The crypto IPO trend continues to gain momentum amid growing regulatory clarity. Circle’s IPO has lit the spark for institutional appetite with confirmed filings for Kraken, Gemini and others. Sellers who exited via secondary markets in the months before Circle’s IPO missed out on tremendous upside during Circle’s post launch surge. This has prompted buyers to rotate into other late-stage names, hoping to catch a similar rerating on new IPOs. With Gemini, Bullish, and FalconX rumored next, the rotation from token-only deals to equity is accelerating.

ZK protocols have burned over $54M in gas in the past couple of years. While the proving market is projected to hit $10B by 2030, it's constrained by infrastructure that makes simple operations too expensive. @ZKVProtocol fixes this by cutting verification costs drastically. Ethereum's roadmap is pivoting to ZK tech, but the infrastructure isn't designed for proof verification at scale. This is a fundamental bottleneck. By creating a dedicated L1 for proof verification, zkVerify cuts verification costs by more than 90%. Both proof types see this cost reduction: Groth16 proofs drop from $20 to $2, while STARK proofs are reduced from $200 to $20 per proof. For reference, Starknet's $168k annual verification costs would drop to $20k with these cost reductions. Developers can simply send proofs to zkVerify's L1 which is verified in a purpose-built environment with native support for all major proving systems. Once verified, zkVerify creates an attestation that works on any chain, eliminating the need for custom verifier contracts entirely. Beyond cost savings, ZK protocols don't need to compromise anymore. • Users can get instant withdrawals instead of long delays • Rollups can verify every hour instead of rationing proofs • Users can benefit from lower fees • Developers can focus on building instead of maintaining compatible verifiers By cutting costs, lowering latency, and simplifying proving, zkVerify fills a crucial design gap in the modular stack: a shared verification layer that is compatible with any ZK primitive or chain.

The four-year cycle may still be relevant • Market structure mirrors late 2017 with BTC ATHs followed by ETH outperformance • Global M2 expansion aligns with historical cycle tops • DAT vehicles echo reflexive risks that ended previous cycles Read more here 👇

While distributed computing remains a broken promise, @theblessnetwork is building the infrastructure to make it a reality. Here are some real use cases that Bless enables. Machine Learning & AI Processing Bless is democratizing AI by creating a distributed AI ecosystem to make machine learning more accessible for everyone. By enabling model sharding across multiple devices, models can be served and fine-tuned across distributed nodes. This reduces costs while improving training times through massive parallelization. The network's federated learning capabilities allow companies to contribute to model training while keeping their data local. Healthcare companies could train AI on patient data without it ever leaving the hospital. Only gradient updates are shared across the network, addressing one of the most significant barriers to AI adoption in regulated industries. Gaming Latency One issue that has plagued online games is latency. Players experience widely different quality gameplay based on their geographical location. Bless solves this through hierarchical state synchronization. Regional game states are maintained locally for responsive gameplay while being synchronized across regions to maintain global consistency. By allowing game physics calculations to happen close to players rather than in distant data centers, gamers benefit from responsive gameplay that feels consistent regardless of location. Partnerships Bless is working with @SpaceandTimeDB to integrate ZK-proof capabilities into their network. AI agents deployed through Bless can access cryptographically verifiable inputs including weather forecasts, grid location projections, and energy prices. This architecture works well for EV charging networks. Agents can analyze real-time grid capacity, electricity prices, and user demand patterns to prevent grid overload and minimize costs for users. Bless has also partnered with @monad to enable deployment of autonomous AI trading agents that run entirely on user devices while interacting on-chain. Monad delivers high throughput and low latency, and Bless complements it by providing the compute layer for local, low-latency agent inference. These agents could analyze market patterns in real time and lower the barrier to entry for HFT. This allows anyone to deploy sophisticated trading agents on their own devices.

The @avax appchain thesis is playing out in real-time. • Daily L1 transactions are up 6x to 18M txs since January • Smart contract deployments exploded from 1k to 250k daily • Demand is led by appchains such as @MaplestoryU & @playoffthegrid • The network now supports 82 active L1s with 890+ validators • ICM volume has seen a 25x surge since April Avalanche is becoming the platform for teams that want to own their stack

Prop AMMs are reshaping Solana • They now route 40% of Jupiter volume by hiding orderflow from MEV • SOL/USDC pairs see 42% of trades through these AMMs with tighter spreads • Prop AMMs dominate liquid pairs while public AMMs are used for memecoins Read more here.👇

Computing is still stuck with expensive cloud monopolies and experiments in decentralization that haven’t lived up to their promises. @theblessnetwork is building a global supercomputer using idle devices – turning consumer tech into nodes in a distributed computing platform that runs AI models, gaming infra, and more. No hardware requirements or complicated configurations needed. Why Distributed Computing Keeps Failing Traditional networks exclude most users through complexity, specialized hardware requirements, and security concerns. Even when people join, the tech fails: • Gaming requires millisecond response times but involves long wait times in practice • Powerful nodes are bottlenecked while consumer devices sit idle • Decentralized apps can't match centralized performance However, Bless takes a fundamentally different approach. Bless splits AI workloads and other computing functions across multiple everyday devices with model sharding. By distributing workloads across the network, Bless cuts costs compared to expensive CPU/GPU farms. Adaptive cryptographic verification matches validation to actual needs. Simple calculations get lightweight verification while critical computations receive multiple validation rounds. The goal is for the network to adapt to real-time conditions to eliminate the bottlenecks that stifle traditional models. Bless also offers a serverless interface for developers to submit workloads without worrying about infrastructure. This combines the simplicity of centralized services with the resilience of decentralized networks. If Bless delivers on its promises, traditional edge computing becomes obsolete. Now in testnet, Bless has surpassed 800 petaFLOPS in compute power, with over 5.4 million nodes registered in 210 countries worldwide. The testnet numbers are impressive. The architecture looks thoughtful. The timing might finally be right. But distributed computing has been "almost here" for a decade. Bless needs to prove it's different this time.

My first @delphi_intel report, 'Browser Wars Got Personal' is out! The era of app-centric computing is coming to an end. As AI learns to understand human intent directly, browsers are emerging as the ultimate frontend for using AI. Welcome to the AI Browser wars. 🧵

My first @delphi_intel report, 'Browser Wars Got Personal' is out! The era of app-centric computing is coming to an end. As AI learns to understand human intent directly, browsers are emerging as the ultimate frontend for using AI. Welcome to the AI Browser wars. 🧵

Introducing the Digital Asset Treasury Dashboard! Track $150B+ in BTC, ETH, and SOL held in crypto treasuries all in one place for free. See Treasury & Fully Diluted mNAVs, portfolio breakdowns (MicroStrategy, Metaplanet & more), and regular updates.