Cognitive Capitalism with AI Characteristics

Artificial intelligence is the latest and greatest buzzword of our technological moment (remember when Facebook changed its name to Meta because the Metaverse was the thing?) - every week brings a new model, a new benchmark, a new claim about the future of cognition and work. There is excitement; there is fear; and there is uncertainty.

One interpretation sees this moment as a technological breakthrough: machines learning to reason, software gaining new abilities, industries preparing to automate tasks once reserved for skilled workers. A different view, often voiced by skeptics, treats the boom as a speculative bubble, inflated by hype, investor optimism, and semi-religious belief destined to correct itself. Both perspectives capture part of the truth, but neither fully explains why the largest technology firms in the world are undertaking the most extensive capital expenditure program in modern history - building data centers, electricity infrastructure, chip supply chains, and model-training pipelines at a pace that rivals wartime industrial mobilization.

Understanding this requires a broader frame. AI is not emerging into an empty landscape; it is arriving at the end of a long arc of economic transformation - the era of cognitive capitalism, in which value flows through the organization of knowledge, communication, and attention rather than primarily through the fabrication of physical goods.

Crucially, knowledge work cannot be outsourced in the same way manufacturing could. A lawyer in another country can’t easily try cases in U.S. courts; a Wall Street trader can’t operate under foreign regulatory regimes. Yet this is exactly where AI becomes so appealing: it promises to “outsource” knowledge work — not to cheaper labor markets abroad, but to machines owned by U.S.-based firms.

Companies that built their power by capturing the surplus generated from knowledge work now see AI as a way to deepen their hold on the most profitable sectors of the economy. This isn’t just about replacing jobs or automating tasks — though that’s part of the story. It’s about becoming indispensable to how knowledge is produced and monetized.

Of course, knowledge work is not just accounting and lawyering; it’s also the way of war - drones, airpower and surveillance. Sometimes I think of the US as a brain in a tank.

TLDR; The AI build-out is not a break with that system of cognitive capital. It is an extension and intensification of it - and, in some ways, its logical culmination.

From Industrial Capital to Cognitive Capital

For most of the 20th century, “advanced economies,” with the US as the archetype, were defined by factories, production lines, and mass manufacturing. Steel mills, automobile plants, petrochemical complexes, and defense industries formed the backbone of economic strength and national power. But beginning in the late 1970s, this model shifted. Globalization and trade liberalization made overseas production cheaper. Deregulation and financial innovation opened new avenues for capital returns. Manufacturing capacity migrated, first gradually and then rapidly, to East Asia.

The shock of deindustrialization was real, and its effects remain visible in many regions; I remember driving through Gary, Indiana, and thinking: this place has seen better times. Yet the story is not solely one of decline. As industrial capacity moved outward, a new economic center formed around knowledge-intensive industries - finance, software, biotech, professional services, media, and later, platform-enabled digital networks. Companies like Microsoft and Apple, born at the dawn of the personal-computer era, matured into architects of global information systems. Google built the index of human knowledge; Meta scaled social connection; Amazon created the logistics and cloud backbone of digital commerce.

The internet wasn’t invented by Al Gore, but it was certainly invented in America.

These firms did not simply create products. They built infrastructures for cognition: systems that manage search, communication, coordination, and decision-making at scale. The raw material was attention, data, and human mental activity; the output was value extracted from organizing that activity more efficiently than before.

Cognitive capitalism elevated the ability to manage information flows as the central economic capability.

Thanks for reading Ranganaut! Subscribe for free to receive new posts and support my work.

Bell Labs was one of the first institutional settings to organize scientific discovery as an integrated system - fusing research, engineering, and industrial application. Its innovations in transistors, lasers, radio astronomy etc etc created a lot of cognitive capital before its time. Owned by Nokia of all fates!

The Tribulations of Scale

By the early 2020s, the largest technology companies reached a size that distorted conventional categories. Together, the “Magnificent Seven” accounted for roughly one-third of all U.S. equity market value. Their revenues rival national budgets. Their platforms mediate cultural discourse, labor markets, and political communication. Their cloud infrastructure underpinned entire industries. They are, in many ways, the central institutions of the global economy.

But extreme scale produces a structural challenge: the problem of reinvestment. When a company is valued in trillions, it cannot maintain its valuation simply by selling more advertising or devices. It must find new domains large enough to absorb capital and produce durable strategic advantage.

For much of the last decade, cloud computing served this role. But even cloud growth has natural ceilings. Artificial intelligence, understood not merely as software but as a new computational and energy infrastructure layer, offers a fresh expansion frontier. It demands data centers, chips, transmission lines, cooling systems, software pipelines, and talent on a scale that few institutions can muster.

The AI boom, in other words, is not only enthusiasm. It is the industrial policy (or should I say, cognitive policy?) of capital at scale: a reinvestment imperative masquerading as technological inevitability.

The Stack as Strategic Terrain

Once committed to AI, hyperscalers face a second question: where in the emerging AI stack should value be captured? The current landscape reveals two broad strategic patterns.

Microsoft has positioned itself as the orchestrator of a distributed ecosystem: a deep alliance with OpenAI, flexible partnerships with Nvidia and other chip suppliers, and Azure as the backbone through which models reach customers. It is a federation strategy - leverage external innovation, integrate it into cloud and enterprise channels, and reinforce network effects. Incidentally, similar to how Windows became the default in the pre-internet era, so MS might be leaning on institutional memory here.

