April Recap
Monthly funding rose 31% from $840 million in March to $1.1 billion in April despite the deal count falling 25% from 106 to 80. April represented the highest monthly funding total going back to November 2022, helped by two nine-figure deals from Auradine and ZenMEV, representing about 27% of total funding. Infrastructure was the leading category, accumulating $639 million in funding across 41 deals, representing over half of all funding activity. From a deal stage perspective, Seed rounds were the most popular with 23 deals totaling $154 million in funding, comprising 29% of all deals but only making up 14% of the funding total. April showed an increase in late-stage deals with 2 Series C deals (only 2 in all of 2024) and 4 Series B rounds, collectively accruing $365 million in funding. There were 3 token sales completed in April, including LayerZero (ZRO), raising $55 million in a private token sale to a16z. a16z previously invested in LayerZero in both 2022 and 2023. Interestingly, LayerZero’s FDV currently sits below its Series B valuation of $3 billion (April 2023).
Funding by Category
Funding by Stage
Deal of the Month
Nous Research, a decentralized AI development company, raised $50 million in a Series A round led by Paradigm, valuing the company at $1 billion, granting Nous unicorn status. Nous is the first unicorn round since Phantom’s Series C round in January. Nous also announced a previous $20 million Seed round from Distributed Global, North Island Ventures and Delphi Digital. Nous Research has been developing AI models and applied research since 2022 with a focus on accelerating decentralized and distributed training.
Why is This Deal of the Month?
Nous Research is establishing itself as a leader in AI development, particularly in distributed training, the process of training AI models across different data centers and GPUs. Distributed training is crucial for AI because it can dramatically reduce training time and enable the use of massive datasets that wouldn’t fit on a single machine. A distributed approach is essential for scaling up advanced models like GPT or image recognition systems. There is a growing supply of underutilized heterogeneous compute resources across global data centers, and Psyche hopes to capitalize on the idle supply to train its models.
Nous has developed a distributed training over the internet (DISTRO) framework and decoupled momentum optimization (DeMo) algorithm (developed by Nous and OpenAI cofounders), which have helped decrease training bandwidth requirements by a factor of 1,000x – 10,000x. Nous DISTRO and DeMO are the key components for Psyche, Nous’ distributed training orchestrator built on Solana to coordinate heterogeneous compute resources with data, methodologies, and gradients for training end-to-end, fault-tolerant, censorship-resistant AI models. Nous hopes Psyche can democratize and decentralize the development of superintelligence by coordinating incentives across network participants.
Psyche has shown notable success in testing, training a 15-billion parameter model in December which scaled from initial training to over 11,000 steps. It was able to dynamically add compute nodes from data centers mid-training, achieve higher training rates with additional nodes, and maintain stability. It also validated extreme fault tolerance with zero impact from individual node failures and optimized bandwidth usage through innovative adaptive transmission rates. Nous is now optimizing Psyche for a person-to-person model so Nous can capitalize on the growing global supply of compute resources across heterogeneous GPUs.
The second phase of Psyche will focus on expanding its impact, offering accessible inference and advanced model capabilities. Nous believes Psyche will be the home for state-of-the-art (SOTA) models with advanced reasoning and reinforcement learning, powering the next generation of AI applications. Psyche is currently on testnet but is expected to launch on mainnet this year alongside its native token (NOUS), which will be used for incentives, governance, and potential product access.
Selected Deals
Symbiotic, a restaking protocol, raised $29 million in a Series A round from Pantera Capital. Other investors included Coinbase Ventures and angel investors from teams like Aave, Polygon, and Starkware. Symbiotic is striving to become the default coordination layer for decentralized economic security. Symbiotic is a direct EigenLayer competitor and was incubated by Paradigm and Cyber Fund. Symbiotic allows any network to plug into a permissionless and modular security infrastructure, allowing them to bootstrap their networks seamlessly and cost-efficiently. Shared security models are growing in popularity as a way to bootstrap their security cost-effectively while maintaining control of logic, tokenomics, and governance systems. Symbiotic keeps its core contracts immutable, offers flexible integrations, and keeps its role minimal, focusing solely on coordinating between networks. Symbiotic launched in the summer of last year and has accumulated $835 million in TVL, down from its peaks above $2.5 billion. Symbiotic will use the funding to accelerate its Universal Staking Framework and expand its team and tooling functionality. Symbiotic is also expected to launch a token this year, but there isn’t a confirmed date for its TGE.
Blackbird, a blockchain-based payments and loyalty platform for restaurants, raised $50 million in a Series B round led by Spark Capital. Other investors included Coinbase Ventures, Amex Ventures, and a16z. Blackbird connects customers to restaurants, reduces payment friction, and establishes a unique loyalty program. With over 1,000 restaurants signed up on the platform, users can leverage the Blackbird app to make reservations and pay for restaurant tabs while earning FLY tokens. Users earn more tokens for things like hitting spending limits or dining more times. As users accrue more FLY, they unlock benefits such as guaranteed reservations, special events, and membership to Blackbird’s upcoming private club. Customers can also pay for their meals using the Blackbird app, either through fiat methods or the FLY token, removing the need to wait for the check. Blackbird is built on Base, and its CEO, Ben Levanthal, formerly founded Resy, the popular restaurant reservation app (acquired by American Express). Leventhal has lots of experience in the restaurant business and has now brought on two strategic partners in American Express and Coinbase.
ZenMEV, a DeFi protocol harvesting maximum extractable value (MEV) into distributable yield, raised $140 million in a strategic round from VC VentureX. VC VentureX was the sole investor in the round and plans to work closely with ZenMEV to enhance its MEV harvesting, user rewards, and broader market growth. ZenMEVs’ strategies include arbitrage, front-running, back-running, and more. ZenMEV also has its own MEV protection engine that protects users from MEV events and turns the transactions into profitable yield. Users can stake ETH, SOL, USDT, and USDC on ZenMEV to start earning MEV rewards, and ZenMEV has accumulated over $40 million in TVL. The funds will be used to broaden ZenMEV’s presence in global DeFi markets, integrating with new blockchains, DEXs, and liquidity sources. The capital will help ZenMEV scale its operations and integrate AI across its product suite for enhanced MEV scanning, transaction bundling, and other creative yield-generating tactics.
