Creative Destruction: From Schumpeter to AI (3/3) 

June 2025 

The Schumpeterian Growth Paradigm: Innovation, Disruption, and the Dynamics of Capitalist Evolution

At the heart of the Schumpeterian growth paradigm lies a vision of capitalism as a restless, shape-shifting system—driven less by equilibrium than by eruption. It is not the slow accumulation of efficiency that propels economies forward, but waves of disruptive innovation. This vision rests on three intertwined forces: the cumulative nature of knowledge, the lure of monopoly profits to entrepreneurs, and the mechanism of creative destruction. In this model, old firms and industries must perish for the new to thrive.
But the engine that drives progress also generates instability. Innovation yields monopoly rents; yet those rents, once secured, tend to foster sclerosis. Incumbents use their winnings not to re-invest in new ideas, but to entrench their dominance—acquiring rivals, stifling competition, and insulating themselves from change. Thus, the Schumpeterian model contains an inbuilt tension: capitalism needs temporary monopolies to innovate, but enduring monopolies to stagnate.

The most recent wave of innovation—spurred by information technology and early biotech—gathered pace in the late 20th century. Microprocessors and the internet reshaped production, spawning global supply chains and intangible-heavy business models. The same period saw breakthroughs in genetic engineering, though regulation and public mistrust slowed their broader uptake. This technological shift coincided with a macroeconomic transformation: the retreat of Keynesianism, the rise of neoliberalism, and a turn toward deregulation and mobile capital.

Yet signs of fatigue are evident. Progress in semiconductors has decelerated as physical limits bite and industry consolidation stifles competition. Biotech, despite its promise, remains hemmed in by high costs and ethical constraints. More broadly, productivity growth in the rich world has slowed since the 2010s—despite the technological fireworks.

That slowdown may reflect deeper structural frictions. Schumpeter saw the dynamism of capitalism rooted in the vitality of new entrants. Today, that vitality is under threat. Dominant firms have learned to buy out budding rivals before they become a threat. Non-compete clauses restrict labour mobility. Inventors hired by incumbents often become less productive, suggesting their talents are being neutralised rather than harnessed. The result is not creative destruction, but defensive stagnation.

This raises a difficult question. Are the profits of today’s corporate giants a reward for innovation, or the spoils of rent-seeking? If the former, antitrust action could blunt incentives. If the latter, reform is urgent. Concentrated markets tend to impede the diffusion of innovation—slowing productivity, fuelling inequality, and entrenching privilege. Growth falters not because new ideas are lacking, but because the old guard blocks their advance.
Empirical evidence suggests innovation is most vigorous when rivals are evenly matched. If one firm pulls too far ahead, others give up the chase—and the leader coasts. Much modern R&D is defensive: not aimed at creating new frontiers, but at fortifying old castles. Market dominance becomes an end in itself, not a by-product of invention.
The geography of innovation also matters. In low-income countries, development is driven by capital accumulation and basic infrastructure. Middle-income economies climb by absorbing foreign technologies. But to make the leap to high-income status, a nation must become an innovator itself. That requires institutional reform: flexible labour markets, competitive ecosystems, and agile regulation. Where such reforms are blocked—as in Japan or South Korea—innovation slows and growth stalls.

Not all innovation is alike. Some innovations complement existing factors—raising labour productivity without reducing headcount. Others automate tasks and shed jobs. The most transformative innovations create new categories of work altogether. Their effects are uneven: industrial robots, for instance, have boosted efficiency in some American regions but destroyed routine jobs in others. Traditional growth models struggle to explain such variation, because they ignore how innovation reshapes the task content of work.

Schumpeterian growth, then, is not a linear march. It is a messy, discontinuous churn. The challenge is to preserve its dynamism without succumbing to its pathologies. Innovation must remain a process of broad renewal, not narrow dominance. That demands vigilance from policymakers, foresight from firms, and pressure from civil society. Capitalism cannot be sustained on disruption alone. It must also find ways to share its spoils.

Economic Policy and the Future of Creative Destruction
For much of the post-war period, rising productivity powered rising living standards. But in recent decades, the engine has sputtered—even in economies closest to the technological frontier. Today, some 95% of firms in rich countries report little or no productivity gains. Despite spectacular advances in digital technologies, the economic returns have been patchy and, increasingly, captured by a shrinking slice of “frontier” firms. The result is a troubling paradox: innovation abounds, yet its benefits remain narrowly held.

