All posts
How DeFi works now: 2026 tech stack explained
Gas-free transactions, restaking, and smart accounts have quietly rewritten how DeFi behaves. Here's what actually shifted between defi 2025 and 2026 — and which changes you'll feel using protocols today.
Insights
Numbers
Proven performance
TL;DR
Key takeaways
DeFi now runs as five layers: settlement, execution, bridging, smart-account UX, and AI automation
Paymasters cover gas for you, so holding native tokens on every chain is no longer required
Blob transactions (EIP-4844) cut L2 fees by 80–95%, making micro-payments and frequent trades viable
For USDC, prefer Circle's burn-and-mint CCTP over wrapped bridges to avoid extra counterparty risk
Bridge hacks drained $2.5B+ from 2021–2024, so check audits and liquidity before fees on big transfers
17 minute reading
Insights
Why DeFi's architecture had to change
So, what is DeFi in practical terms? It's on-chain finance where smart contracts replace intermediaries and users control funds via a DeFi wallet.
In 2025, DeFi is not a single chain or application category; it is a layered technical stack spanning settlement, execution, interoperability, and user experience layers simultaneously.
The 2025 technical stack has five identifiable layers:
Settlement layer: Ethereum, Solana, Bitcoin, and Cosmos chains hold canonical asset state and provide cryptographic finality guarantees
Execution layer: ZK rollups (zkSync Era, Scroll, Linea) and optimistic rollups (Arbitrum, Base, OP Mainnet) process transactions at low cost while inheriting L1 security
Coordination layer: Cross-chain messaging protocols (LayerZero, Wormhole, Chainlink CCIP, Hyperlane) relay asset and data proofs across chains, enabling cross-chain swaps and omnichain token standards
UX layer: Account abstraction wallets with paymaster gas sponsorship, session keys, and intent solvers that translate user goals into optimized on-chain transactions
Automation layer: On-chain AI execution agents that monitor positions and execute strategies without manual approval per transaction
Total value locked across all chains exceeded $150 billion in 2025 (source: DeFiLlama), driven by restaking growth, stablecoin expansion past $300 billion in circulating supply (source: DeFiLlama Stablecoins), and institutional DeFi adoption from firms previously operating exclusively in traditional finance.
The phrase "DeFi roadmap 2025" now refers to a converging set of technical standards — ERC-4337, EIP-7702 (which lets EOAs temporarily behave like smart accounts), cross-chain messaging, and ZK proof verification — not individual protocol feature releases.
Why DeFi's architecture needed to change
DeFi's architecture needed to change because multi-chain growth made liquidity, UX, and security costs scale faster than user value.
The five structural pressures that forced architectural change:
1. Liquidity fragmentation: Deploying the same protocol across 20+ chains split TVL into isolated pools, worsened slippage on large trades, and made cross-chain arbitrage inefficient.
2. DeFi wallet complexity: Requiring users to hold native gas tokens on each chain created a hard onboarding wall. Definition (Paymaster): A paymaster is an ERC-4337 contract that sponsors gas fees for a user's smart account, removing the "need ETH for gas" requirement for most user flows.
3. Settlement cost: Pre-EIP-4844, L2 data posting costs made micro-transactions economically unviable. Blob transactions dropped L2 fees by 80–95% (source: L2Beat), unlocking DeFi-native payments and high-frequency strategy execution.
4. Regulatory pressure: Separating execution, settlement, and custody layers gave protocols compliance flexibility without abandoning non-custodial guarantees.
5. AI integration readiness: AI liquidity management became viable only after on-chain execution costs dropped and structured oracle data feeds became reliable enough to support automated decision-making at scale.
The combined result: protocols that couldn't abstract complexity away from users lost TVL to those that could.
How DeFi's core technical stack works in 2025
DeFi's 2025 stack is a five-layer architecture where each layer handles a specific function and can be upgraded independently:
1Layer 1 — Settlement: Ethereum, Solana, Bitcoin, and Cosmos chains hold canonical asset state. No DeFi operation is economically final until it settles here.