Google has chosen a vertically integrated route: proprietary tensor-processing units, internal research through Google DeepMind, extensive model-training capacity, and a unified product stack. It aims for internal coherence and efficiency gains from controlling the chain end-to-end. Google is the Apple of AI.

These strategies differ, but the underlying logic is shared: own or coordinate the infrastructure for producing and distributing machine cognition. In the industrial age, control over steel, rubber, and assembly lines created durable advantage. In the cognitive age, that logic applies to silicon, data, compute clusters, and inference pipelines.

This is why so much current innovation - and capital spend - concentrates not at the application layer but in chip design, data-center architecture, training infrastructure, and energy access. AI is not simply a software revolution; it is a hardware and power-infrastructure revolution disguised as one.

Ford Motor Company’s early success was rooted in controlling an integrated supply chain, from materials to manufacturing to distribution. This vertical integration reduced costs and increased reliability. Today’s technology firms pursue a similar logic across compute, energy, software, and distribution channels. It is Fordism applied to cognition.

A Mirror from Elsewhere: The Chinese Production Lesson

The rationale becomes even clearer when viewed through global comparison. In the electric-vehicle sector, China has built a vertically integrated industrial ecosystem. Companies like BYD manufacture batteries, design chips, control mineral supply chains, and operate massive domestic factories — supported by long-term industrial policy and state coordination.

This strategy has already reshaped the global EV market. Western automakers, which favored outsourced production and “asset-light” strategies, now find themselves facing a competitor that treats manufacturing scale not as a burden but as a moat.

I talked about the US response in my posts about compute and energy sovereignty: not only in EVs (by doubling down on ICE tech - apropos in Trump’s America…), but also in semiconductors and AI. The CHIPS Act, IRA energy provisions, and hyperscaler infrastructure build-out all reflect an effort to restore or reconstruct sovereign industrial capability in critical domains. In one sphere - EVs - China has led materially. In another - AI - the U.S. is moving to lead cognitively and computationally.

In the 1970s, Japanese automakers stunned global markets by combining efficiency, quality, and integrated supply planning. Their success forced Western firms to rethink manufacturing and supply-chain strategy. China’s EV sector plays a similar role today -demonstrating what coordinated industrial capability can achieve.

Entering the Metabolic Phase

To make sense of this convergence, it helps to return to a concept we have explored previously: metabolism. Societies depend on flows of energy and material. Coal powered the steam era. Oil powered the automotive and aviation eras. Electricity reorganized manufacturing and urban life.

Digital networks added a new metabolic layer: information and attention. AI adds a further one: computational metabolism, i.e., the ability to convert energy and data into automated inference.

Crucially, this is not (only) abstract. Data centers require vast electricity supply, land, cooling water, specialized hardware, and highly trained labor. The cloud was once described as “weightless”; AI reveals its weight. The system that organizes cognition turns out to be deeply material.

We may be returning, in a new form, to a world where the core institutions of economic power are not only digital platforms but infrastructural builders - firms that command energy, compute, chips, and talent.

The metaphor of “the cloud” is giving way to actual geography: substations, fiber routes, transformer stations, and industrial-scale computing campuses. AI, in this sense, brings digital capitalism back into contact with the physical world.

Speculation and the Lessons of History

None of this eliminates the possibility of a bubble. Indeed, history suggests that major infrastructure shifts often coincide with periods of financial excess. The 1920s radio boom, the railroad speculation of the late 19th century, and the dot-com surge of the 1990s shared characteristics with the current moment: transformative technology, unclear business models, and optimism that sometimes exceeded reality.

But history also shows that speculation can coexist with structural change. Most early railroads failed financially but rail networks still reshaped the world. The dot-com bubble burst but the internet matured into core infrastructure. Many early radio firms folded but radio transformed communication.

No doubt bubbles are irrational, but they can also accompany transitions in the underlying metabolism of an economy. Crises often create the regulatory frameworks and governance norms that stabilize the next era. After 1929 came the SEC and financial reform. After 2008 came new capital rules. A future AI recalibration could produce oversight, safety institutions, antitrust interventions, energy planning frameworks, or new forms of industrial coordination.

Where We Are Headed

The AI moment looks less surprising when viewed through this longer lens. What appears at first as a technological leap or a speculative craze also reflects deeper structural forces:

• the maturation of cognitive capitalism,

• the investment needs of firms operating at unprecedented scale,

• the reemergence of industrial-scale integration,

• the geopolitical race for technological and metabolic sovereignty, and

• the beginning of a new infrastructure cycle centered on compute and energy.

AI turns knowledge work into an energy-intensive, resource-dependent industrial system (builders love that, climate be damned!) - one that still involves human judgment, but now interacts with machine inference and automated prediction.

This shift does not answer every social or ethical question that AI raises. Nor does it guarantee efficiency or productivity gains on the timeline markets expect. Some expectations will prove prescient; others will fade. But the underlying trajectory toward integrated systems that combine energy, compute, data, and cognitive labor feels like the foreshadowing of a cyborg future: an emerging era defined by systems that integrate computation, energy, and intelligence into a single metabolic whole.

We are still early in this process. It will take time to see what endures, what adjusts, and what requires rethinking. But the scale and character of current investments suggest that the world is not returning to older models.