Globalisation, once a powerful conveyor belt for spreading best practices, has lost steam. Since the global financial crisis, trade as a share of GDP has plateaued. The pandemic further fractured supply chains. And emerging markets, far from forging new regional blocs, remain poorly integrated with one another. The weakening of global competition has also blunted a key disciplining force—one that once encouraged firms to raise their game or fall behind.
Inequality has grown alongside this divergence. The firms that sit at the technological edge enjoy outsized profits and pay much higher wages. But they employ only a minority of the workforce. For the median worker, particularly those born after the 1960s, wage stagnation and falling intergenerational mobility are the new normal. The problem is not too much growth, but too little inclusion. A capitalism that fails to share its spoils erodes its own legitimacy.
That does not mean embracing degrowth or abandoning innovation. Rather, the task is to spread innovation’s gains more widely. Redistribution via taxation is necessary but insufficient. A more robust agenda must focus on diffusion—ensuring that technologies, skills and capabilities reach beyond the elite. That calls for active policies: better education and training systems, pro-competitive regulation, and smarter industrial strategies. Innovation can lift top incomes, but if designed well, it can also broaden opportunity and reduce rent-seeking.

Creative destruction remains a central feature of capitalism. But it does not operate in a vacuum. Roughly 30% of all jobs are reallocated every five years, even within narrowly defined sectors. Transitions such as the fall of Blockbuster and the rise of Netflix may look clean in hindsight, but the underlying churn is messy and painful. Aggregate statistics often hide this turmoil.

Policy determines whether this churn strengthens or weakens the economy. Recessions, for example, can have a “cleansing” effect by weeding out inefficient firms. But they can also induce “sulling”, where long-term scarring prevents renewal. After the euro crisis, much of Southern Europe fell into the latter trap. Cheap credit and regulatory forbearance propped up large, unproductive incumbents, while starved smaller firms struggled to grow. The result: zombie capitalism and stagnant productivity.

Technological change is not destiny. Whether artificial intelligence is used to automate jobs or augment them is not determined by engineers alone. It reflects political economy: who holds power, what institutions exist, and how incentives are shaped. Much corporate innovation is defensive. Big firms deploy patents as barriers, engage in regulatory capture, and buy rivals to preempt disruption. The outcomes are predictable: more digital investment, fewer broad-based gains.

Nowhere is this clearer than in Britain. Since 2008, productivity has barely budged. Business investment has languished. Brexit offered an opportunity to rethink industrial policy, but bold rhetoric has yielded timid action. Infrastructure spending has been patchy. Skills policies have lacked follow-through. The result: a proliferation of zombie firms, propped up by cheap debt and hesitant regulators, draining dynamism from the wider economy.

Reinvigorating creative destruction demands more than nostalgia for Silicon Valley or mimicry of America’s venture capital model. European policymakers must double down on foundational assets: education, research, infrastructure. Competition policy must be modernised to tackle the peculiarities of the digital age—data moats, winner-takes-all markets and platform monopolies. Civil society should help hold both firms and states to account. Climate innovation, meanwhile, requires public-private alliances strong enough to overcome market inertia.

Just as vital is knowing what to avoid. Over-taxing capital will stifle investment. And degrowth is not a serious path to prosperity. Nor can innovation strategies be copy-pasted from one jurisdiction to another; they must reflect institutional nuance and local industrial capacity.

Innovation remains capitalism’s most powerful engine. But unless its fruits are shared more broadly, its legitimacy—and its future—will be at risk. Without diffusion, creative destruction becomes exclusionary. And growth, far from being a tide that lifts all boats, turns into a narrow current accessible to only a few.

AI and Creative Destruction: A Keynesian Reflection on the Future of Growth, Disruption, and Direction

There are epochs in which the economist, like the prophet, must peer beyond equations and aggregates to discern the deeper movements of civilization. Ours is such a moment. Artificial intelligence now stands not as a mere technical augmentation nor as an auxiliary to human effort, but as a general-purpose force—a system of systems—capable of reconfiguring the very structure of production, labour, and social order. If Schumpeter saw capitalism as an engine of creative destruction, it is now the algorithm that threatens to become both engine and engineer. The question that confronts us is not whether AI will transform our economy, but whether that transformation will be governed by design or drift.