Layer 2 — Execution: ZK rollups and optimistic rollups process transactions at low cost with Ethereum security inheritance.
Layer 3 — Interoperability: Cross-chain messaging protocols including LayerZero, Wormhole, Chainlink CCIP, and Hyperlane relay asset and data proofs between chains.
Layer 4 — Account/Intent: ERC-4337 smart accounts with session keys and paymasters allow users to express goals — "swap 1,000 USDC for ETH at the best rate across all L2s" — and solver networks compete to fulfill those intents optimally.
Layer 5 — Automation: On-chain AI execution agents monitor positions, execute rebalancing, and manage risk without requiring manual approval for each transaction.
Crypto staking platforms sit across layers 1 and 3 — staking occurs at the settlement layer, while liquid staking tokens (LSTs) and liquid restaking tokens (LRTs) circulate across chains via Layer 3 messaging. Lido issues stETH on Ethereum mainnet, and that token is actively used across Arbitrum, Base, and other L2s through bridging mechanisms.
Therefore, DeFi in 2025 works by settling security on L1, executing cheaply on L2, moving value via interoperability layers, and abstracting user actions through smart accounts and automation.
Omnichain stablecoins: best cross-chain USDC and USDT bridges
Choosing a bridge chain route (source chain → destination chain) matters most for stablecoins because liquidity and finality differ by network. Native USDC cross-chain transfers via Circle's Cross-Chain Transfer Protocol (CCTP) became the institutional standard in 2025, reducing counterparty risk relative to wrapped USDC bridge models by burning on source and minting on destination with cryptographic proof verification.
Three practical options for stablecoin bridging:
CCTP (burn/mint) for institutional USDC transfers — lowest custodial risk, no wrapped token exposure
Across/Stargate (liquidity network) for fast L2↔L2 stable moves — optimized for speed and low slippage
Aggregators (Bungee/Rubic) for route optimization across multiple protocols simultaneously
For non-stable routes, a cross chain swap aggregator can combine DEX routing plus bridging into one quote.
Over $2.5 billion was lost to bridge exploits between 2021 and 2024 (source: Immunefi), making security criteria more important than fee optimization for large transfers. Key risk factors for stablecoin bridge selection: smart contract audit history, insurance coverage, oracle manipulation resistance, and liquidity depth.
A detailed comparison of best cross chain bridge options for USDC/USDT — including fees and security models — is covered in our dedicated guide.
Best MCP servers for DeFi: AI agents and on-chain execution
Definition (MCP): Model Context Protocol is an open standard that connects AI models (GPT-4o, Claude, Gemini) to external tools and APIs. In DeFi, MCP tool servers act as middleware — translating natural language goals into structured on-chain transactions with gas estimation, slippage tolerance, and route optimization built in.
Active production use cases include flash loan arbitrage AI (agents identify multi-hop opportunities and execute complete loan-trade-repay cycles within a single block) and AI risk management DeFi (monitoring collateral ratios on lending protocols and triggering protective actions before liquidation thresholds are breached). Portfolio rebalancing across chains and oracle verification against multiple price feeds are also common.
Session key scoping limits agents to specific tokens, maximum amounts, and time windows. MCP-enabled agents with broad wallet permissions introduce novel attack vectors: malicious tool servers injecting false price data, prompt injection attacks redirecting agent execution, and automated strategies that can destabilize thin liquidity markets. Transaction simulation before execution is an essential safeguard.
Protocol-level comparisons of MCP definition and best MCP servers for DeFi are detailed in our dedicated guide.
Cross-chain interoperability: how it works in DeFi
A DeFi bridge (a crypto bridge) and bridge aggregators are the infrastructure that moves assets and messages between chains for swaps and liquidity routing. Bridge security remains the primary selection criterion — exploits have drained billions from the ecosystem, and the attack surface expands with each additional chain integration.
