H1 2026 Progress Report – Curve Ecosystem Grant

[michwill]

Grant Period: January 2026 to June 2026

At a Glance: Key Highlights & Action Items

The summary below highlights the period’s key achievements and the items requiring DAO attention. Full detail follows in the sections below.

Key Highlights (H1 2026)

• Llamalend V2 shipped: completed its ChainSecurity audit, launched in production on Optimism, and was deployed to Ethereum mainnet on 13 July, introducing LP token and PT collateral support and a new source of DAO admin-fee revenue.
• FXSwap matured: moved into production-led optimisation, using live YieldBasis and FX pilot performance to improve rebalancing, fee allocation and liquidity placement.
• FastBridge launched: cut crvUSD withdrawal times from L2s to Ethereum from around seven days to roughly 15 minutes.
• Phase 2 grant funding secured: 8,725,000 CRV plus a 2,500,000 CRV and 520,000 crvUSD top-up have been received, funding operations through 31 December 2026.

Action Items for the DAO

Protocol fee governance proposal: Swiss Stake AG will submit an on-chain governance proposal to increase the DAO’s protocol fee share from 10% to 30% (adjusting the fee_receiver allocation), per the Mid-Term Review commitment.

• Swiss Association (Verein): preparatory legal work is underway to establish the non-profit stewardship entity; formal establishment work resumes later in Q3.
• Security audit budget: total audit expenditure is being managed toward the CHF 900,000 to 1,000,000 target for 2026.

1. Executive Summary

Covering the period from January to June 2026, the first half of the Phase 2 grant period marked a significant step forward for Curve across multiple fronts. Swiss Stake AG focused on advancing core protocol development, completing key security audits, expanding cross-chain capabilities, and building the foundation for the next generation of Curve products.

The central technical milestone of H1 2026 was the completion of the Llamalend V2 audit and its first production deployment on Optimism. The codebase completed its external security audit with ChainSecurity during H1, with Ethereum mainnet contracts deployed shortly after the reporting period on 13 July. Llamalend V2 introduces LP token and PT collateral support, enables non-crvUSD lending markets with direct admin fees accruing to the Curve DAO, and represents the most significant product release in Curve’s lending roadmap to date.

FXSwap moved into a deeper phase of production-led optimisation during H1. Performance from live YieldBasis markets and FX pilots informed improvements to rebalancing and fee allocation, alongside further research into liquidity placement and external pricing. Cross-chain infrastructure expanded through the launch of FastBridge and work on a more resilient Blockhash Oracle design using LayerZero and Chainlink.

Business development activity focused on the Llamalend V2 market pipeline, supporting FXSwap adoption among prospective asset issuers, and providing technical assistance for third-party integrations across the Curve ecosystem. Separately funded activities outside the Curve DAO grant scope, including the Stellar implementation, are summarised in Annex A. The team made meaningful progress toward the structural commitments made to the DAO in the Mid-Term Review, including preparatory work for the Swiss Association and the protocol fee governance proposal.

We thank the Curve community for its continued engagement and oversight. H1 2026 has laid important groundwork and we look forward to delivering on the product roadmap in H2 2026.

2. Accomplishments and Progress

2.1 Core Protocol Smart Contract Developments

During H1 2026, core protocol development was concentrated around Llamalend V2 and FXSwap. Llamalend V2 progressed through external audit and into production on Optimism, while supporting work on the Gauge Factory, monetary policies and price-oracle infrastructure prepared the broader Ethereum deployment completed shortly after H1. FXSwap moved from early deployment into production-led optimisation, using live market performance, simulation and first-principles research to improve rebalancing, liquidity placement and fee allocation. In parallel, research expanded into AMM fee economics, LP outcomes, correlated lending risk and crvUSD peg dynamics, providing a stronger basis for future parameter and market design.

Llamalend V2

Llamalend V2 was the largest smart-contract development effort during H1 2026. The upgrade expands Llamalend from a system primarily built around crvUSD into a more flexible framework for isolated lending markets, while retaining Curve’s LLAMMA liquidation mechanism. H1 work took the protocol from a substantially refactored codebase through external audit and its first production deployment on Optimism, followed by preparation for Ethereum mainnet.

