Author: TinTinLand
In the past year, there has been a noticeable change in the global venture capital market: capital is flowing into AI at an unprecedented speed, while the investment pace in the crypto industry has significantly slowed.
Let's take a look at two sets of data:
On one side is the crypto market: the total number of global crypto VC investments plummeted from about 2900 cases in 2025 to 1200 cases, a decline of more than 60%.
On the other side is the AI market: according to OECD data, the scale of global VC investment in the AI sector reached $258.7 billion in 2025, accounting for 61% of the total global venture capital, while this proportion was only 30% in 2022.
This structural pressure difference is quietly pushing a group of established crypto VCs to adjust their course and begin deploying capital into the AI track.
On February 28, according to industry media reports, top crypto VC Paradigm is raising a new fund of up to $1.5 billion, with investment directions no longer limited to the crypto field but expanding to artificial intelligence, robotics, and other frontier technologies.
Are VCs really abandoning crypto and shifting to AI?
A careful analysis reveals that most leading crypto VCs have not simply turned to AI, but are seeking new opportunities arising from the collision of AI and Crypto. AI requires computing power, data, identity, and payment infrastructure, which are precisely the problems blockchain excels at solving.
TinTinLand reviewed 6 of the most representative institutions to see which AI projects they specifically backed in the past year.
1️⃣ Paradigm: $1.5 billion new fund bets on AI
Since its establishment in 2018, Paradigm has been known for its research-driven investment style, with Uniswap, Coinbase, and dYdX being classic achievements.
As early as 2023, Paradigm briefly removed 'Web3' related statements from its official website, sparking market speculation about whether it is undergoing a strategic transformation.
In response to external doubts that 'Paradigm is abandoning crypto', co-founder Matt Huang explicitly stated, 'We have never been more excited about crypto; we continue to invest at all stages,' but also emphasized, 'The progress of AI is too interesting to ignore. We do not believe that Crypto and AI are zero-sum competition; there is a lot of overlap between the two.'
Paradigm's recent representative layout
? Nous Research (April 2025)
Decentralized AI startup Nous Research completed $50 million in Series A financing in 2025, led by Paradigm, with the project valuation reaching $1 billion.
The company is dedicated to developing and training open large language models, contrasting sharply with the closed paths of centralized giant laboratories.
? EVMbench (February 2026)
In early 2026, Paradigm and OpenAI jointly launched EVMbench.
This is a benchmark testing tool for assessing the capabilities of AI models in the field of smart contract security audits, providing a technical foundation for future AI automated audit tools.
? $1.5 billion new fund (February 2026)
According to several industry media reports, Paradigm is raising a new fund targeting up to $1.5 billion, expanding its investment scope to AI, robotics, and other frontier technology fields.
The fund is still led by Paradigm's existing research and engineering team, which means its AI investment path continues its consistent style: prioritizing underlying technologies and infrastructure.
2️⃣ a16z: AI+Crypto dual-line battle
Founded in 2009, Andreessen Horowitz (a16z) is one of the most influential VCs in Silicon Valley. In 2018, it established the a16z crypto fund, systematically laying out blockchain infrastructure and DeFi. With the explosion of generative AI, a16z quickly opened up a second front.
In early 2026, a16z announced the completion of fundraising for a new fund exceeding $15 billion, with approximately $1.7 billion specifically earmarked for AI infrastructure investment.
Several a16z crypto partners have publicly stated that 2026 will be a crucial year for AI Agents to transition from prototype stages to practical application pilots. AI and crypto networks have natural intersections in identity, payment, and incentive mechanisms, and are likely to form new technology integration directions in the future.
a16z's recent representative investments
? Thinking Machines Lab (June 2025)
In 2025, AI startup Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, announced the completion of $2 billion in seed round financing, with a post-investment valuation of $12 billion, led by a16z. This is one of the largest seed rounds in Silicon Valley history.
The company focuses on building safer and more general generative AI systems and has attracted a large number of researchers from organizations such as OpenAI and Meta.