In 2025, four canonical models handle the majority of cross-chain swap volume:
Model | How It Works | Example Protocols | Key Risk |
|---|---|---|---|
Lock-and-mint | Asset locked on source, wrapped token minted on destination | Legacy WBTC, early bridges | Custodial risk at lock contract |
Liquidity network | Relayers front destination liquidity, reimbursed from source pools | Across, Stargate | Relayer liquidity depth |
Native burn-and-mint | Asset burned on source, cryptographic proof minted on destination | Circle CCTP, LayerZero OFT | Smart contract correctness |
Cross-chain swap (DEX/intent-based) | Token A on Chain X swapped directly for Token B on Chain Y | THORChain, deBridge | Routing complexity, slippage |
Chain abstraction layer — emerging from protocols like Particle Network, NEAR chain signatures, and ERC-7683 — extends this further: users control assets across all chains from a single account interface and never manually select a network.
A technical breakdown of interoperability protocols, messaging standards, and security models is covered in our dedicated guide: Cross-Chain Interoperability: How It Works in DeFi.
Best DeFi restaking protocols: yields and risks compared
Definition (Liquid staking): Liquid staking is the process of staking a proof-of-stake asset and receiving a liquid token (stETH from Lido, rETH from Rocket Pool) that represents the staked position and accrues rewards — enabling staked assets to remain usable across DeFi while still earning base staking yield.
Definition (Restaking): Restaking means using already-staked assets or LSTs as security collateral for Actively Validated Services (AVSs) — oracle networks, bridges, and data availability layers — each paying additional rewards on top of base staking yield. EigenLayer pioneered this on Ethereum; Symbiotic and Karak extended it to other assets.
Crypto staking platforms comparison (2025):
Protocol | Type | Asset | Notable Feature | Key Risk | Best Fit |
|---|---|---|---|---|---|
Lido | Liquid staking | ETH | ~30% of all staked ETH, stETH widely accepted | Centralization, governance | Broad DeFi collateral use |
EigenLayer | Restaking | ETH/LSTs | Largest restaking TVL, AVS marketplace | Slashing cascades | Developers building AVSs |
Symbiotic | Multi-asset restaking | ETH + others | Accepts non-ETH collateral, modular AVS | Newer, less battle-tested | Multi-asset restaking strategies |
Karak | L2-native restaking | Multiple | L2-first architecture | Liquidity depth | L2-native applications |
Pendle Finance | Yield tokenization | LSTs/LRTs | Fixed/variable yield splitting | Complex position management | Fixed-rate yield strategies |
Babylon | Bitcoin staking | Native BTC | BTC secures PoS chains without bridging | Bitcoin scripting limits | BTC holders avoiding bridges |
Rule of thumb: If you can't model slashing and LRT liquidity risk, treat restaking yield as risk-premium, not "extra APR."
Liquid restaking tokens — eETH (EtherFi), ezETH (Renzo), pzETH (Swell) — trade on secondary markets at variable liquidity premiums. During market stress, LRT secondary market prices adjust faster than redemption queues process, creating temporary depegs that can trigger cascading liquidations in protocols accepting LRTs as collateral.
Yield figures, slashing histories, and protocol-level risk comparisons are detailed in our dedicated guide: Best DeFi Restaking Protocols: Yields and Risks Compared.
Intent-based architectures: the next wave of DeFi
Intent systems are increasingly built directly into the DeFi wallet UX. Intent-based architectures replace the transaction construction model with a goal specification model where users declare desired outcomes and solver networks compete to fulfill them optimally.
What is account abstraction? In DeFi, it means wallets become programmable smart accounts (ERC-4337) that can sponsor gas and automate execution. Definition (Account abstraction): Account abstraction (ERC-4337) replaces standard externally owned accounts (EOAs) with programmable smart contract accounts — enabling gas sponsorship via paymasters, session keys for automated execution, and social recovery for lost credentials. EIP-7702 extends this to existing EOAs without requiring account migration.