• A broader lending framework: V2 allows supported assets on both sides of a lending market, meaning crvUSD is no longer required as the borrowed asset. It also introduces support for Curve LP tokens, principal tokens and other suitable yield-bearing assets as collateral. Each market remains isolated and has its own oracle, monetary policy, parameters and DAO-controlled borrow cap. Markets also generate admin fees for Curve DAO, creating a new source of protocol revenue.

• Security review and production hardening: The V2 codebase completed a full external audit with ChainSecurity during H1. Findings were addressed before the production rollout. The team also had to complete and test several supporting components required to operate V2 markets safely, including factory infrastructure, price oracles and monetary policies.

• Initial Optimism rollout: Optimism served as the first production deployment, allowing the core V2 contracts and market architecture to be introduced through a controlled initial set of markets. The first markets covered ETH/wstETH, wstETH/USDC and WBTC/USDC. This provided an initial environment for observing the new contracts and user flows before the broader Ethereum rollout.

• Additional Ethereum infrastructure: Deploying on Ethereum was not simply a matter of redeploying the Optimism contracts. The team still needed to complete the Gauge Factory, additional price-oracle infrastructure and the monetary policies required for a broader range of markets. The Gauge Factory standardises how new lending-market gauges are created and connected to Curve’s incentive system. Price oracles provide the valuations needed by LLAMMA and market risk controls, while monetary policies determine how borrowing rates respond to market utilisation.

• Ethereum deployment after the reporting period: The Ethereum mainnet contracts were deployed on 13 July, shortly after the end of H1. Initial markets have been prepared and deployed.

• Immediate priorities: Near-term work includes preparing the next markets and designing an appropriate monetary policy for them. The team is also replacing the Semilog monetary policies used by the initial Optimism markets with Hyperbolic policies, alongside launching the first markets using Curve LP tokens as collateral.

• Further oracle research: The team plans to evaluate whether Chainlink price feeds can be integrated safely with LLAMMA’s range-based liquidation process. If the research is successful, the first Chainlink-based lending markets may follow. This could give future markets more flexibility in how their prices are sourced, but the interaction between external feeds and soft liquidation needs to be understood before deployment.

Beyond the near-term priorities, the team will evaluate additional directions for Llamalend V2, including the potential use of Chainlink SVR for hard liquidations, support for third-party market curators, and integrations with external protocols. These remain exploratory areas at this stage rather than committed roadmap items.

FXSwap development

During H1 2026, FXSwap moved from initial deployment into a deeper phase of production-led research and optimisation. Performance from live pools, particularly YieldBasis markets and FX pilots, provided practical evidence of how rebalancing, fee allocation and liquidity placement behave under real market conditions. These observations were fed back into simulation and backtesting, helping the team identify where the mechanism could be improved. Work focused on three closely connected questions: where the pool should concentrate liquidity, how quickly that concentration should move as prices change, and how the pool should fund those movements. The resulting work led to contract improvements and shaped the next phase of research into external pricing and more actively managed pool designs.

• Improved rebalancing and funding: Earlier work improved how FXSwap moves its liquidity concentration as prices change. A configurable reserve-profit fraction was subsequently added, allowing flexibility on the part of the fees earned by the pool that is allocated to its rebalancing budget. This reduces reliance on external donations or refueling, opening up for more autonomous FX pools. While allocating more fees to the budget leaves a smaller immediate share for LPs, stronger pool positioning may improve execution, attract more volume and produce better overall LP returns.

• Price-scale and dynamic-fee research: Research during Q2 increasingly showed that correct placement of price_scale, the centre around which liquidity is concentrated, is critical to pool performance. If the market price has moved but liquidity remains centred around an outdated level, traders receive worse execution and the pool uses its capital less efficiently. Dynamic fees remain important for protecting LPs against informed flow, but they need to be considered together with the oracle and liquidity movement rather than tuned as an isolated parameter.

• Lightweight hooks and propAMMs: An experimental lightweight hooks implementation was developed for CryptoSwap/FXSwap, allowing selected pool controls, particularly the dynamic fee and price_scale, to be driven through approved external logic. This creates a path for market makers or block builders to supply fresher price information before arbitrage occurs. A pool could then behave more like an informed propAMM, improving execution while reducing value lost through stale pricing. This remains early-stage work, with no live external market-maker integration yet.

