Andreessen Horowitz’s crypto unit said the common industry story about DeFi and traditional finance converging into a clean hybrid system is largely the wrong frame.
Its argument is simpler: if blockchain helps incumbents run existing businesses better, they will use it. They are not doing that because they have signed on to decentralization. They are doing it because the economics work. The technology can cut costs, improve settlement, widen distribution, and tighten control over customer relationships.
Under that view, institutions are not “merging” with DeFi. They are selecting the parts of DeFi that fit their operating constraints, stripping out the parts that do not, and rebuilding the stack around institutional requirements. What emerges, according to a16z, will look neither like legacy finance nor like today’s DeFi. It describes that category as programmable financial infrastructure running on blockchain rails but optimized for institutional constraints.
What institutions actually adopt
a16z said a blockchain component usually needs to satisfy two tests before a traditional financial institution will use it. First, it must improve cost, risk, or distribution. Second, it cannot undermine control and accountability. Features commonly associated with crypto-native systems, including open access, anonymity, and immutable execution, may clear the first test but often fail the second.
That makes institutional adoption predictable rather than random. The firm said founders can treat this as a design test. If a product creates value only by reducing institutional control, it is likely to be reworked or rejected, no matter how elegant the design may be.
The article walked through several examples. Atomic settlement removes the delay between trade execution and final settlement, cutting counterparty risk and freeing collateral that would otherwise be tied up against unsettled trades. Shared ledgers can reduce reconciliation, one of the larger hidden costs in financial back offices. Programmable money can automate coupon payments, margin calls, and corporate actions through code rather than long chains of manual instructions. Automated market maker, or AMM, curve mathematics can also be repurposed as a pricing engine for onchain foreign exchange and tokenized money market fund net asset values once the permissionless wrapper is removed.
a16z’s point was that each of these components can improve a line item on a profit-and-loss statement or remove an operational risk and its cost. None of them requires an institution to believe in decentralization.
Using blockchain for familiar financial functions
The firm said JPMorgan’s permissioned chain for institutional deposits and tokenized money market funds from BlackRock and Franklin Templeton should not be read as enterprises experimenting with DeFi. In its telling, these projects are still handling familiar functions such as interbank payments and settlement, fund subscription management, and yield-bearing product distribution. The difference is the pipeline underneath them.
Those deployments use blockchain’s technical properties, including programmability, transparency, and atomic settlement. At the same time, they deliberately leave out the properties that allow native DeFi to function, such as open access, anonymity, and trustless execution. a16z said that is not a failure and not a compromise. It is a deliberate architectural choice that shows where the market is moving.
Institutional buyers follow different rules
The article also pushed back on the idea that institutional adoption is simply a larger distribution channel for existing DeFi infrastructure. Institutions evaluate protocols differently from crypto-native users. For them, this looks more like selecting software vendors and infrastructure partners. They weigh operational risk, compliance controls, and the long-term ownership of critical systems through standard enterprise processes.
That means success in DeFi does not automatically translate into success in institutional markets. Companies rarely buy the best technology in isolation, a16z said. They buy the technology that best fits existing workflows, risk models, procurement processes, and internal constraints.
The piece compared that pattern with earlier technology cycles. The internet was reshaped inside enterprises through firewalls and intranets. Cloud computing went through private clouds, VPCs, and FedRAMP certification. Artificial intelligence is now being adapted through on-premises deployment, data residency requirements, and model governance. Blockchain, it said, will go through the same kind of reshaping in institutional settings.
Compliance and enterprise value delivery
a16z said the reshaping happens along two main axes.
The first is compliance. Know-your-customer rules, anti-money laundering checks, sanctions screening, investor accreditation, and regulatory reporting are not optional for most institutions. Permissionless systems do not naturally support those requirements. Institutions often need the ability to freeze assets, reverse transactions, and identify counterparties. DeFi was not originally designed around these needs, and meeting them often means substantial architectural changes.
The article added that this could change over time as regulatory frameworks mature. It cited legislation such as the CLARITY Act as an example that may eventually make it easier for institutions to access permissionless systems while still meeting regulatory requirements. For now, though, institutions still judge blockchain infrastructure through the lens of control, accountability, and operational risk.
The second axis is enterprise value delivery. Institutions are not adopting blockchain because they are committed to permissionless principles. They are adopting it when it reduces costs, lowers reconciliation friction, opens new distribution channels, or lets them embed themselves more deeply in customer relationships. If the value proposition is not expressed in those terms, it may not even survive procurement.
Stablecoins were presented as the clearest example. Banks, payment companies, and fintech groups are increasingly treating them as useful settlement infrastructure because they allow dollars to move faster across networks and regions. Very few of those firms, the article said, are embracing the ideology of permissionless finance. They are adopting programmable dollars because they are useful, not because they want to rebuild finance around DeFi principles.
a16z used Circle to illustrate that shift. It said Circle’s Arc network shows how blockchain infrastructure is being packaged for institutional buyers: with an emphasis on compliance, operational control, trusted counterparties, and integration with existing workflows, rather than permissionless access and composability. The product being sold is faster settlement, global reach, and better capital efficiency, delivered in a form institutions can actually use.