Key components of the intent-based stack:
Paymaster gas sponsorship: dApps pay transaction gas on behalf of users in any token, removing the ETH-for-gas requirement
Solver networks: CoW Protocol, UniswapX, and 1inch Fusion run competitive solver auctions where users receive guaranteed output amounts with MEV protection built in
Session keys: Pre-authorized execution permissions allowing automated agents to act within defined parameters — specific tokens, maximum amounts, time windows
Social recovery and batched transactions: Smart account recovery via trusted contacts eliminates seed phrase loss as a catastrophic failure mode; multiple operations execute atomically in a single user action
Intent-based transactions reduce on-chain footprint: a cross-chain swap that previously required 4–6 separate transactions collapses into a single settlement transaction, lowering fees and MEV exposure simultaneously.
Protocol comparisons, solver economics, and implementation patterns are covered in our dedicated guide: Intent-Based Architectures: The Next Wave of DeFi.
BTCFi: building Bitcoin-native DeFi applications
Bitcoin-native DeFi moved from theoretical to operational in 2025. The core challenge: Bitcoin's UTXO scripting model was never designed for complex smart contract logic, so BTCFi protocols must work around fundamental architectural constraints.
Two primary execution models emerged:
Stacks / RGB protocol: Smart contract and client-side validation layers built above Bitcoin. Stacks settles transactions on Bitcoin mainchain via proof-of-transfer consensus, while RGB runs contract execution off-chain with Bitcoin commitments as anchors. Both approaches inherit Bitcoin's settlement security without modifying the base protocol.
Babylon: Bitcoin staking protocol allowing BTC holders to secure proof-of-stake chains via Bitcoin-native timestamping. BTC never leaves the Bitcoin network — instead, stakers commit time-locked UTXOs that can be slashed if validation duties are violated. This makes Babylon the only production restaking protocol that avoids cross-chain bridge exposure entirely.
The practical tradeoff: Stacks and RGB offer programmability but add execution complexity. Babylon offers yield with minimal bridge risk but limited DeFi composability — staked BTC can't simultaneously serve as collateral in lending protocols the way stETH can on Ethereum.
Current BTCFi TVL remains small relative to Ethereum DeFi (~$3–5 billion vs. $80+ billion), but institutional interest is high because Bitcoin holders represent the largest pool of unstaked crypto capital. The key structural limitation is that most BTCFi yield beyond Babylon ultimately depends on cross-chain bridge exposure or off-chain coordination, retaining some custodial risk even when marketed as "native Bitcoin" products.
Protocol architecture and risk models are detailed in our dedicated guide: BTCFi: Building Bitcoin-Native DeFi Applications.
How to integrate cross-chain liquidity on L2s: SDK comparison
To integrate cross-chain liquidity on L2s, apps typically use third-party bridge/DEX routing SDKs that handle quoting, route selection, and transaction building across chains.
Top SDKs for cross-chain liquidity integration:
SDK | Underlying Protocol | Chain Coverage | Best For |
|---|---|---|---|
Li.Fi SDK | Multi-protocol aggregation | 20+ chains, 30+ bridges | Apps needing maximum route optionality |
Socket Protocol | Intent-based routing | 10+ chains | Developer-friendly integration |
Across SDK | Optimistic bridge | Ethereum + major L2s | Stablecoin transfer applications |
Ethereum upgrades on the 2026 roadmap — specifically the Fusaka upgrade, which introduces PeerDAS for further data availability scaling — will affect staking economics and L2 data availability capacity. Developers integrating LST returns into L2 applications should monitor Ethereum upgrade timelines, as changes to validator economics directly affect stETH yield rates and LRT pricing.
Detailed SDK benchmarks, integration code samples, and fee comparisons are covered in our dedicated guide: How to Integrate Cross-Chain Liquidity on L2s: SDK Comparison.
Real-world implementations: how protocols deploy these technologies
Aave v4's unified liquidity layer is the most architecturally significant protocol upgrade of 2025: single pool contracts accessible from any supported chain allow users to borrow on Arbitrum against collateral posted on Ethereum mainnet — without any manual bridging step.
