The crypto space is buzzing with talks about Bitcoin miners and their role in powering AI-related computations. As you've probably heard, the biggest mining companies are gradually shifting their operations to serve the AI industry, in a pursuit of higher profits and lower volatility. Within just a couple of days, several major updates came to light:
CleanSpark won a major AI‑data‑center site in Wyoming, beating out Microsoft, and partnered with engineering company Submer to develop AI-focused campuses across North America. The plan combines CleanSpark’s energy and land assets with Submer’s liquid‑cooled, high‑density systems.
TeraWulf plans to raise $500M via a convertible senior note offering (with an option for an additional $75M) to fund the construction of a new data-center campus in Abernathy, Texas. This comes amid TeraWulf’s broader pivot toward AI-focused infrastructure: it’s reportedly working with Morgan Stanley on a separate up-to-$3B financing effort (with a $1.4B backstop from Google) and has a $3.7B hosting deal with Fluidstack.
The shareholders of Bitcoin miner Core Scientific voted against a proposed $9B all-stock merger with AI computing company CoreWeave, causing the deal’s termination. Key reasons for the rejection: some investors believed the valuation undervalued Core Scientific’s prospects.
Bitcoin Miner Core Scientific Investors Nix $9 Billion CoreWeave Merger - Decrypt
Core Scientific investors have rejected a $9 billion deal that would have seen AI compute firm CoreWeave buy the Bitcoin miner.
Meanwhile, Riot Platforms, Inc. recorded a $93.3M increase in its mining revenue in Q3 2025, so not everything is lost for Bitcoin miners. Yet, large‑scale Bitcoin miners are essentially becoming “utilities” for AI compute rather than only mining coins, so this naturally raises questions about the heath of the Bitcoin infrastructure.
Mining infrastructure (cheap power, land, cooling, industrial scale) overlaps with high‑performance compute/AI infrastructure needs. It's absolutely understandable that when the “hashprice” (i.e., revenue per hashrate) falls, hosting AI workloads under fixed‑dollar contracts may offer better economics in $/MWh terms.
Futuriom just published a fascinating report, describing the global “arms race” to build large‑scale AI clouds (“neoclouds”) and GPU‑heavy data centers, often consuming gigawatts of power and requiring massive capital expenditure. It defines neoclouds as clouds purpose‑built for AI training/inference (rather than general‑purpose public clouds), many arising from crypto‑mining or edge‑compute firms pivoting into AI.
The report estimates that more than $1 T has reportedly been committed to AI data center build‑outs to date.
It also highlights that securing power supply and land is now a key competitive factor in AI cloud infrastructure. Moreover, it doesn't exclude the possibility of a “bubble” given the circular dynamics: e.g., NVIDIA supplies GPUs to neoclouds, which in turn build infrastructure that needs more NVIDIA hardware.
AI, GPU Clouds, and Neoclouds in the Age of Inference
The explosion of GPU clouds and neoclouds is complicated to sort out. In our new report, we did all the work for you.
A couple of days ago, I covered that NEAR Protocol had failed to garner enough community support to enact a decrease of its token emission rate. Well, despite that, the so-called "halving" is now a reality.
NEAR has implemented a network upgrade that cuts its annual inflation rate from 5% to roughly 2.4%, although a governance vote on the matter attracted only ~45% voted in favor, below the required 66.67% super-majority threshold.
The change aims to reduce token dilution (around 60 million fewer NEAR tokens minted yearly) and align staking incentives (staking yield dropping from ~9% to ~4.5% assuming half the supply is staked).
Some validators criticized the move, arguing it sets a “dangerous precedent” by enabling the core team to enforce economic changes even when the formal vote didn’t pass. On the other hand, NEAR's Bowen Wang commented:
NEAR Cuts Inflation Rate by Half Despite Failed Community Vote - "The Defiant"
NEAR Protocol has upgraded its network with a new update that reduces its annual token inflation rate from 5% to roughly 2.4%, even though the community vote intended to authorize the change failed to reach the required threshold.
MetaMask's Marco De Rossi, one of the authors behind Ethereum Foundation's Trustless Agents standard, just presented Agent0, a Python SDK for agent probability, discovery, and trust on ERC-8004.
Agent0 enables agents to register, advertise their capabilities and give each other feedback and reputation signals. All this using blockchain infrastructure (ERC-8004) and decentralized storage, enabling permissionless discovery without relying on proprietary catalogues or intermediaries.
The Agent0 SDK empowers developers to:
Create and manage agent identities: Register your AI agent on-chain with a unique identity, configure presentation fields (name, description, image), set wallet addresses, and manage trust models with x402 support.
Advertise agent capabilities: Publish MCP and A2A endpoints, with automated extraction of MCP tools and A2A skills from endpoints.
Enable permissionless discovery: Make your agent discoverable by other agents and platforms using rich search by attributes, capabilities, skills, tools, tasks, and x402 support.
Build reputation: Give and receive feedback, retrieve feedback history, and search agents by reputation with cryptographic authentication.
Public indexing: Subgraph indexing both on-chain and IPFS data for fast search and retrieval.
If you need more information on Ethereum's ERC-8004, read here, and if you're looking for clarification on Coinbase's x402, check out my latest deep dive here.
Fhenix, a protocol delivering FHE-powered privacy solutions, has announced a strategic investment from BIPROGY Inc., a major Japanese IT services firm, and Translink Capital’s Sustainability & Innovation Fund. The funding is aimed at accelerating Fhenix’s entry into the Japanese privacy-focused stablecoin and Web3 fintech market, leveraging their homomorphic encryption technology to offer privacy-first blockchain infrastructure in Japan.
BIPROGY’s backing gives Fhenix invaluable access to enterprise clients, local technology ecosystems, and regulatory intelligence — a rare advantage for a privacy-native protocol aiming for real adoption in Japan.
In Japan’s maturing crypto landscape, many users and institutions are pushing back against public exposure of portfolio and transaction data. As more DeFi protocols move on-chain, privacy becomes not a niche, but a baseline expectation. This capital injection positions Fhenix to lead that shift.
Fhenix Announces Strategic Investment from BIPROGY & TransLink Capital, Accelerates Entry into Japanese Privacy-focused Stablecoin and Web3 Fintech Market
Tel Aviv / Tokyo - October 22, 2025 - Fhenix, the leading infrastructure protocol powering Private DeFi via homomorphic encryption, today announced a strategic investment from BIPROGY and Translink Sustainability & Innovation Fund. BIPROGY Inc. is one of Japan's largest IT service providers, with a strong footprint in the financial sector.
Semantic Layer secured $5M in a Series A funding round led by Greenfield Capital to advance on-chain AI autonomy. The funding follows a prior seed round of $3M led by Figment Capital.
The new capital will be used to accelerate the development of “asset ordering sovereignty” infrastructure, enabling dApps, agents, and assets to control transaction sequencing rather than leaving it entirely to external validators.
Semantic Layer Funding $5M Series A
Semantic Layer has secured $5 million in Series A funding to drive on-chain AI autonomy, dApps and asset sequencing sovereignty.
Thank you for reading! The next edition is coming next week.
I invite you to subscribe to The Web3 + AI Newsletter to stay in the loop on the hottest dAI developments.
I'm looking forward to connecting with fellow Crypto x AI enthusiasts, so don't hesitate to reach out to me on social media.
Disclaimer: None of this should or could be considered financial advice. You should not take my words for granted, rather, do your own research (DYOR) and share your thoughts to create a fruitful discussion.