This thesis analyses artificial intelligence as a hybrid Schumpeterian technological regime, focusing on the relationship between firm entry, innovative concentration, and industrial dynamics. Drawing on the literature on technological regimes and Schumpeterian models of innovation, the study interprets AI as a technological domain characterised by the coexistence of Mark I and Mark II dynamics. On the one hand, artificial intelligence displays broad technological opportunities, cross sectoral applicability, and relatively open entry conditions, especially in downstream and application-oriented segments. On the other hand, innovation in core and infrastructural segments remains shaped by strong cumulativeness, high appropriability, and control over complementary assets such as data, computational infrastructure, platforms, and advanced research capabilities. The empirical analysis is based on a patent dataset concerning AI-related inventions over the period 2006–2024. The dataset includes 19,459 AI-related patents, 95,172 patent–technology class observations, and 7,019 distinct applicants. The analysis examines the temporal evolution of patenting activity in AI, the technological composition of patents through IPC classifications, the distribution of inventive activity across applicants, concentration ratios, the Herfindahl-Hirschman Index, and entry dynamics. The results show a strong expansion of AI patenting activity, particularly from the mid-2010s onwards, indicating the presence of broad technological opportunities. At the same time, the distribution of inventive activity appears highly asymmetric: a large number of applicants participate in AI innovation, but a restricted group of leading firms accounts for a disproportionate share of patenting output. This evidence is consistent with the interpretation of AI as a hybrid technological regime, in which broad participation and experimentation coexist with cumulative advantages, persistent leadership, and selective concentration. The thesis therefore argues that AI should not be understood simply as a democratizing technology, but rather as a stratified technological system, characterized by a broad periphery of new entrants and a concentrated core of dominant innovators.
Questa tesi analizza l’intelligenza artificiale come un regime tecnologico schumpeteriano ibrido, concentrandosi sulla relazione tra ingresso delle imprese, concentrazione innovativa e dinamiche industriali. Basandosi sulla letteratura sui regimi tecnologici e sui modelli schumpeteriani dell’innovazione, lo studio interpreta l’IA come un dominio tecnologico caratterizzato dalla coesistenza di dinamiche di tipo Mark I e Mark II. Da un lato, l’intelligenza artificiale presenta ampie opportunità tecnologiche, applicabilità trasversale ai settori e condizioni di ingresso relativamente aperte, soprattutto nei segmenti a valle e orientati alle applicazioni. Dall’altro lato, l’innovazione nei segmenti centrali e infrastrutturali resta influenzata da una forte cumulatività, da un’elevata appropriabilità e dal controllo di asset complementari quali dati, infrastrutture computazionali, piattaforme e capacità avanzate di ricerca. L’analisi empirica si basa su un dataset brevettuale relativo alle invenzioni connesse all’IA nel periodo 2006–2024. Il dataset comprende 19.459 brevetti legati all’IA, 95.172 osservazioni brevetto–classe tecnologica e 7.019 richiedenti distinti. L’analisi esamina l’evoluzione temporale dell’attività brevettuale nell’IA, la composizione tecnologica dei brevetti attraverso le classificazioni IPC, la distribuzione dell’attività inventiva tra i richiedenti, i rapporti di concentrazione, l’indice di Herfindahl-Hirschmann e le dinamiche di ingresso. I risultati mostrano una forte espansione dell’attività brevettuale nell’IA, in particolare a partire dalla metà degli anni 2010, indicando la presenza di ampie opportunità tecnologiche. Allo stesso tempo, la distribuzione dell’attività inventiva risulta fortemente asimmetrica: un ampio numero di richiedenti partecipa all’innovazione nell’IA, ma un gruppo ristretto di imprese leader rappresenta una quota sproporzionata dell’output brevettuale. Questa evidenza è coerente con l’interpretazione dell’IA come regime tecnologico ibrido, nel quale ampia partecipazione e sperimentazione coesistono con vantaggi cumulativi, leadership persistente e concentrazione selettiva. La tesi sostiene quindi che l’IA non debba essere intesa semplicemente come una tecnologia democratizzante, ma come un sistema tecnologico stratificato, caratterizzato da un’ampia periferia di nuovi entranti e da un nucleo concentrato di innovatori dominanti.
INTELLIGENZA ARTIFICIALE COME REGIME TECNOLOGICO SCHUMPETERIANO IBRIDO: INGRESSO, CONCENTRAZIONE E DINAMICHE INDUSTRIALI
MONTAGNA, SIMONE
2025/2026
Abstract
This thesis analyses artificial intelligence as a hybrid Schumpeterian technological regime, focusing on the relationship between firm entry, innovative concentration, and industrial dynamics. Drawing on the literature on technological regimes and Schumpeterian models of innovation, the study interprets AI as a technological domain characterised by the coexistence of Mark I and Mark II dynamics. On the one hand, artificial intelligence displays broad technological opportunities, cross sectoral applicability, and relatively open entry conditions, especially in downstream and application-oriented segments. On the other hand, innovation in core and infrastructural segments remains shaped by strong cumulativeness, high appropriability, and control over complementary assets such as data, computational infrastructure, platforms, and advanced research capabilities. The empirical analysis is based on a patent dataset concerning AI-related inventions over the period 2006–2024. The dataset includes 19,459 AI-related patents, 95,172 patent–technology class observations, and 7,019 distinct applicants. The analysis examines the temporal evolution of patenting activity in AI, the technological composition of patents through IPC classifications, the distribution of inventive activity across applicants, concentration ratios, the Herfindahl-Hirschman Index, and entry dynamics. The results show a strong expansion of AI patenting activity, particularly from the mid-2010s onwards, indicating the presence of broad technological opportunities. At the same time, the distribution of inventive activity appears highly asymmetric: a large number of applicants participate in AI innovation, but a restricted group of leading firms accounts for a disproportionate share of patenting output. This evidence is consistent with the interpretation of AI as a hybrid technological regime, in which broad participation and experimentation coexist with cumulative advantages, persistent leadership, and selective concentration. The thesis therefore argues that AI should not be understood simply as a democratizing technology, but rather as a stratified technological system, characterized by a broad periphery of new entrants and a concentrated core of dominant innovators.| File | Dimensione | Formato | |
|---|---|---|---|
|
Tesi S.M. 19.06.2026 last.pdf
accesso aperto
Descrizione: Master's Degree Thesis
Dimensione
954.24 kB
Formato
Adobe PDF
|
954.24 kB | Adobe PDF | Visualizza/Apri |
È consentito all'utente scaricare e condividere i documenti disponibili a testo pieno in UNITESI UNIPV nel rispetto della licenza Creative Commons del tipo CC BY NC ND.
Per maggiori informazioni e per verifiche sull'eventuale disponibilità del file scrivere a: [email protected].
https://hdl.handle.net/20.500.14239/35992