Startups financed by token sales innovate less than their VC-backed peers, driven by differing incentives and investor types; meanwhile, generative AI is shrinking the organizational scale of AI entrepreneurship, enabling more solo founders but changing product novelty and quality outcomes.

Main Finding

This dissertation contains two empirical essays showing how two digital disruptions—blockchain-based token financing and generative AI (GenAI)—reconfigure entrepreneurial innovation.
- Essay 1 (blockchain financing): Startups that raise via public token offerings (PTOs/tokens) display lower product-innovation activity than comparable VC-backed startups. Misaligned incentives between token-holders (often customers/retail investors) and entrepreneurs, plus differences in investor value‑adding (vs. professional VCs), explain the decline. Certain design and founder factors (e.g., token retention design, founder experience) mitigate the negative effect. PTOs also affect employment composition and long-run firm outcomes (growth/failure).
- Essay 2 (GenAI and teams): After the emergence of GenAI, launching AI products is associated with smaller founding teams and a marked rise in solo entrepreneurship. GenAI lowers the organizational scale required to build AI products (feasibility-shifting), changing product structure and producing heterogeneous downstream effects on product quality and novelty.

Key Points

• Tokens vs. Equity
  • PTO-backed startups decrease publicly observable innovative activity (measured via topic-annotated tweets and patent-related outcomes) relative to matched VC-backed peers.
  • Mechanisms: (1) agency and incentive misalignment in token financing (tokens often reward usage/short-term engagement rather than long-term R&D); (2) investor-type differences—customer/retail investors willing to finance go-to-market or demand-side investments rather than deep product R&D; (3) reduced VC value-added (strategy, network, governance) when tokenization substitutes for equity.
  • Bottom-line consequences include altered employment growth and job composition; mixed effects on success/failure depending on innovativeness and token design.
  • Moderators: serial/experienced founders, token-retention mechanisms, and third‑party platform transparency can attenuate negative innovation impacts. Robustness: startup-specific trends, synthetic control alternatives, and within-PTO analyses support results.
• GenAI and Organizational Scale
  • GenAI emergence correlates with a shift toward smaller teams for AI product launches; many projects become “solo” endeavors.
  • Mechanism: GenAI reduces technical fixed costs and raises feasibility of delivering AI capabilities with fewer specialized personnel (tools automate model development, content generation, and parts of engineering).
  • Heterogeneity: effects vary by market competitiveness, target users, and product generativity (how much the product can produce new content/features).
  • Downstream: product structure changes (more single-author/simple products), with nuanced effects on product quality and novelty; the dissertation tests mediation via survey and instrumented strategies.
• General synthesis
  • Digital technologies broaden access—token sales broaden funding sources; GenAI lowers entry requirements for AI product creation—but they also reshape incentive alignment and organizational complements to innovation, with potential trade-offs for long-run frontier innovation.

Data & Methods

• Multi-source empirical approach across both essays, combining textual, administrative, and quasi-experimental methods.
• Essay 1 (Tokens vs Equity)
  • Data: PTO/token sale records, VC investment records, startup social media (tweets), patent filing data, employment data, CoinMarketCap listing info, and token-transfer metadata.
  • Measurement: topic modeling / supervised topic classification of startup tweets to quantify product-innovation–related activity; patent counts as hard innovation outcomes; employment counts and job-type categories for bottom-line impact.
  • Identification: matched-sample comparisons (PTO vs VC-backed startups), cross-sectional regressions, difference‑in‑difference (DiD) event-study around PTO dates, synthetic control robustness checks, within-PTO analyses using token transfer patterns.
  • Robustness: inclusion of startup fixed effects, startup-specific time trends, alternative matching and donor pools, and moderator analyses (founder experience, platform transparency, token design features).
• Essay 2 (GenAI and Teams)
  • Data: product-launch announcements/posts, startup/team size and founding-team composition records, product descriptions; LLM-based annotation to measure "AI proximity" (how AI-centric a product is), domain and target-user coding, and survey responses for mechanism checks.
  • Measurement: AI proximity score (multiple embedding/text strategies, LLM annotation for generativity and target users), binary/continuous indicators for AI product launches, team-size and solo vs non-solo classification, and product-level quality/novelty metrics.
  • Identification: event-study around the GenAI inflection, instrumental-variable (IV) constructions to address endogeneity, difference-in-differences and panel regressions, heterogeneity analyses by domain competitiveness and product generativity, and mediation via survey evidence.
  • Robustness: alternative embedding strategies, sensitivity tests for IV, supplementary LLM annotation pipelines, and cross-validation with survey responses.

Implications for AI Economics

• Technology as a fixed-cost shifter: GenAI empirically reduces the effective fixed costs and specialized labor requirements of AI product development. Models of AI innovation should incorporate endogenous organizational scale—GenAI changes the cost function so smaller teams (and solo founders) become viable entrants, increasing firm-level heterogeneity and potentially the number of market entrants.
• Financing and innovation incentives: Tokenization alters the investor mix (toward customers/retail holders) and reward structures, shifting startup investment from long-horizon R&D toward demand-side, engagement, or short-term features. Economic models of innovation should treat financing instruments as endogenous determinants of R&D effort and innovation direction—not only as capital supply.
• Returns to scale and market structure: Lower entry costs (via GenAI) may intensify competition in AI product markets, compress incumbents’ advantages, and increase small-scale, niche innovation. At the same time, token financing may accelerate commercialization of user-facing features at the expense of deep, risky innovations—potentially reducing technological frontier progress if VC-like professional investors are crowded out.
• Labor demand and skill composition: With more solo and small-team AI projects, demand may tilt from large engineering teams toward individuals with prompt-engineering, system-integration, and model-tuning skills. This has implications for wage structures, labor mobility, and human-capital investment in AI-related skills.
• Policy and institutional design:
  • For token markets: regulatory and platform mechanisms that improve information, align long-term retention (e.g., vesting, retention tokens), and encourage professional intermediary roles could preserve incentives for deeper R&D. Policymakers should consider investor protection and disclosure standards to avoid short-termism.
  • For GenAI-enabled SMEs/solo founders: support for quality assurance, safety, and standards may be important as product proliferation increases—e.g., certification for models used in safety-critical domains, funding for high-risk R&D that solo founders are unlikely to pursue, or incubator programs to help scale promising solo innovations.
• Future research directions for AI economics:
  • Quantify the long-run welfare trade-off between broader access (more entrants, faster commercialization) and potential slowdown in frontier R&D when financing shifts away from professional investors.
  • Model interactions between financing instruments and technology that changes organizational scale—how do financing choice, market competition, and cost-shifting technologies co-determine innovation trajectories?
  • Study labor-market dynamics as GenAI changes task composition across firm sizes; evaluate impacts on productivity, employment quality, and inequality.

Overall, the dissertation documents that digital disruptive technologies rewire both the supply of finance and the organizational scale of production. For AI economics, this means treating financing mechanisms and AI tool availability as joint determinants of who can innovate, what they build, and how innovation aggregates into market-level outcomes.