Unlike the steam engine or the electric motor, artificial intelligence does not emerge into a blank landscape. It is born into a world already saturated with data, computation, and digital logic. Its power lies less in its novelty than in the peculiar form of leverage it enjoys—rooted in prediction, shaped by learning loops, and propelled by access to enormous proprietary datasets. The natural language interface, which renders this intelligence legible and usable by the non-specialist, lowers the barriers of entry and spreads its reach into law offices, logistics hubs, medical clinics, and marketing departments. Yet in doing so, it consolidates power around those who own the means of data production, giving rise to a new asymmetry—informational rather than industrial.

As in all great technological upheavals, the outcomes are indeterminate. AI's impact unfolds along three principal axes—productivity, inequality, and industrial structure—each fraught with promise and peril. Will AI be a genuine spur to productivity, or merely another case of technological optimism outrunning organizational readiness? If integration lags, if institutions resist, if incentives falter, then AI may fall into the category of “so-so” technologies—those praised in PowerPoints but impotent in practice. But the same systems, if recursively deployed and well-complemented by human agency, may become compounding catalysts—augmenting not only manual effort but cognitive output, generating knowledge at a pace hitherto unimaginable. In such a world, the production function itself shifts, and productivity becomes not the sum of tasks done faster, but the birth of entirely new capacities.

So too with inequality. The current trajectory suggests a bifurcated outcome. Unchecked, AI favours capital-rich incumbents and highly skilled elites, threatening the employment and bargaining power of routine and even mid-tier cognitive workers. But this is not destiny. Where AI is made accessible—through copilots, assistants, and guided interfaces—it can uplift the novice and the layperson, enhancing rather than replacing their effort. In the end, the distributive effects of AI will be determined not by its code, but by its context—by the institutions that frame its adoption, the policies that mediate its diffusion, and the ethics that govern its reward.

The industrial structure is likewise in flux. AI favours scale. The computational and infrastructural demands of large models tilt the playing field toward the giants—those who possess the cloud capacity, the talent, the regulatory fluency, and above all, the data. As in previous industrial epochs, the risk is that innovation becomes synonymous with consolidation, that disruption gives way to entrenchment. Yet countervailing currents are stirring. Open-source initiatives, collaborative research communities, and modular development frameworks point to another path—one where innovation is plural, distributed, and less beholden to capital intensity. The outcome, again, is undecided. The technology may centralize or decentralize. The deciding variable is not technical but political.

The productivity of AI may also be delayed and distorted by our instruments of measurement. Like electricity in the early twentieth century, AI’s impact is likely to be underestimated by official statistics. Knowledge work, creative processes, and decision augmentation elude easy quantification. Gains appear as “silent productivity”—real but invisible, present in quality of service, speed of response, and cognitive surplus, but absent from GDP tables. This is the paradox of the productivity J-curve: disruption precedes documentation; the returns of transformation come only after turbulence has passed.

From this uncertainty emerge three plausible futures. One is a path of continuity, in which AI merely amplifies existing economic structures without altering their trajectory. Inequality deepens, institutions adapt slowly, and the promise of AI turns bitter. A second path offers renewal—a new paradigm in which AI converges with biotechnology, transforming education, medicine, and discovery. But such a future demands institutional imagination: governance frameworks, ethical norms, labour adaptation, and infrastructure design that rise to meet the moment. The third path is breakdown. AI amplifies dislocation without generating cohesion. Mistrust grows, polarization hardens, and growth gives way to volatility. In such a world, the disruption is not creative—it is corrosive.

Already, we see frictions in the Schumpeterian machine. Creation outpaces destruction. Startups, unable to compete on equal terms, are forced to leapfrog or perish. Incumbents entrench themselves not through productive dynamism but through acquisition, litigation, and regulatory capture. Innovation becomes defensive—an insurance policy against disruption, not an engine of renewal. The cycle stalls. In such a climate, creation without destruction becomes stagnation in disguise.

What then is the economist—or the policymaker—to do? As in the 1930s, we must turn not to laissez-faire resignation but to deliberate economic engineering. The policies required are not esoteric. Promote labour-complementary AI. Broaden access to infrastructure. Reform intellectual property to prevent enclosure of algorithmic commons. Support open-source communities. Redesign safety nets to ease transitions rather than cushion collapse. These are not merely acts of redistribution. They are investments in resilience—in the capacity of society to absorb disruption without falling into disorder.

Artificial intelligence is malleable. It is not fated to centralize or democratize, to enrich or impoverish. Its effects are contingent on how we choose to govern it—how we frame the incentives, structure the institutions, and define the metrics of progress. In time of fracture and flux, stability is not the absence of change but the successful navigation of it. In the face of great transformation, renewal must be more than a sentiment. It must be a programme.