Executive Summary

Start with the most fundamental question in the data economy: who owns the data you produce every day? Legally, "data ownership" remains undefined. The EU Data Act grants rights of access, use, and portability -- but deliberately avoids the word "ownership." A Hohfeldian property-rights analysis shows this omission is intentional: because data is non-rivalrous, traditional ownership concepts do not map cleanly onto it.

The market distortion this legal gap creates is visible in the numbers. The global data broker market stands at $291-313B (2025), while pure data marketplace platforms account for just $1.81-1.86B. That is a 170x gap. Opaque intermediary trading thrives, yet transparent markets where data producers participate directly have barely formed.

1.1 Platforms Harvesting Data Labor

Reddit signed AI licensing deals worth $60M/year with Google, an estimated ~$70M with OpenAI, and a cumulative $203M in total. Stack Overflow's $115M revenue (+17%) increasingly depends on AI licensing. Both platforms monetize user-generated content at scale, yet none of that revenue flows back to the contributors. The "data as labor" framework proposed by Arrieta-Ibarra (2018) is gaining renewed attention for good reason.

1.2 Exploding Demand for AI Training Data

The AI training dataset market is projected to grow from $2.82-3.2B (2024) to $7.23-8.6B (2030). The synthetic data market is expanding even faster at a CAGR of 35-42%. Data labeling alone is a $2.23-3.77B business. At the very moment data demand is exploding, there is no consensus on who owns it -- a structural time bomb.

1.3 The Rise of Data Sovereignty Frameworks

Teichmann's three-dimensional model of data sovereignty identifies the core imbalance. Of the three dimensions -- "protection" (GDPR), "participation" (economic reward), and "provision" (public benefit) -- the market has focused almost exclusively on protection while sidelining economic participation. The EU Data Act, DGA (Data Governance Act), and GAIA-X aim to restore this balance, but the Data Governance Trilemma (DGT) remains an inherent tension: rights protection, economic value, and public benefit are difficult to satisfy simultaneously.

A CNIL survey of EU consumers found that the willingness-to-accept price for personal data is EUR 120-360/year, with 35% refusing at any price. South Korea's data industry grew from KRW 27.15 trillion (2023, confirmed) to an estimated KRW 30.75 trillion (2024) -- roughly $20-23B -- driven by the MyData 2.0 initiative and expanding data voucher programs.

Web3 addresses the data ownership gap through three mechanisms: collective ownership (DataDAO), decentralized collection (DePIN), and contribution verification (Proof-of-Contribution). Yet quality verification and governance centralization remain structural limitations.

2.1 DataDAO: Communities Own the Data

A DataDAO is best understood as a "blockchain-native data cooperative." In Buehler's taxonomy of data cooperatives, trusts, commons, and unions, it represents the most programmable model. Contributors submit data to a community treasury and receive token rewards proportional to their contribution. Governance runs on token voting.

MIT's Pentland research group has compared value creation between platforms and data cooperatives as digital commons. Li proposed a public data trust model that licenses AI training data and shares revenue. The theoretical foundation is solid.

However, Kioupkiolis's analysis warns of structural flaws in DAO governance: declining participation, re-centralization of decision-making, and failure to adapt to changing conditions. More DataDAOs do not automatically mean better governance.

2.2 DePIN: Decentralizing Physical Data Infrastructure

DePIN (Decentralized Physical Infrastructure Networks) is a model where individuals operate physical hardware -- dashcams, hotspots, GPS sensors -- and earn token rewards. Under Zichichi's three-axis classification (distributed ledger, cryptoeconomics, physical infrastructure), AI-related projects account for 59.3% of all DePIN activity.

The core incentive insight behind DePIN is that speculative token value can bootstrap infrastructure before market demand materializes. Token rewards attract early participants, and network effects eventually generate real demand. Research shows, however, that monetary incentives alone are insufficient for sustainability; non-monetary incentives -- community belonging, data access rights -- must complement them.

2.3 Proof-of-Contribution and Data Valuation

Proof-of-Contribution is the mechanism that verifies data submissions and distributes token rewards. The central challenge is measuring the value of a contribution. Shapley-value-based data valuation has become computationally feasible even for LLM fine-tuning. Yet in practice, most Web3 projects limit quality checks to schema validation, and none apply ML-grade contribution assessment at scale.

2.4 Compute-to-Data: Privacy-Preserving Exchange

Privacy-preserving data exchange is another pillar of the Web3 data economy. SMPC (Secure Multi-Party Computation) combined with blockchain, and the Compute-to-Data (C2D) pattern, bring computation to the data instead of moving data to the computation. The D2M framework integrates on-chain auctions, off-chain federated learning, and incentive-compatible revenue sharing. ZKP (Zero-Knowledge Proof)-based verifiable ML is also advancing toward full-pipeline verifiability across training, testing, and inference.

As of 2026, a clear gap is emerging between projects that have achieved product-market fit (PMF) and those still validating their model. The DePIN market cap has corrected from a peak of $19.2B to ~$9-10B, and valuation multiples have normalized from 1,000x+ to 10-25x revenue.