• External price inputs: The team began evaluating how external prices could complement FXSwap’s internal oracle. This can be applied to both FX and Cypto use cases. Testing external feeds and improving how quickly fresh pricing reaches the pool will therefore be a major H2 research direction.

• Broader market support: The underlying architecture is not limited to foreign exchange. It could also support tokenised equities, commodities, indices and other real-world assets. These markets require further work around trading hours, primary and secondary price sources, keeper design and asset-specific recentering policies. The research and tooling developed for Bitcoin and FX provide a base for this expansion, but the same policy cannot simply be applied unchanged across every asset class.

• Simulation and production validation: The supporting research stack expanded substantially during H1, including large-scale backtesting, automated parameter optimisation and increased compute capacity. Simulated outcomes were compared against the performance of live pools, particularly YieldBasis markets, to test whether observed effects came from the mechanism itself or from specific market conditions. This created a feedback loop between production performance, research and subsequent contract improvements.

Oracles

H1 included continued work on more precise and reusable price infrastructure for Curve pools and lending markets.

• LP-token oracles: New oracle components were developed for StableSwap and FXSwap LP tokens. Reliable LP valuation is required for using these positions as collateral in Llamalend V2 and makes Curve LP tokens easier for external protocols to integrate.

• Aggregated crvUSD pricing: The team also developed an aggregated crvUSD price against the US dollar, providing a common price source for lending markets, crvUSD-related systems and third-party integrations.

Protocol Research

Alongside smart-contract development, the research team expanded its first-principles analysis of Curve’s AMM and lending mechanics. The objective is to understand why particular mechanisms work, when they may fail and how they should change across different market conditions. Historical backtesting remains important, but the research process also uses analytical models and independent simulations to distinguish genuine economic effects from simulator-specific behaviour.

• Dynamic fees and arbitrage: Research examined how pool fees should be set relative to short-horizon market movement and the time it takes arbitrageurs to react. Fees that are too low can expose LPs to informed flow, while fees that are too high can make Curve uncompetitive for ordinary users and aggregators. In a simplified constant-product benchmark, the team found that higher fees per arbitrage trade can be substantially offset by a lower frequency of arbitrage. As a result, arbitrage-related fee income can remain surprisingly stable across a broad range of fee levels.

• Liquidity placement and LP economics: The research also studied the interaction between the internal oracle, price_scale and dynamic fees. Moving liquidity too slowly leaves the pool quoting around an outdated market price, while moving too quickly can cause it to follow short-term noise and incur unnecessary rebalancing costs. This work directly informs the FXSwap and CryptoSwap research into better liquidity placement and recentering policies.

• A more complete view of LVR: In the same simplified benchmark, the team examined loss versus rebalancing, or LVR. Once the fees paid by arbitrageurs are included, the headline LVR figure can materially overstate the associated net loss to LPs. This does not mean LPing is risk-free, and the result does not yet represent the complete economics of Curve’s production invariants. It shows that LVR should not be interpreted as the full LP profit-and-loss calculation in isolation.

• Total execution cost: The work also found that lowering the headline pool fee does not always reduce the user’s total execution cost proportionally. When a transaction is visible before execution, lower fees can also make it cheaper to extract value around the trade. Part of the apparent savings may return as additional slippage or MEV. This supports optimising for total execution quality rather than the displayed pool fee alone.

• Lending and crvUSD risk: A new research direction began during Q2 following LlamaRisk’s departure. Initial work includes factor models designed to identify collateral assets that appear distinct in normal markets but may respond to the same underlying shock during periods of stress. The team also began modelling how crvUSD borrowing costs, scrvUSD yield and yields available on other stable assets can create persistent pressure on crvUSD demand and its peg. This helps distinguish temporary liquidity deviations from more structural yield-related imbalances.

The first public AMM research output was published in Q1, followed by a paper in Q2:

• Arbitrage and AMM fee dynamics, February 2026

• Skorokhod-based AMM analysis, May 2026

During H2, the team plans to extend the research from simplified constant-product models to Curve’s production invariants, connect correlated-risk models to LLAMMA and liquidation liquidity, and test parameter recommendations through independent simulation and a non-executing shadow system before any on-chain changes are considered.