It made a similar point about SWIFT. The organization’s work on interoperability for tokenized assets is not aimed at replacing existing financial institutions, according to a16z. It is meant to help those institutions coordinate more effectively through the SWIFT network. The pattern repeats across the market: blockchain adoption is reinforcing existing financial networks rather than replacing them.
Two opportunities, not one
a16z said the industry is looking at two separate opportunities.
The first is helping institutions adopt the infrastructure they are ready to use now. Each time an institution adopts a component, whether that is atomic settlement, programmable money, or tokenized collateral, it validates the technology, improves the shared rails, and brings real transaction volume and capital onchain.
The second is continuing to build open, crypto-native financial systems that institutions are not ready to use yet.
Those paths do not exclude each other. Open networks can keep producing new components, markets, and forms of innovation. Institutions may eventually take those outputs and use them. If both sides succeed, convergence may happen on its own, not because one side absorbs the other, but because both come to depend on the same underlying infrastructure.
Why founders need to choose their market
At the ecosystem level, institutional adoption and open networks can strengthen each other. At the company level, a16z said, trying to do both at once is usually a mistake. Selling to institutions requires knowledge of procurement, compliance, internal controls, channel partners, and long enterprise sales cycles. Building for open networks means optimizing for developers, liquidity, composability, and network effects.
The customer, distribution model, product requirements, and measures of success are often completely different. The point is not that one opportunity is better than the other. The point is that founders need to decide which market they are serving, while recognizing that both are linked by the same underlying public blockchain settlement rails.
The article said working with institutions and building a parallel financial system do not conflict. If both work, they can amplify each other. Permissioned layers can bring volume, legitimacy, and capital. Open layers can keep producing the components that permissioned systems may adopt next. If convergence happens, a16z said, it will happen at the rail level rather than through one side surrendering to the other.
Build from scratch or adapt existing products
The firm outlined two ways to build for this new category of programmable financial infrastructure: start from scratch or adapt products that already exist.
It pointed to networks such as Canton as an example of the first path. Rather than adapting existing DeFi infrastructure, Canton was designed around institutional requirements for privacy, compliance, and controlled interoperability from the start. The goal is not to pull banks into DeFi, but to use blockchain-based coordination while preserving the governance, confidentiality, and operational controls institutions require.
The second path looks different. a16z cited Morpho as a case where a team did not abandon its DeFi components, but instead focused on making them easier for institutions and asset issuers to use. It gave the example of Apollo’s ACRED fund, which incorporated Morpho into its onchain lending strategy, combining a DeFi-native lending component with institutional-grade distribution, compliance, and fund structure.
a16z said the end state will be neither pure DeFi nor a fully isolated institutional stack. It will be a model in which institutions selectively adopt existing crypto infrastructure and repackage it around their own requirements for control, compliance, and distribution. That category is built for institutional constraints. It draws from DeFi, but it operates in a more permissioned and compliant form.
The article added that teams like Morpho, which have successfully adapted crypto-native infrastructure for institutional use cases, do exist. But founders should not treat that path as the default. Institutions are a distinct customer segment with distinct needs. In many cases, designing around those needs from day one may be more effective than retrofitting products that were built for open networks.
Open networks remain the main source of innovation
a16z argued that many of the innovations institutions are adopting today did not come from banks, asset managers, or incumbent financial infrastructure. They came from open networks, where founders could experiment freely with new market structures, coordination mechanisms, and financial components.
That difference matters, the article said. Institutions are not the sector’s main engine of innovation. Permissioned layers often sit downstream from open ones. That leads to a broader strategic point: if the entire industry focuses on selling to banks and asset managers, it risks mistaking one large customer segment for the whole opportunity. TradFi is an important customer base, but not the only one.
a16z said building for institutional needs is legitimate and valuable, but it is one lane rather than the whole road. Durable companies tend to be the ones that stay clear about who they are building for.
Its conclusion was that institutional adoption may become a large opportunity, but it is not a simple extension of DeFi. Success in one market does not guarantee success in the other. Founders building for institutions should fully commit to understanding customers, procurement, and institution-specific product design. Founders building for open networks should keep going rather than abandoning their thesis because institutions are the loudest buyers in the market right now.
The two paths are complementary, a16z said. One commercializes, adapts, and scales innovations that have already been proven. The other discovers those innovations in the first place. TradFi is not adopting DeFi in full. It is selectively adopting the parts that fit its own model. The real opportunity for builders is not to chase every market at once, but to decide which market they are building for and execute accordingly.