The following table organizes major projects by tier.

3.1 What Tier 1 Has Proven: Real-Economy Partnerships

The common thread among Tier 1 projects is unmistakable: real-economy partnerships -- not token speculation -- drive growth. DIMO partnered with Progressive and Liberty Mutual for insurance data. Helium contributes to AT&T's telecom infrastructure. Hivemapper's map data feeds VW's autonomous driving and Lyft's dispatch systems. These projects have validated a clear business model: "collect data via token incentives, sell data to enterprises."

3.2 Lessons from Failures and Pivots

In Q1 2026, Polynomial, ZeroLend, and Parsec shut down. Over 300 gaming dApps went inactive (Q2 2025), and X's API policy changes hit InfoFi projects hard. The lesson these cases teach is consistent: token incentives alone are not sustainable. Projects that depend on token price without connecting to real demand do not survive a downturn.

National regulations are both an accelerant and a constraint for Web3 data ownership models. The EU Data Act (effective September 2025) codifies data access rights, lending momentum to decentralized ownership -- but the potential classification of tokens as securities creates a regulatory headwind.

4.1 EU: Codifying Data Sovereignty

In the four-model taxonomy of digital sovereignty, the EU represents the "rights-based" approach. The EU Data Act's core obligations took effect on September 12, 2025; product design obligations follow on September 12, 2026; and cloud switching obligations on January 12, 2027. Violations carry fines of 4-5% of global revenue.

GAIA-X now has 350+ confirmed members working to standardize European data spaces. The DGA (Data Governance Act) introduced "data altruism" -- a legal framework for individuals to voluntarily contribute data for the public good. This regulatory ecosystem is structurally compatible with Web3's community-based data ownership model.

4.2 South Korea: MyData 2.0 and the Data Industry

Following its Data Three Acts (2020), South Korea launched MyData 2.0 in June 2025. The system has linked 165 million data records, with KakaoPay alone at 20 million subscribers. The push to transform KDX (Korea Exchange) into a data marketplace, combined with high demand in data voucher programs, signals strong appetite for data trading infrastructure -- and a growing need for objective data pricing mechanisms.

4.3 United States and Singapore: Market-Driven Approaches

The United States lacks a federal data privacy law, with 19 states enforcing individual privacy statutes. However, SEC Chair Atkins signaled at ETHDenver 2026 that no-action letters for tokenization and DePIN are under consideration. Singapore balances PDPA with a data innovation sandbox. Legal recognition of DAOs is progressing in Wyoming (2021), the Marshall Islands (2022), and UAE RAK.

Gartner projects that 25% of enterprises will use Web3 services by 2027. Regulatory uncertainty remains, but the trend is clear: codification of data access rights is spreading globally.

The AI agent market is projected to grow from $5.1B (2024) to $47.1B (2030) at a CAGR of 44.8%. The M2M services market is expected to reach $164.2B (2030, CAGR 25.5%). An economy where agents autonomously buy and sell data is no longer hypothetical -- it is materializing.

5.1 Payments Are Solved. Trust Is Not.

Payment infrastructure for the agent economy is coming together fast. Coinbase's x402 protocol standardizes M2M payments using the HTTP 402 status code. Google's AP2 (Agent Payment Protocol) implements trust-based agent payments. Stablecoin infrastructure is also maturing. Each of these is covered in dedicated analyses, so we will not repeat the details here.

The crux is this: "how to pay" is being solved, but "what to trust and buy" remains wide open. Already, 37% of The Graph API users are AI agents. In a world where agents autonomously source data, there is no "trust structure" to verify the quality and provenance of what they are buying.

5.2 DataDAOs and DePIN Become the Agent Supply Chain

In the five-layer architecture of the agent economy, the "data asset layer" sits between payment infrastructure and data sourcing. DataDAOs become the agent's data sourcing channel; DePIN becomes the real-time data supply chain.

Concretely, autonomous driving agents could purchase DIMO's vehicle data directly, or delivery agents could source Hivemapper's latest map data in real time. AI Agent crypto projects already number 550+, with a combined market cap of $4.34B. io.net launched Agent Cloud in March 2026, providing compute infrastructure purpose-built for AI agents.

5.3 The Missing Layer: Data Quality Proof

IBis proposed blockchain-based provenance tracking for AI training data -- bidirectional tracing between datasets and models. Longpre identified the fundamental inadequacy of current data provenance frameworks: "authentication, consent, credit, and compensation" are all broken.

A full stack connecting data quality proof, ownership proof, and automated payment must be in place for agent data transactions to function at scale. The payment layer is covered by x402 and AP2. Data sourcing is handled by DataDAOs and DePIN. But the "trust bridge" between these two -- the quality proof layer -- is empty. The "diagnose-remediate-certify" pipeline analyzed in our Data Value Proof report addresses precisely this gap.

The four threads analyzed in this report -- ownership shifts, project landscape, regulatory environment, and the agent economy -- converge on a single intersection. Token incentives drive explosive data volume, but quality goes unverified. Most projects cap their validation at schema checks. This structural gap is Pebblous's opportunity.