2.2 Cross-Chain Software Infrastructure

During H1 2026, cross-chain work focused on making existing infrastructure more resilient and easier for the DAO to control. This is primarily enabling infrastructure rather than a standalone user-facing product, but it reduces operational risk, supports governance across Curve deployments, and standardises how protocol fees generated across networks are processed.

• More resilient Blockhash Oracle: The original Curve Blockhash Oracle relied on LayerZero to broadcast Ethereum blockhashes to other networks. During H1, the team has been working on an additional path using Chainlink’s CCIP and Chainlink Runtime Environment, supported by a new public Vyper library. The planned design combines LayerZero, Chainlink and an Emergency DAO fallback through a two-of-three model, reducing reliance on any single messaging provider. The new adapters are awaiting audit, with production deployment targeted for late Q3 or Q4.

• Cross-chain governance and boosts: Work continued on xGov and the wider goal of more robust DAO ownership of Curve’s cross-chain deployments. The Blockhash Oracle also allows networks to read Ethereum veCRV balances, supporting cross-chain boost delegation. Governance actions are also being standardised through the Curve voting library, making it easier for DAO votes on Ethereum to control contracts deployed across other networks.

• FastBridge: FastBridge launched in March for Arbitrum, Optimism and Fraxtal, reducing crvUSD withdrawal times to Ethereum from around seven days to roughly 15 minutes. It improves the user experience while also allowing arbitrageurs to respond more quickly when crvUSD prices diverge between Ethereum and L2s.

• Cross-chain fee processing: Infrastructure was added to standardise the handling of protocol fees generated outside Ethereum. Fees accumulated on Arbitrum were processed through the existing onchain fee-conversion mechanism and transferred to the designated treasury address on Ethereum. Support for additional networks are expected to follow during H2.

• Bridge risk controls: Following the precautionary LayerZero bridge pause in April, the team reviewed the bridge setup and reduced daily limits to constrain potential exposure. Bridging was subsequently restored across the affected networks under more conservative limits.

2.3 Backend & Analytics

During H1 2026, backend development focused on supporting Llamalend V2, improving API performance and reliability, and reducing infrastructure overhead. The team reviewed slow endpoints across the API and brought almost all queries that previously took more than five seconds to well below one second. New routing and solver infrastructure also improved execution for users and made the conversion of DAO fees materially more efficient.

• Llamalend V2 support: The backend was made version-aware across the complete data pipeline. This included V2 factory discovery, new controller and AMM interfaces, event decoding, market and user snapshots, API changes, and updated oracle and token handling. This required several thousand lines of new and refactored code and was necessary to support the production rollout of Llamalend V2.

• API performance: Every endpoint with response times above five seconds was reviewed. Improved indexing, query design and caching brought almost all of them below one second. The largest improvement involved a crvUSD endpoint querying a table of approximately 50 million rows. Its response time fell from more than ten seconds to less than ten milliseconds.

• API consolidation: Work began on a cleaner and leaner second version of the Curve API. The new API is intended to consolidate the current infrastructure and replace several older JavaScript APIs that can then be deprecated. New endpoints were also introduced for frontend redesigns, protocol research and external integrators.

• New routing infrastructure: The team developed CurveRouterV2, a new Vyper router contract supported by a Rust-based route solver. It is currently deployed on Ethereum, Base, Gnosis, Arbitrum and Optimism. The system supports split routing and balanced LP-token withdrawals, improving execution for larger and more complex trades. Quote generation is between 10 and 100 times faster than with the previous routing system. In simulations covering orders between $10,000 and $100,000, the new router produced better execution than the previous router in every tested case.

• FXSwap analytics and refueling: New backend functionality was added to track pool-specific FXSwap activity and health. This data is exposed through an internal dashboard used by the smart-contract team to monitor production pools. Refueling functionality was initially added to this dashboard and was later integrated into the public Curve frontend.

