#8: AI's biggest VC quarter
Plus: Top reads, alpha, the first GPU cloud rating, Gensyn launch and top agents.
Hi, it’s Marc.
AI just had its biggest VC quarter in 3 years.
💰 1 in 5 VC deals in Q1’25 went to AI.
📈 Median deal size? $3.5M — a new high.
🚨 OpenAI’s $40B round led the charge.
Investors are moving from infra to industry-specific AI — with healthcare, robotics, and automation getting the love.
Today, we cover:
Top Reads
Alpha
Spotlight: Gensyn
Spotlight: The first GPU Cloud Rating
Top AI Agents
“The future of crypto lies in Specialized AI Agents - powered by Specialized Language Models (SLMs) that understand context, nuance, and the fast-moving nature of Web3.”
Let’s dive in 👇
The Ultimate Vendor Intelligence Platform
Discover, compare, and track AI vendors in real time — powered by research and data, not hype. Join the waitlist for exclusive early access 👉
Top Reads
Reasoning models don't always say what they think. Anthropic. Link
AI Can Cause The PMF Threshold To “Inflect”, Causing Product Market Fit Collapse. Reforge. Link
State of Venture Q1’25 Report. CB Insights. Link
Detecting AI Content or Playing a Game of Guess? Link
Google Has Finally Dethroned ChatGPT. Link
How DeepSeek-R1 Pushes the Limits of Language Models — A Mathematical Dive into Group Relative Policy Optimization (GRPO). Link
Everything You’ve Heard About AI In Game Development Is Wrong. Link
The power of private composable data in advancing DeSci. Link
What is DePAI (Decentralized Physical AI)? Link
The Bias Problem: AI's Most Dangerous Flaw. Link
Alpha
Fetch.ai's ASI-1 Mini, a Web3 LLM, integrates with Llama Index to enhance RAG pipelines, enabling agentic reasoning and decentralized context-awareness for more intelligent AI applications.
Universal Machine Time (UMT) by peaq introduces the first onchain Precision Time Protocol, syncing machines and DePINs with nanosecond accuracy for real-world applications like autonomous vehicles and AI coordination.
Cortensor's Proof of Inference (PoI) system uses decentralized nodes to validate AI outputs, addressing trust issues in AI by ensuring no single point of failure.
Autonolas launched Mech Marketplace.
The Silicon Crowd concept builds on the "wisdom of the crowd" idea, where aggregating predictions from many large language models (LLMs) creates more accurate forecasts that rival human crowd accuracy.
NetMind.AI has integrated with n8n, a workflow automation tool known for its 400+ app integrations and native AI capabilities.
Partnerships:
Inference Labs partnered with Friend3AI, a social dApp focused on decentralized monetization, to create a user-controlled, trustless social platform.
Nillion's collaboration with the Coinbase Developer Platform introduces a private data storage vault for AI agents.
Pond and Sentient's partnership leverages Pond's expertise in AI models trained on on-chain data, aiming to enhance Sentient Chat with crypto-specialized agents for better user insights.
👉Enjoying 51x? Subscribe to our Web3 newsletter for business leaders
Gensyn is live
The Gensyn testnet is live [Github, Announcement]
Why it matters: Gensyn raised $50.6M from a16z, EdenBlock, and others. They’re an emerging player in the decentralized GPU space.
Zooming in: Gensyn Protocol is a layer-I protocol for deep learning computation that rewards participants for their compute time. It facilitates task distribution and payments through smart contracts, ensures the verification of completed ML work and creates a decentralized Al compute network.
Who can participate?
Swarm Participants: Run a node, track contributions, and train local models.
Developers: Deploy self-owned ML models and update parameters in real-time.
ML Researchers: Create custom swarms, train on new datasets, incentivize participation.
Punchline: Gensyn’s P2P compute network could cut GPU costs by up to 85%—offering $0.62/hour for Nvidia A100 vs. $4.10/hour on AWS. This makes AI training cheaper while boosting privacy and ESG benefits. However, scalability, reliability, and trust remain key challenges for decentralized AI computing.
More links on :
First Ever GPU Cloud Rating
SemiAnalysis just dropped ClusterMAX™, the first-ever rating system for GPU cloud providers covering 90% of the market.
Why it’s important: Most decentralized GPU networks—like SaladCloud and Massed Compute—got slapped with “UnderPerform” ratings due to key weaknesses:
Weak Security: Many decentralized providers lack key security certifications like SOC2 or ISO 27001, exposing user data to risks like internet traffic leaks and third-party logging.
Unreliable Service: Frequent outages and slow issue resolution.
Poor Technical Support: Lack of expertise and poor support make issue resolution and setup difficult.
Misleading Information: Some providers use confusing or incorrect marketing, including AI-generated content that misleads customers.
Examples:
Security Flaws: Providers like SaladCloud admit that third-party equipment might log data traffic, raising privacy concerns.
Harmful Marketing: Massed Compute’s inaccurate articles clutter the internet with misinformation, making it harder for users to find reliable facts.
Zooming in: ClusterMAX™ is a rating system assesses providers based on factors like security, reliability, technical support, and pricing.
How Decentralized GPU Providers are Rated
Decentralized GPU providers, often categorized as "Emerging Neoclouds," are newer players in the GPU rental market. Emerging Neoclouds struggle in GPU rentals, with many rated UnderPerform by ClusterMAX™
Punchline: Decentralized GPU is still early and competing against established hyper scalers is hard. he market shift to a buyer’s market creates an opportunity, but only for those who improve fast.
More links on :
Decentralized Cloud Computing: How Blockchain Reinvents Data Storage
Decentralized GPU Networks: Demand, Challenges, and Future Opportunities
Top AI Agents by mindshare
Some agents you want to keep an eye on:
$VIRTUAL ($365M): A framework and infrastructure designed to create, deploy, and monetize autonomous AI agents across various platforms and applications, such as gaming, entertainment, social interactions, and more.
$TRAC ($173M): A decentralized knowledge-sharing protocol and blockchain-based platform that uses a Decentralized Knowledge Graph (DKG) to organize, verify, and share trusted data and knowledge assets across industries.
$AI16Z ($167M): Solana-based web3 platform that employs a unique AI agent-driven VC investment model and community-driven DAO governance.
$FAI ( $141M): AI built with publicly available large language models (LLMs) and operates on the Base blockchain.
$PAAL ($98M): Personalized, advanced AI assistants and applications, particularly for crypto investors and broader digital communities.
$PHA ($74M): A decentralized blockchain platform focused on providing secure, privacy-preserving computation for Web3 applications.
$AIXBT ($69M): A crypto-savvy AI that scans markets and social media to deliver trading insights and alpha opportunities.
That’s it for now.
– Marc & team