? Cursor (2024-2025)
AI programming assistant Cursor received multiple rounds of investment from a16z between 2024-2025, and after the D round financing in November 2025, its valuation reached $29.3 billion.
Cursor is currently one of the fastest-growing AI development tools. It deeply integrates large language models, enabling developers to complete complex tasks such as code generation, debugging, and refactoring through natural language. According to several industry media reports, Cursor's annual recurring revenue (ARR) has surpassed $2 billion.
? Catena Labs (May 2025)
Catena Labs was founded by Sean Neville, co-founder of Circle and inventor of USDC, and is dedicated to building customized financial infrastructure for AI Agents, including smart payments, identity verification, and low-cost settlements.
In 2025, the project completed a $18 million Seed round financing, led by a16z crypto, with participation from institutions such as Coinbase Ventures.
3️⃣ Polychain: Layout AI trust and identity layer
Polychain Capital was founded in 2016 by former Coinbase employee Olaf Carlson-Wee and has been positioned as a crypto-native investment institution since its inception.
Polychain has not explicitly stated 'shifting from Crypto to AI', its investment logic is closer to viewing AI as a new layer of blockchain infrastructure opportunity.
In Polychain's view, the core bottlenecks of AI, such as computing power, data, and incentive mechanisms, are precisely the problems that blockchain excels at solving.
Polychain Capital's recent representative investments
? Billions Network (August 2025)
Billions Network is an identity infrastructure project that combines AI, privacy computing, and zero-knowledge proofs, allowing users to complete identity verification without disclosing personal data and providing a trusted identity layer for AI systems.
In August 2025, the project announced the completion of approximately $30 million in financing, with investors including Polychain Capital.
? Talus Labs (2024-2025)
In September 2025, Talus Labs completed a $10 million financing round, led by Polychain Capital, marking Polychain's third investment in this project.
Talus Labs proposed the concept of PredictionAI: combining AI Agents with on-chain prediction markets, allowing AI models to compete, predict, and make decisions in a transparent and verifiable environment.
? Grass (2025)
Grass is a data network project aimed at AI models. Grass uses the DePIN model to allow users to contribute bandwidth and network data resources and uses this data for AI model training.
In 2025, Grass completed approximately $10 million in financing, with investors including Polychain Capital. This is also Polychain's third investment in this project.
4️⃣ Pantera Capital: Increasing investment in DePIN and decentralized computing power
Pantera Capital was founded in 2003, initially as a global macro hedge fund. It fully transitioned to investing in crypto assets in 2013 and is an early supporter of projects like Circle, Coinbase, and Ripple.
The AI investment logic of Pantera Capital is based on several core judgments: AI needs new computing and data infrastructure; Web3 can provide decentralized computing power and data networks; DePIN + AI will become a new infrastructure layer.
Pantera Capital's recent representative investments
? Gradient Network (June 2025)
Gradient Network is a decentralized AI inference network aimed at utilizing idle computing power from personal computers and mobile devices to provide distributed inference capabilities for AI models, thereby reducing AI computing costs.
In 2025, Gradient Network completed $10 million in Seed financing, led by Pantera Capital and Multicoin Capital. The project is regarded as an important exploration in the DePIN × AI infrastructure track.
? Based (February 2026)
Based is a Web3 super app that integrates trading, payment, and social functions, introducing AI Agent as the primary interaction method for on-chain operations, allowing users to automatically execute transactions and asset management through AI.
In early 2026, Based completed approximately $11.5 million in financing, led by Pantera Capital, with participation from institutions such as Coinbase Ventures.
5️⃣ 1kx: Betting on the Agent Internet era
1kx was established in 2018 and is headquartered in Berlin. It is an early supporter of projects such as Lido, Arweave, and Gitcoin, establishing significant influence in the DeFi and decentralized infrastructure fields.
1kx's core judgment is that the internet is entering the 'Agent Internet' era, where numerous AI Agents will perform tasks, transactions, and collaborations on behalf of users, and Web3 infrastructure is the key support layer for this transformation.