• Infrastructure consolidation: A broader audit of Swiss Stake’s AWS usage identified services and instances that could be removed or downsized. The resulting changes reduced recurring infrastructure expenditure by more than $10,000 per month, without introducing additional external service dependencies. Backend services for Fantom and Taiko were also disabled as part of the wider effort to reduce maintenance requirements for lower-priority networks.

• Monitoring and reliability: Alerts were introduced for stale RPC connections, indexing delays, indexing failures and API errors. Grafana dashboards now track API latency, traffic and queue-processing times. Reported backend uptime remained well above 99% during H1. The largest mainnet indexing delay was approximately five hours in January, followed by two delays of around one hour in May and June.

• H2 priorities: Work during the remainder of the year will focus on completing API V2, deprecating legacy APIs, supporting the frontend redesign, and expanding Llamalend V2 monitoring and analytics. Further priorities include making indexers more robust and self-healing, improving router execution, and updating Curve-Lite so that additional lower-priority networks can be consolidated. If capacity permits, the team also plans to expand user-focused analytics around behaviour, routing preferences and protocol revenue.

2.4 Front-End Software Improvements

Frontend development during H1 focused primarily on Llamalend and improving the lending experience around it. Market discovery, position management, leverage execution and data loading were substantially reworked, while major parts of the legacy Llamalend codebase were removed. The result is a clearer user experience and a cleaner technical foundation that should allow the team to iterate faster across both Llamalend and the Curve DEX.

• Llamalend V2 support: The frontend was updated to support the new Llamalend V2 market architecture and its additional features. This supported the initial Optimism rollout and left the frontend ready for subsequent deployments, including Ethereum. Most of the underlying interface is chain-agnostic, with only market availability and routing configuration requiring chain-specific work.

• Market discovery and navigation: The markets list and individual market pages received a major UX and visual update. Supply and borrow opportunities are easier to distinguish, information is presented more consistently, and search and filtering were improved. Users can also open and inspect market pages without first connecting a wallet.

• Better market information: Market tables were made faster and easier to navigate. Lending markets now include historical supply and borrow rates, interest-rate-model charts, and an improved display of LLAMMA bands. Work also began on clearer market-health information, including a redesigned health bar that remains in beta.

• Position management: User position tables were redesigned and made more stable while markets and balances are loading. The actions for opening, managing and closing positions were reorganised to make the available next steps clearer.

• Improved leverage routing: Leverage routing was rebuilt around a flexible routing architecture rather than depending on a single provider. The frontend launched with support for Enso, the Curve Router, the Curve Solver and Odos. Routes can be selected and validated more consistently, reducing brittle provider-specific behaviour and giving users clearer execution choices.

• Rewards discovery: Merkl campaign data and APRs were integrated into the Curve interface, allowing users to identify incentivised Llamalend markets without relying on an external campaign page.

• Backend and API integration: Llamalend tables, activity data and transaction flows were moved towards the new API and routing infrastructure. This provides more consistent data across chains, fewer incomplete or stale loading states, and a more reliable foundation for swaps and leveraged transactions.

• Removal of legacy code: Major parts of the legacy Llamalend implementation were removed after the replacement flows had been proven in production. This included duplicated form and input components, the previous charting dependency, and older data-fetching patterns. Shared interface components were also consolidated. These changes reduce the number of separate code paths that need to be maintained and make future product changes less likely to introduce inconsistent behaviour.

• DEX groundwork: Work also began on bringing the DEX interface onto the same technical and UX foundation as Llamalend. Early changes available behind the beta flag include an improved pool list, better filters, Prices API-powered multichain data, and support for the newer router and solver options. These changes remain groundwork rather than a completed DEX redesign.

• FXSwap refueling: A dedicated FXSwap refueling page was built using the new frontend architecture. It was the first entirely new page developed on this foundation and could be delivered quickly. In addition to making pool refueling easier, it provides an initial basis for a broader pool-management interface aimed at asset issuers and liquidity providers.

• H2 priorities: The main objective for H2 is to transfer the UX, performance and architectural improvements developed for Llamalend to the Curve DEX. Other priorities include improving the Llamalend mobile experience, completing migration to the new API, supporting additional V2 markets and yield-farming use cases, expanding analytics, and continuing performance improvements across the Curve applications.

