Installation and Deployment Phases
Carlota Perez's framework, developed in her 2002 Technological Revolutions and Financial Capital, distinguishes two phases in the life cycle of each major technological revolution. The installation phase is the early period during which the technology develops, early adopters experiment with it, and financial capital drives speculation that produces both infrastructure and bubbles. The deployment phase is the subsequent period during which the technology becomes broadly adopted, its benefits are widely distributed, and institutional capital governs its integration into the economy. The transition between phases is typically crisis-driven and requires what Perez calls institutional innovation — new rules, new norms, new arrangements that channel the technology's power toward broad benefit. The Andreessen — On AI volume deploys this framework to position the current AI moment as the transition between installation and deployment, and to identify the institutional innovation the moment requires.
The Pattern as Rationalization — Contrarian ^ Opus
There is a parallel reading in which Perez's framework functions less as discovery than as retrospective sense-making—a schema that imposes coherence on messy historical transitions by selecting the evidence that fits. The two-phase pattern appears robust because the framework defines what counts as a 'technological revolution' by whether it exhibits the pattern, creating a selection effect that guarantees confirmatory cases. The crises identified as turning points (1930s depression, 2008 financial crisis) were overdetermined events with multiple causes; reading them primarily as technology-phase transitions privileges one explanatory thread over economic, geopolitical, and ecological factors that may have been equally or more determinative.
The application to AI inherits these problems while adding new ones. Whether AI represents installation or deployment depends entirely on how you draw the boundaries—is it continuous with digital computing (deployment), or discontinuous enough to warrant new-revolution status (installation)? The framework cannot answer this question; it can only restate it in different terms. More critically, the framework's premise—that financial crisis creates productive pressure for institutional innovation—mistakes correlation for mechanism. The 1930s produced the New Deal, but also produced fascism. The 2008 crisis produced Dodd-Frank, but also produced austerity and the conditions for authoritarian resurgence. Crisis creates pressure, but whether that pressure yields broad-based institutional innovation or elite capture depends on power dynamics the framework does not model. Applying it to AI risks mistaking a descriptive pattern for a prescriptive path—assuming the turning point will arrive and institutional innovation will follow, when both remain radically contingent on choices not yet made.
In the AI Story
Perez's framework emerged from detailed historical study of five technological revolutions since 1771: the industrial revolution, the age of steam and railways, the age of steel and electricity, the age of oil and mass production, and the age of information and telecommunications. In each case, Perez identified the same two-phase pattern, separated by a turning point typically associated with financial crisis.
The installation phase is characterized by specific features: financial capital dominates, speculative bubbles form and burst, infrastructure is built ahead of demand, and the economy bifurcates between the new sector (growing rapidly) and the old economy (stagnating). The deployment phase replaces these features with their complements: production capital dominates, broad-based prosperity becomes possible, institutional arrangements channel the technology toward mass adoption, and the economy integrates the new technology across all sectors.
The turning point between phases is, in Perez's framework, typically associated with financial crisis. The 1930s depression was the turning point for the age of steel and electricity. The 2008 financial crisis was, in Perez's reading, the turning point for the age of information. The crisis exposes the limits of the installation-phase arrangements and creates political pressure for the institutional innovation the deployment phase requires.
The framework's application to AI is contested. Some readings — including Andreessen's — treat AI as a continuation of the information age, meaning the current moment represents deployment-phase dynamics playing out. Other readings — including Perez's own recent work — treat AI as potentially the opening of a sixth technological revolution, meaning the current moment represents a new installation phase. The distinction matters for policy: deployment-phase institutional innovation looks very different from installation-phase experimentation.
For the Andreessen — On AI volume, the framework provides analytical distance from the immediate intensity of the AI discourse. The Software Death Cross, the speculative valuations, the institutional lag — these are features the framework predicts and specifies. The volume argues that navigating them requires the specific form of institutional innovation Perez's framework identifies, and that the absence of adequate institutional innovation is the primary risk of the current moment.
Origin
Perez developed the framework across a career that began with her 1983 article on technological change and economic cycles, and culminated in her 2002 book. The framework extended and systematized observations by Schumpeter, Kondratiev, and Chris Freeman, organized around the specific finding that financial and production capital play systematically different roles across the phases of each revolution.
Key Ideas
Two-phase structure. Each technological revolution unfolds in an installation phase (speculative, financial-capital-driven) followed by a deployment phase (institutional, production-capital-driven), separated by a crisis-driven turning point.
Financial vs. production capital. The two phases are distinguished by which form of capital dominates — financial capital's preference for liquidity and speculation during installation, production capital's preference for patient investment during deployment.
Institutional innovation requirement. The transition between phases requires the construction of new institutional arrangements — labor law, regulatory frameworks, distributive mechanisms — that the installation phase does not produce on its own.
Crisis as turning point. The passage from installation to deployment is typically triggered by financial crisis, which exposes the inadequacy of installation-phase arrangements and creates political pressure for reform.
AI transition ambiguity. Whether AI represents a new installation phase or the deployment phase of the information revolution determines what institutional innovation the moment requires — an analytical question with major policy consequences.
Debates & Critiques
The framework is the subject of continuing debate in both academic and policy contexts. Its critics argue that the two-phase pattern is imposed on the historical record rather than derived from it, and that the specific timing of phase transitions cannot be predicted in ways the framework suggests. Its defenders argue that the general pattern is robust even where specific details are contested, and that its application to contemporary questions — including AI — remains productive. The question of whether AI represents a new revolution or the continuation of the information age is specifically open within the framework itself.
Appears in the Orange Pill Cycle
Framework as Conditional Map — Arbitrator ^ Opus
The right weighting here depends on what question you're asking. If the question is "Does history exhibit recurring patterns in how societies absorb major technologies?"—the answer is substantially yes (75% Perez). The two-phase structure, the role of speculative capital in building infrastructure, the institutional lag that produces crisis—these appear across enough cases to constitute genuine pattern. If the question is "Can we predict when phase transitions will occur or what form institutional innovation will take?"—the answer tilts heavily contrarian (70%). The framework identifies necessary conditions but cannot specify sufficient ones; it names what must happen without determining whether it will.
The AI application turns on a third question: "What does the current moment structurally resemble?" Here the framework's value is methodological rather than deterministic (60% Perez, 40% contrarian synthesis). It provides analytical distance—a way to recognize installation-phase dynamics (speculative capital, infrastructure build-ahead, institutional lag) without treating them as unprecedented. But the contrarian reading is right that the framework cannot adjudicate its own key ambiguity: whether AI is a new revolution or the information age's deployment. This matters because the institutional innovation required differs radically between cases.
The synthetic frame the topic benefits from treats Perez's framework not as prediction but as conditional mapping: IF we are in an installation-to-deployment transition, THEN these are the characteristic dynamics and required innovations. The framework's value lies in naming what to look for and what forms of intervention might matter—not in guaranteeing those interventions will occur. The risk it identifies (inadequate institutional innovation) remains real even if the pattern that surfaces it is partly constructed. Crisis creates possibility, not inevitability; recognizing the pattern is the beginning of institutional work, not its substitute.
Further reading
- Carlota Perez, Technological Revolutions and Financial Capital (2002).
- Carlota Perez, "Technological Revolutions and Techno-Economic Paradigms," Cambridge Journal of Economics (2010).
- Carlota Perez, "AI's Place in History" (2024).
- Mariana Mazzucato, The Entrepreneurial State (2013).