, and mass production.
The same mistake is being made about China today. And I believe the late-cycle framework, with its emphasis on manufacturing depth and the seeds of the next technological revolution, is one of the few lenses large enough to take China seriously on its own terms.
3/ China is turning electricity into the foundation of the next technological revolution
As the current age of semiconductors, computing and networks enters its late-cycle maturity phase, I believe China—just like the US in the late 19th century—is laying the groundwork for the next technological revolution, and that revolution will lead us into the age of electrification.
Historically, every “great surge of development” revolves around a new, pervasive, low-cost input. Today, that input is electricity—increasingly generated by renewables—and its emblematic products are electric vehicles, batteries, drones, and robotics.
Three analytical frameworks help make sense of this shift:
First, Carlota Perez’s model of techno-economic paradigms reveals how the seeds of a new revolution are planted during the exhaustion of the old. While the West fixates on AI—which to me is an efficiency innovation extending the current computing paradigm—China is leveraging massive scale to trigger the “big bang” of the electrification era, transitioning from a fossil-dependent economy to the world’s first true “electrostate”.
Second, the concept of “programmable electricity” helps us understand how the new grid will function. In the age of electrification, energy is no longer a static commodity; it is a software-defined network where generation, storage, and consumption are dynamically orchestrated by the most advanced version of software—that is, AI.
Finally, Sam D’Amico and Noah Smith’s “Electric Tech Stack”—comprising lithium-ion batteries, electric motors, power electronics, and semiconductors—provides a lens for understanding the industrial base in the age of electrification. This stack serves as a universal, modular template for modern hardware, emerging as the modern equivalent of the American system of manufacturing. Also see a variant: the “Electro-industrial stack”.
Through these frameworks, it becomes clear that future geopolitical power belongs to those who master the physical generation and intelligent distribution of electricity.
4/ The AI bubble is real, but it is the wrong kind of bubble
A recent essay by Speedinvest’s Sameer Singh deserves attention from anyone following the late-cycle framework. In it, Sameer distinguishes between mid-cycle bubbles, which leave behind useful infrastructure, and late-cycle bubbles, which trigger overconsumption of the input that powered the whole surge. This, like my own work, maps onto Carlota Perez’s model.