2.5 Integrations

Integration work during H1 focused on external protocols and services building on Curve’s AMMs or connecting users to Curve markets. Several of these integrations generated meaningful activity and demonstrated uses of Curve infrastructure beyond the primary Curve interface.

• YieldBasis using Curve as core product infrastructure: YieldBasis continued to build its BTC and ETH products directly around Curve pool technology. Across four legacy and four V3 pools, these markets generated approximately $1.97 billion in H1 volume, equal to around ~8% of the H1 trading volume. This was generated against average net TVL of approximately $165.5 million, representing roughly 11.9 times volume relative to average TVL. The V3 pools launched during May and June and became some of the first production users of Curve’s upgraded FXSwap implementation. Approximately $10.98 million in LP fees were generated by Yieldbasis. This is one of the clearest examples of Curve’s AMM technology being embedded as a foundational product component rather than simply serving as a venue for a listed token.

• Onchain FX markets on Polygon: Curve participated in coordinating the launch of six FX pools with DFB, Frax and Polygon. The initial markets paired frxUSD with BRZ, IDRX, tGBP, AUDF, KRWQ and USDT. DFB led much of the market coordination, while the Curve team supported simulations, modelling and parameter selection. By the time of reporting, the BRZ/frxUSD pool had processed approximately $5.5 million in cumulative volume.

• External product integrations: The team worked with Yield.xyz to make scrvUSD, Llamalend markets and Curve LP opportunities available through its API. Enso was integrated as a routing provider for leveraged Llamalend positions, as part of broader work to support multiple execution providers. The team also worked with Merkl on Llamalend V2 reward distribution and supported DZap’s one-click entry into Curve LP positions. These are selected examples from a wider set of integrations supported during H1, many of which were primarily developed by third-parties in cooperation with the team providing technical assistance where required.

No additional Curve-Lite deployments were completed during H1, with a deployment on Robinhood Chain following shortly after the reporting period. With few new networks currently offering sufficient liquidity, demand or ecosystem support, the team is prioritizing deeper integrations and stronger markets on existing deployments. Broad network expansion is therefore not expected to be a major focus in H2, although selective opportunities will still be considered where there is a clear strategic case.

2.6 Business Development

Business development during H1 was concentrated around FXSwap and the rollout of Llamalend V2. Both depend on more than deploying contracts. Viable markets require suitable assets, liquidity, reliable oracle paths, sometimes market-maker support and, in many cases, incentives. Much of the work therefore involved coordinating issuers and liquidity providers with Curve’s development and research teams, then feeding experience from live markets back into product development.

• FX issuer outreach and market development: The team held extensive discussions with issuers of fiat-backed and tokenized currencies about launching FXSwap markets. These conversations covered existing adoption and trading demand, available liquidity, target execution quality, redemption paths, market-maker requirements and potential incentive support. Several discussions progressed into live pools or pilots, while others remain longer-term opportunities as the underlying assets and their onchain usage develop.

• Market-maker and implementation coordination: DFB was one of the team’s closest collaborators for turning this pipeline into operational markets. DFB coordinated several issuers and provided market-making infrastructure around the Polygon rollout, while Curve supported simulations, pool modelling, parameter selection and technical implementation. Experience from these markets is feeding into research on price-scale placement, rebalancing and external pricing, where established market makers could more directly be involved in pool operation and optimization, circumventing the need to bootstrap new liquidity.

• Llamalend V2 market pipeline: The team is working with asset issuers that already have Curve liquidity, as well as issuers willing to establish a Curve pool alongside a lending market. Coordination is underway around potential mainnet markets with Resupply and lending markets involving YieldBasis. Planning also covers the eventual migration of existing mint markets to V2 and the later expansion into Curve LP-token collateral.

• Market selection and bootstrapping: On Ethereum, new markets require both sufficient secondary liquidity and a suitable oracle and liquidation path. Assets with established Curve pools are therefore easier to support because the pools provide onchain EMA pricing and a venue for liquidations. Some Chainlink-based approaches remain too gas intensive for use throughout mainnet soft liquidations, making the ongoing oracle research directly relevant to the business development pipeline. Yield-bearing stablecoins are a particularly natural fit because their underlying yield can attract supply with less reliance on incentives.