Uniswap v4 hooks allow developers to attach custom logic — dynamic fee curves, TWAP automation, on-chain limit orders — directly to liquidity pool contracts, transforming the AMM into a platform for product differentiation across EVM chains.
Lido remains the foundational liquid staking base layer, distributing stETH across Ethereum and L2s via cross-chain messaging. EigenLayer builds on top as the AVS restaking marketplace, while deBridge provides cross-chain swap infrastructure across 20+ EVM and non-EVM chains with intent routing.
Gauntlet and Chaos Labs deployed AI risk management models that adjust borrow caps, collateral factors, and oracle parameters on Aave and Compound via on-chain governance proposals — reducing governance response latency from weeks to hours during market stress events.
Limitations and risks of the 2025 DeFi stack
Higher architectural sophistication increases attack surface area. The 2025 DeFi stack introduces new failure modes at every layer it adds.
Cross-chain bridge risk: Bridge exploits remain the single largest source of DeFi losses historically (source: Immunefi). The attack surface expands with each additional chain integration, and risk is highest when liquidity is thin or contracts are unaudited.
Restaking slashing cascades: If a major AVS is slashed, all restakers providing security to that AVS lose a proportional percentage of staked capital. LRT holders face additional depegging risk as secondary market prices react faster than redemption queues can process withdrawals.
AI agent risk: MCP-enabled agents with broad wallet permissions introduce novel attack vectors: malicious tool servers injecting false price data, prompt injection attacks redirecting agent execution, and automated strategies that can destabilize thin liquidity markets.
Account abstraction risks: Paymaster contracts drained by high gas usage can block pending transactions; session keys with overly broad scopes create single points of compromise; smart account upgrade mechanisms governed by admin keys reintroduce centralization.
Highest APY crypto staking strategies: Advertised rates of 20–50%+ typically combine base staking yield, restaking rewards, liquidity mining incentives, and leveraged recursive positions — each layer compounds underlying risk, not just returns.
Regulatory risk: Cross-chain protocols operating across jurisdictions face inconsistent AML/KYC requirements. Chain abstraction layers that obscure asset provenance may conflict with FATF travel rule compliance requirements.
Future outlook: DeFi's technical direction through 2026
The defining theme of DeFi's 2025–2026 technical roadmap is abstraction: of chains, of gas, of transaction construction, and increasingly of strategy execution itself.
Chain abstraction completion: Particle Network, NEAR chain signatures, and ERC-7683 are targeting a UX where users interact with "DeFi" rather than named networks. What is chain abstraction crypto? It is the point at which the user's mental model collapses to "balance" and "action" — the underlying chain becomes an implementation detail.
Ethereum upgrades 2026: The Fusaka upgrade introduces PeerDAS for further data availability scaling and EIP-7742 for uncoupled blob counts — enabling L2 transaction costs to drop further and supporting higher throughput DeFi applications. (See: Ethereum Foundation blog on PeerDAS and EIP-7742)
AI governance and restaking evolution: On-chain AI execution frameworks are moving toward semi-autonomous operation within pre-approved risk bounds, reducing governance latency from weeks to hours. Competition between EigenLayer, Symbiotic, and Karak will drive AVS economics toward equilibrium yields. The distinction between liquid staking vs. restaking will blur further as protocols stack yield layers automatically based on user risk tolerance settings.
The convergence point is a financial system where cross-chain execution, gas management, and yield optimization are handled entirely at the infrastructure layer — leaving users with a bank-equivalent interface, backed by non-custodial, auditable, permissionless smart contracts.
Lead Growth Product Manager
FAQs
Got questions?
Still have questions? Contact us and we’ll help you out.
01
What are the DeFi trends for 2026?
DeFi in 2026 is defined by a converging set of technical standards rather than individual protocol releases — ERC-4337 smart accounts, EIP-7702 (which lets regular wallets temporarily behave like smart accounts), cross-chain messaging, and ZK proof verification. The industry is also shifting from raw TVL growth toward productive, capital-efficient liquidity concentrated in fewer, deeper venues. On-chain AI execution agents and intent-based swaps round out the picture.