Business Alignment: Four Solutions, One Web3 Position

The following table maps Pebblous solutions to their roles in the Web3 data ecosystem.

Positioning: If Chainlink Does Price, Pebblous Does Quality

Chainlink became the trust foundation of DeFi by providing on-chain price-feed oracles. By the same logic, a "data quality oracle" that publishes AI-Ready quality scores for DataDAO/DePIN data on-chain could become the trust foundation of the Web3 data ecosystem. Combining ISO/IEC 5259 with Web3 provenance proofs satisfies both regulatory compliance and market trust.

Scale of the Opportunity

The synthetic data market is growing at a CAGR of 35-42%. The enterprise data monetization market is projected to expand from $4.78-5.22B (2025) to $48.55B (2035). Enterprise data monetization adoption has already reached 68%. DePIN startups alone raised $1B in funding in 2025. Across all these markets, demand for quality verification partners is rising.

When DIMO's vehicle data and Hivemapper's mapping data are repurposed for AI training, quality verification is not optional -- it is mandatory. In South Korea, the KDX data marketplace transition and data voucher programs both require technical solutions for data pricing. In the Data Governance Trilemma, the "quality assurance" layer that bridges "economic value" and "rights protection" is precisely where Pebblous sits.

Frequently Asked Questions

Key questions and answers about Web3 data ownership, DataDAOs, DePIN, and the agent economy.

References

Academic Papers and Research

1. Teichmann, "Data Sovereignty: Three Dimensions of Protection, Participation, and Provision"
2. Hohfeldian property-rights analysis of the EU Data Act: access-use-portability rights bundle
3. Four models of digital sovereignty: rights-based (EU), market-driven (US), centralized, state-based
4. Buehler, taxonomy of data cooperatives / trusts / commons / unions
5. MIT Pentland, digital commons value creation in platforms vs. data cooperatives
6. Data Governance Trilemma (DGT): rights protection vs. economic value vs. public benefit
7. Li, public data trust model: AI training data licensing + revenue sharing
8. Kioupkiolis, structural flaws in DAO governance: declining participation, re-centralization
9. Vipra & Mahari, unique economic properties of data: non-rivalry, context-dependence, emergent rivalry
10. Arrieta-Ibarra, "Should We Treat Data as Labor?" proposal (2018)
11. Zichichi, DePIN three-axis taxonomy: DLT + cryptoeconomics + physical infrastructure
12. Multi-agent simulation: efficiency analysis of decentralized infrastructure
13. DePIN framework: speculative-value-driven infrastructure bootstrapping before market demand
14. SMPC + blockchain privacy-preserving data exchange
15. BlockDaSh reference architecture: data processing-sharing-storage three-layer model
16. D2M: on-chain auction + off-chain federated learning + incentive-compatible revenue sharing
17. Incentive design: effectiveness of combined monetary and non-monetary incentives
18. Shapley-value data valuation: practical-scale computation for LLM fine-tuning
19. IBis: blockchain-based AI training data provenance tracking
20. ISO/IEC 9126-based blockchain data provenance quality attributes
21. Longpre, fundamental inadequacy of current data provenance frameworks
22. ZKP-based verifiable ML: full-pipeline verification across training/testing/inference

Industry Reports and Data

23. Grand View Research / GlobeNewswire -- Data broker market $291-313B (2025), marketplace $1.81-1.86B
24. Grand View / Straits Research -- AI training datasets $2.82-3.2B (2024) → $7.23-8.6B (2030)
25. MarketsandMarkets -- AI agent market $5.1B → $47.1B (2030), CAGR 44.8%
26. Research & Markets -- M2M services market $164.2B (2030), CAGR 25.5%
27. Messari -- DePIN FY25 on-chain revenue $72M, startup funding ~$1B (2025)
28. DePINscan -- 41.8M active devices
29. CoinGecko -- DePIN market cap ~$9-10B (current), peak $19.2B
30. Han et al. -- DAO total treasury $24.5B, 13,000+ DAOs
31. Vana official -- 300+ DataDAOs, 1.3M users, 12.7M data points
32. DappRadar -- AI Agent crypto 550+, combined market cap $4.34B
33. TechCrunch -- Reddit AI licensing cumulative $203M
34. CNIL France -- EU consumer willingness-to-accept for personal data €120-360/year
35. KDATA (Korea Data Agency) -- South Korea data industry KRW 27.15T (2023) → KRW 30.75T (2024 est.)
36. Financial Services Commission -- MyData 2.0: 165M records linked, KakaoPay 20M subscribers
37. IAPP -- 19 US state privacy laws enforced, no federal law
38. Gartner -- 25% of enterprises to use Web3 services by 2027
39. EU / Clifford Chance -- EU Data Act fines at 4-5% of revenue
40. House of Chimera -- 650+ active DePIN projects
41. Precedence / Mordor -- Enterprise data monetization $4.78-5.22B (2025) → $48.55B (2035)
42. InsightAce -- Web3 blockchain market $2.7B → $114.9B (2034), CAGR 45.6%