• Liquidity and incentive coordination: Swiss Stake does not control a dedicated liquidity-incentive budget. Market development is therefore focused on assets with organic demand, issuers prepared to support liquidity, or independently funded incentive programmes. This makes the selection and coordination of such third-parties particularly important for Llamalend V2, where technically viable markets still need sufficient initial supply and borrowing demand.

• Ecosystem support: The team maintained working relationships with market makers, asset issuers, data providers, incentive platforms, vault developers and other integrators. This included bringing ecosystem players into direct contact with the relevant developers, resolving integration questions, supporting launches and coordinating follow-up work after products went live.

Following LlamaRisk’s departure at the end of June, Swiss Stake is also supporting the DAO-led transition to new risk assessment and market-monitoring providers. Maintaining effective risk coverage is particularly important for expanding Llamalend V2 and onboarding new markets safely.

For H2, Llamalend V2 market development is the leading business development priority, with a focus on expanding the number and adoption of viable markets. FX expansion, combined with ongoing research and closer market-maker participation, will be the other major focus.

2.7 Education & Communications

Education and communications during H1 focused on supporting major product launches, improving technical resources and maintaining Curve’s presence across the wider industry. This work continued alongside a reduction in communications staffing as part of the cost controls agreed through the renewed budget.

• Stable Summit and event presence: Curve hosted Stable Summit in Cannes during ETHCC week and a second edition in New York during EthConf. Team members presented on Llamalend V2, FX market structure, YieldBasis and AMM research, and stablecoin market design.

• Educational content: The team launched the Curve Knowledge Hub and continued publishing product explainers, documentation, monthly recaps and market updates. H1 coverage included FXSwap, FastBridge, Llamalend V2, protocol research and major governance developments.

• Interviews and public discussions: Michael Egorov participated regularly in interviews, panels and long-form discussions covering Curve’s products, crvUSD, FX markets, AMM design and decentralized governance. Other team members also represented Curve through technical presentations and media appearances.

• Grant transparency and communications efficiency: In March, Swiss Stake published its Mid-Term Review and Grant Extension Proposal, providing the DAO with a detailed account of its cost base, organizational changes, grant runway and longer-term funding roadmap. Following the review and budget approval, communications capacity was reduced and responsibilities were consolidated across fewer contributors. The remaining resources are being focused on product launches, documentation, governance reporting and content that directly supports users and integrators.

For H2, Curve expects to maintain a limited but targeted presence at major events, including TOKEN2049 and Devcon where Stable Summit will be hosted as well.

3. Financial Overview

3.1 Fund Allocation Overview – Current Period and Cumulative

The overview below outlines the allocation of grant funds for H1 2026 (January to June 2026), as well as on a cumulative basis covering the full grant period to date:

Category	Amount Spent – Cumulative prev. Grant 2025	Amount Spent – Q1 2026	Amount Spent – Q2 2026	Amount Spent – Cumulative 2026	% of Total Funds Allocated 2026
Security Audits	1,401	49	0	49	3%
Software Development	1,148	289	267	556	32%
Community & Tech Support	1,192	256	203	459	27%
Front End Development	900	173	155	328	19%
Infrastructure	359	75	110	185	11%
Research & Analytics	298	55	84	139	8%
Total	5,298	897	819	1716	100%

All numbers expressed in CHF in '000

3.2 Focus and Alignment

Swiss Stake AG continues to align its workstreams with the objectives of the DAO-granted mandate, focusing on high-impact protocol development. crvUSD remains a key contributor to DAO revenue, reinforcing the priority placed on its stability and continued growth.

Llamalend V2, now live on Optimism and Ethereum mainnet, introduces an admin fee mechanism for non-crvUSD markets. FXSwap extends the DEX into the foreign exchange segment, a nascent area with strong structural growth potential as EUR and other non-USD stablecoins expand on-chain. Together, these two initiatives represent a clear and direct monetisation path for the protocol, directly aligned with the DAO’s long-term sustainability goals.