02
What is the DeFi stack and what are its layers?
The 2025–2026 DeFi stack has five identifiable layers: a settlement layer (Ethereum, Solana, Bitcoin, Cosmos) that holds canonical asset state, an execution layer (ZK and optimistic rollups), an interoperability layer (cross-chain messaging like LayerZero and CCIP), an account/intent layer (ERC-4337 smart accounts), and an automation layer (on-chain AI agents). Each layer handles a specific function and can be upgraded independently.
03
What technology does DeFi use?
DeFi runs on permissionless blockchains where smart contracts replace intermediaries, and users control funds through non-custodial wallets that sign transactions directly. The financial logic lives in smart contracts on-chain, while rollups handle cheap execution and cross-chain messaging protocols move value between networks. No bank account, approval, or counterparty trust is required in typical non-custodial flows.
04
How is DeFi different from traditional finance?
DeFi is a permissionless system where you control assets via non-custodial wallets instead of banks or brokers — no account approval, no withdrawal limits, and no counterparty trust in standard flows. It removes many middlemen, which lowers discretionary fees and enables 24/7 settlement. The tradeoff is that you still pay blockchain network fees, which can rise during congestion.
05
How do ZK rollups and optimistic rollups work in DeFi?
ZK rollups (zkSync Era, Scroll, Linea) and optimistic rollups (Arbitrum, Base, OP Mainnet) process transactions off-chain at low cost while inheriting the security of their underlying L1. After EIP-4844 introduced blob transactions, L2 fees dropped 80–95% according to L2Beat, unlocking DeFi-native payments and high-frequency strategies. They periodically settle proofs back to Ethereum for finality.
06
What is account abstraction and how does it improve DeFi UX?
Account abstraction (ERC-4337) enables smart contract wallets with paymaster gas sponsorship, session keys, and intent solvers. A paymaster sponsors gas fees so users no longer need to hold native gas tokens on every chain, while intent solvers compete to fulfill goals like "swap 1,000 USDC for ETH at the best rate across all L2s." This abstracts away the complexity that previously created a hard onboarding wall.
07
What is the best way to bridge USDC and USDT across chains?
For native USDC, Circle's Cross-Chain Transfer Protocol (CCTP) became the institutional standard in 2025 by burning on the source chain and minting on the destination with cryptographic proof — avoiding wrapped-token risk. Liquidity networks like Across and Stargate are optimized for fast, low-slippage L2↔L2 stable moves, while aggregators like Bungee and Rubic optimize routes across protocols. With over $2.5 billion lost to bridge exploits between 2021 and 2024, security matters more than fees for large transfers.
08
What is total value locked (TVL) in DeFi and why does capital efficiency matter now?
Total value locked across all chains exceeded $150 billion in 2025, driven by restaking, stablecoin supply past $300 billion, and institutional adoption. In 2026 the focus is shifting from raw TVL toward productive TVL — concentrating capital in fewer, deeper venues rather than many shallow pools. One analysis estimated over $12 billion in DeFi liquidity sits idle, highlighting the push for better capital recycling.
Learn more
Bridge guide
Ethereum 2026 upgrades & DeFi: the full breakdown
From Glamsterdam to the data-publication tweaks that quietly slashed L2 fees — here's what actually moved Ethereum DeFi this year, and where TVL and staking yields are heading next.
Read article
Best cross-chain bridge: 7 top crypto bridges compared
Bridge fees, speed, and security don't always move together — pick wrong and you overpay or get stuck. We break down which cross chain bridge actually fits your DeFi moves in 2026.
Read article
Best crypto bridges: cheapest cross-chain swaps
Bridging fees can quietly eat 5% of your transfer if you pick wrong. We break down how the top cross-chain bridges stack up on real costs, settlement times, and chain coverage so your next DeFi move isn't a gamble.
Read article
Swap crypto across 50+ networks
Non-custodial. No KYC. Connect your wallet and get started.
