Ten Years In, Blockchain Finance Research Is Still Mostly About Plumbing

A new structural topic model of 2,401 blockchain-finance papers landed in Frontiers in Blockchain on February 10, 2026. The authors - Sanghvi, Ubarhande, Chandani and a team across Symbiosis, Jaipuria, IMI-Kolkata and Integral - ran the entire Scopus corpus from 2016-2025 through Python-based STM. The result is an uncomfortable map of where academic attention actually goes.

Roughly 60% of the literature is about blockchain technology itself - security, infrastructure, smart contracts. Only about 30% engages actual financial applications. A single topic - "Blockchain, Technology, Security" - captured 35% of the dataset by itself.

Ten years of papers. And the field is still mostly studying how the pipes work.

What the model surfaced

Fifteen latent topics. Five thematic clusters. The dominance ranking matters more than the topic count.

After the technology-and-security cluster, the next biggest contributors are Topics 2 and 13 - both broad blockchain-and-data themes - covering 15-20% of the corpus. DeFi, lending, financial research, and risk modeling sit in the mid-tier. The smallest topics, each under 5% of the corpus: voting, electricity-and-attacks, Hyperledger-and-law, and peer-identity-buyer mechanisms. Niche, sparsely studied, mostly orphaned.

The correlation matrix tells the same story. Topic 6 (technology/security) correlates with Topic 13 (smart/data) at 0.66 and with Topic 11 (data/IoT) at 0.62. The technology cluster talks to itself. Finance-specific topics correlate less densely with each other.

The five clusters

The authors group everything into five buckets: Blockchain + AI/Data Intelligence; Blockchain + Supply Chain Financing; Blockchain for Governance, Identity and Regulation; Blockchain Infrastructure/Risk/Smart Systems; and Blockchain in Finance and Markets.

Look at what the actual "applied finance" cluster contains - DeFi, lending platforms, fintech research. It's the fifth cluster, built from just three of the fifteen topics. The largest cluster by far is infrastructure-and-risk, pulling in five topics including the dominant Topic 6.

The 60/30/10 ratio holds at the cluster level. Most blockchain-in-finance research isn't really about finance.

Why this is awkward

The application layer keeps moving. DeFi protocols, tokenization, on-chain credit - these aren't theoretical anymore. The authors acknowledge this directly: blockchain research is "increasingly emphasising financial applications" even as foundational topics swallow the corpus.

Two readings of this gap. One: publication incentives reward methodological novelty, which favors cryptographic and systems contributions over messy applied-finance work that's hard to peer-review against fast-moving protocols. Two: applied blockchain finance moves faster than journal pipelines can track, so by the time an empirical study clears review, the protocol it studied has forked, migrated to L2, or collapsed.

Both can be true.

What's actually missing

The authors propose research questions across all five clusters. The interesting gaps are regulatory and governance. DAO legal personhood. Layer-2 scalability under real load. Supervisory technology that uses blockchain transparency for real-time systemic-risk monitoring. Regulatory sandboxes that don't require a year of paperwork.

Almost none of this gets serious treatment in the existing 2,401 papers. Topic 8 - Hyperledger, fabric, law - has the highest semantic coherence in the entire model (0.58) but appears in less than 5% of the corpus. The work that is most internally consistent is also the least pursued.

The authors are blunt about it: studies blending financial, technological, and regulatory aspects are "limited." They flag this as a research-gap, but it reads more like an indictment of the field's incentives.

What's left unresolved

If 60% of a decade's research went into infrastructure, what does it mean that blockchain finance still struggles with adoption, regulation, and institutional trust? Either the technical foundations aren't the actual bottleneck - and the field has been studying the wrong question for ten years - or they are the bottleneck, and the work has not yet solved them despite the volume.

The authors don't pick. Their data doesn't either.

Sources: Sanghvi, Ubarhande, Chandani, Pathak, Wagholikar, Bagade, and Atiq, "Trends in blockchain in finance: unveiling latent research topics using a structural topic modelling approach," Frontiers in Blockchain (February 10, 2026); Roberts et al. on STM methodology (2014, 2019); Röder et al. on coherence measures (2015); Patel et al. bibliometric review (2022); Weerawarna et al. content analysis (2023).