Spending during H1 2026 reflected the revised full-year budget of CHF 4 million (previously CHF 4.9 million incl. updated security audit costs), incorporating savings of CHF 400,000 to 500,000 against the original CHF 5.3 million budget through headcount efficiency and reduced audit expenditure.

This spending profile reflects a disciplined allocation toward protocol development and research, consistent with the DAO’s priorities for the grant period. Overall, H1 2026 spending was in line with, or slightly below, the originally budgeted figures for the period.

3.3 Grant Fund Management and Financial Outlook

As outlined in the initial grant proposal, Swiss Stake AG has managed the funds received from the DAO responsibly and with a focus on long-term sustainability. Swiss Stake AG received Phase 1 of the amended grant (8,725,000 CRV) in December 2025 at an average price of approximately CHF 0.30 per CRV, providing an intended runway of approximately CHF 2.6 million. A significant decline in the CRV price to approximately CHF 0.175 during the period materially reduced the effective value of Phase 1, creating a significant funding gap against the planned H1 budget. As communicated in the Mid-Term Review, Swiss Stake AG managed these constraints responsibly and adjusted operational priorities accordingly.

Swiss Stake AG submitted a Phase 2 grant request to the Curve DAO (8,725,000 CRV plus a top-up of 2,500,000 CRV and 520,000 crvUSD) to restore the runway shortfall and fund operations through 31 December 2026. This request was subsequently approved, and the Phase 2 funds have since been received.

4. Sustainability & Outlook

Swiss Stake AG remains focused on sustainable operations and long-term alignment with the DAO. In parallel with core software development, the company continues to pursue complementary revenue streams, including Curve-Lite deployments, the Stellar grant project, and ecosystem-specific cooperations. These activities strengthen operational continuity and support further investment in protocol infrastructure.

Progress was made during H1 2026 on the structural commitments made to the DAO in the Mid-Term Review:

• Swiss Association (Verein): Preparatory work is underway with our legal counsel for the establishment of a non-profit stewardship entity covering community management, website, events, and UI/frontend activities. Efforts will be picked up in the later part of Q3.

• Protocol fee governance proposal: Following the commitment made in the Mid-Term Review, Swiss Stake AG will submit an on-chain governance proposal to increase the DAO’s protocol fee share from 10% to 30% (adjusting the fee_receiver allocation), consistent with prior DAO communications. The proposal reflects continued input from independent veCRV holders and represents a structural step toward reducing the DAO’s reliance on grant funding.

• Security audit efficiency: Total audit expenditure is being managed toward the CHF 900,000 to 1,000,000 target for 2026, taking into consideration the already realized transition of audit funding to direct DAO grant mechanisms.

Looking ahead, Swiss Stake AG will focus on scaling Llamalend V2 market adoption, continued FXSwap provider development, and expanding Curve’s reach across new chains and markets. The team is well-positioned to deliver on the H2 2026 roadmap, with Phase 2 funding now secured.

With the protocol fee proposal prepared for submission and the Swiss Association groundwork underway, Swiss Stake AG considers the structural sustainability commitments made in the Mid-Term Review to be substantially on track. The team remains committed to full transparency with the DAO as these initiatives progress into H2 2026.

Annex A: Complementary Activities Outside the Curve DAO Grant Scope

The activities described in this annex were funded independently of the Curve DAO grant and are not included among the grant-funded deliverables or expenditure covered by the main report. They are included to provide a complete overview of Swiss Stake’s activities during the reporting period.

Curve on Stellar and Rust implementation

Following a feasibility study and planning work in Q1, implementation began in Q2 under a project funded by the Stellar Development Foundation. A dedicated Rust engineer was hired specifically for this work, adding capacity without diverting existing grant-funded developer resources.

By the time of reporting, an early StableSwap implementation had undergone testing on both Stellar testnet and mainnet. The project is expected to continue throughout H2 and establishes a path toward Curve’s first non-EVM implementation.

The intended open release and the additional Rust expertise developed through the project may also support simulators, quotation engines, bots and aggregator integrations. Swiss Stake’s business-development work around the project includes coordination with the Stellar Development Foundation and exploration of potential payments, tokenised-asset and institutional use cases within the Stellar ecosystem.

[Cryptoinvest]

Thanks

Congratulations, huge work for Curve Team