1kx's recent representative investments
? Olas (February 2025)
Olas (formerly Autonolas) is a decentralized AI Agent network that allows developers to create and deploy autonomously running AI agents and incentivizes multiple agents to collaborate on tasks through on-chain mechanisms.
In February 2025, Olas completed a $13.8 million financing led by 1kx. The project plans to launch an AI Agent app store called Pearl, allowing users to own and control their AI agents on-chain and use them in scenarios like automated trading and social media operations.
? Camp Network (April 2025)
As generative AI models increasingly use internet content for training, copyright disputes between content creators and AI companies are on the rise. Camp Network is dedicated to building an on-chain storage and licensing platform for content creators, exploring new business models for entertainment and IP assets in the AI era.
The project completed a $25 million Series A financing in 2025, jointly led by 1kx and Blockchain Capital.
6️⃣ YZi Labs: Investing in AI applications and product innovation
YZi Labs was originally Binance Labs, officially independent from the Binance brand in 2025, distinctly listing Web3, AI, and biotechnology as its new three major investment directions.
In October 2025, YZi Labs announced the launch of a fund approximately $1 billion in size, the BNB Builder Fund, to support emerging projects including AI, RWA, DeFi, and infrastructure.
Compared to some crypto VCs focused on underlying technologies, YZi Labs' AI investment strategy is more application-oriented.
YZi Labs AI's recent investment cases
? VideoTutor (October 2025)
VideoTutor is an education AI Agent aimed at K12, capable of automatically generating animated instructional videos from user questions and explaining complex concepts through voice and visualization. Its founder is a 20-year-old Chinese university student, Kai Zhao.
In October 2025, YZi Labs led the project with a $11 million Seed round financing. The product attracted 20,000 users within 10 days of launch and generated 20,000 instructional videos.
? USD.AI (August 2025)
In August 2025, YZi Labs announced a strategic investment in USD.AI. USD.AI is an AI infrastructure financial protocol that allows GPU or AI server operators to obtain loans through hardware collateral and issue stablecoins related to computing resources, thus providing funding sources for AI computing power construction.
? Summary: Which AI tracks do crypto VCs prefer?
From an overall trend perspective, the AI investment tracks most favored by crypto VCs are mainly concentrated in the following directions:
① Decentralized AI infrastructure
This is the most favored track, with representative projects including Nous Research, Gradient Network, Grass, and others.
The core logic is that the training and inference costs of AI models are extremely high, and coordinating globally idle GPUs through blockchain incentive mechanisms could potentially build a lower-cost, more decentralized alternative.
② AI Agent payments and finance
Projects like Catena Labs and Olas represent this main line.
As AI Agents begin to automatically execute tasks and transactions, demands for autonomous identities, automated settlements, and on-chain payments become urgent. The slow speed, high barriers, and compliance requirements of traditional financial systems are extremely unfriendly to AI Agents, creating a real infrastructure gap.
③ AI identity and trust layer
Representative project: Billions Network.
In an internet environment flooded with AI-generated content and rampant bots, 'proving you are human' is becoming a key demand. The combination of zero-knowledge proofs and on-chain identity mechanisms provides a solution for completing identity verification without disclosing data.
④ AI × IP copyright confirmation
Representative project: Camp Network.
With the development of generative AI, the sources of training data and copyright issues are becoming increasingly important, and blockchain provides a traceable and executable copyright registration and profit-sharing mechanism.
⑤ AI application layer
Some crypto VCs have also directly bet on user-facing AI products like Cursor and VideoTutor — not emphasizing Web3 attributes, but rather directly using AI to lower service barriers and expand inclusive boundaries.
? In conclusion
AI Agents need to make autonomous payments, computing power needs decentralized coordination, identities need on-chain verification, and copyrights need traceable confirmation — these demands are not narratives created by Web3 but represent fundamental infrastructure gaps inevitably brought about by AI development.
Crypto VCs have not truly 'abandoned' Web3. The shift of crypto VCs to AI is a bet on its next explosion point.