Bitcoin Magazine
Benchmark Raises Hut 8 Price Target After Bitcoin Miner Signs $9.8 Billion AI Data Center Deal
Investment bank Benchmark has raised its price target on Bitcoin miner Hut 8 following news that the company had signed a second 15-year lease worth $9.8 billion for its AI data center.
Equity research analyst Mark Palmer reiterated his “buy” rating on the Toronto- and Nasdaq-listed miner in a note Wednesday, raising its price target to $195, up from $165 — an 80% upside from Hut 8’s current share price of nearly $110 a pop.
Palmer argued that the deal validates Hut 8’s “power-first” approach to building out AI infrastructure.
Hut 8 announced Monday that it had signed a second 15-year lease for 352 megawatts of IT capacity at its Beacon Point campus in Nueces County, Texas — doubling the site’s tenant to 704 MW of contracted capacity and fully commercializing the campus against its 1,000 MW of utility capacity. Hut 8 shares climbed over 10% on the news, closing near $101 after peaking above $106.
Palmer estimates the new Beacon Point lease alone could contribute roughly $655 million a year in net operating income once stabilized, pushing the campus’s total contract value as high as $50.2 billion if renewal options are exercised.
Hut 8 is among a growing list of publicly traded Bitcoin miners pivoting toward AI and high-performance computing as mining margins get squeezed by a falling Bitcoin price and rising difficulty.
The company struck a Google-backed deal in December with Anthropic and Fluidstack to build out as much as 2.3 gigawatts of AI data center capacity in the U.S.
Rivals including Terawulf, IREN, and Cipher Mining have signed similar multi-year HPC contracts with Google and Microsoft, while Bitfarms said last year it would wind down its mining operations entirely to focus on high-performance computing.
Hut 8, by contrast, has kept its mining business running through its majority stake in American Bitcoin, the mining venture backed by Eric Trump and Donald Trump Jr.
Hut 8 is scheduled to report second-quarter earnings on August 4.
This post Benchmark Raises Hut 8 Price Target After Bitcoin Miner Signs $9.8 Billion AI Data Center Deal first appeared on Bitcoin Magazine and is written by Mathew Di Salvo.
Bitcoin Magazine
Lightning Labs Launches Wavelength: “Bitcoin on Easy Mode” for Developers and Autonomous Agents
Lightning Labs, the company developing core Lightning Network software including Lnd, Loop, and Taproot Assets, has released the alpha version of Wavelength. The toolkit enables developers and AI agents to add self-custodial Bitcoin payments to applications through a simple non-custodial API, without running nodes, managing channels, or sourcing liquidity.
In a July 21, 2026 blog post, Lightning Labs described Wavelength as “Bitcoin on Easy Mode for Agents and Humans.” The company stated that the Lightning Network already delivers instant, global, low-fee payments under user control, but previously required infrastructure most builders preferred not to operate. Wavelength closes that gap by turning the hard parts of Bitcoin and Lightning integration into a handful of API calls.
Wavelength embeds a self-custodial wallet that runs inside web or mobile apps (via WebAssembly or compiled binaries) or as a standalone client. Users control their own keys on-device. The system supports on-chain Bitcoin, Lightning payments via atomic swaps, and an Ark-like settlement layer for fast, low-cost off-chain transfers that can settle in batches to the blockchain. Every off-chain payment uses a standard BOLT 11 invoice, so the wallet interoperates with the existing Lightning Network from the first integration.
Lightning payments route through Loop for deep, reliable liquidity. A coordination service settles transfers between users but never takes unilateral control of funds. According to the announcement, users can always perform a unilateral exit to on-chain Bitcoin at any time via an explicit exit command, without needing cooperation, the Wavelength SDK is open source.
The same Wavelength API is exposed to AI agents as typed tool calls through the Model Context Protocol (MCP). Agents can hold balances and pay for API calls, data feeds, or other agent services in fractions of a cent. Wallet creation and unlocking designed to remain outside the agent channel so seeds and passwords are not exposed to the model. This pairs with L402, Lightning Labs’ protocol for machine-native authentication and per-request Lightning payments.
Core commands cover the full lifecycle: create/unlock, balance, recv (for addresses or invoices), send, activity, and exit. Integration options include the embedded SDK, a gRPC/REST API, browser WASM package, and an MCP server. Documentation is structured for both human developers and agents, including llms.txt indexes and agent onboarding guidance.
Wavelength is available immediately on Signet and testnet. Mainnet access is invitation-only; interested parties can request it after installing the toolkit. Bitcoin is supported at launch. Stablecoin support is planned via Taproot Assets so the same API surface can handle both. Future work includes deeper mobile embedding and optional direct Lightning channel support using Lnd.
Lightning Labs noted in its announcement that during the closed alpha, Lightning transactions carry a minimal 1 basis point service fee (plus standard network routing fees), with ordinary Bitcoin network fees applying for on-chain activity. Pricing may evolve.
On X, Lightning Labs summarized the release: “Announcing Wavelength, the easiest way to integrate bitcoin for agents and humans. With a simple non-custodial API, anyone can integrate Lightning into their app and get instant, high volume, low fee transactions. Machines can pay machines. Humans can pay humans. Anywhere.” A follow-up post directed builders to a form for early mainnet access.
The release positions Wavelength as infrastructure that lowers the barrier for application developers, “vibe coders,” and autonomous agents to offer self-custodial Bitcoin payments by default rather than as a specialist feature. Full documentation, quickstarts, and the open-source repository are available at wavelength.lightning.engineering and the linked GitHub project.
This post Lightning Labs Launches Wavelength: “Bitcoin on Easy Mode” for Developers and Autonomous Agents first appeared on Bitcoin Magazine and is written by Juan Galt.
AI Seer integrates RealityDetector and FacticityAI to launch ArAIstotle, combining advanced truth verification technologies using $FACY token.
The post ArAIstotle integrates RealityDetector and facticityai technologies appeared first on Crypto Briefing.
Teacher unions in the US are integrating DEI in AI education, potentially paving the way for decentralized AI technologies in schools.
The post Teacher unions embed DEI in AI classroom guidance appeared first on Crypto Briefing.
The White House accused Moonshot AI of covertly distilling Anthropics Fable model to develop Kimi K3 through a concealed internal platform.
The post White House accuses Moonshot AI of using Anthropic’s Fable to build Kimi K3 appeared first on Crypto Briefing.
Drip introduces a micropayment model for AI to pay creators in USDC, challenging traditional subscriptions while focusing on financial content.
The post Drip empowers AI agents to pay creators, sparking a content monetization shift appeared first on Crypto Briefing.
Discover Claude's new 'Teach a Skill' feature for Pro, Max, and Team plans, redefining AI interactions and automating workflows.
The post Claude launches skill-teaching feature for pro, max, and team plans appeared first on Crypto Briefing.
Render Completes 98% Of Solana Migration As RENDER Replaces RNDR
Render Foundation says 98.4% of token supply has now migrated from Ethereum-based RNDR to native RENDER on Solana, bringing one of the network’s most important infrastructure transitions close to completion.
The migration shifts render task settlement onto Solana’s high-throughput rails. For a project focused on decentralized GPU rendering, that matters because speed, transaction cost, and network efficiency can affect how smoothly compute-related jobs are coordinated and paid for.
Render has long sat at the intersection of crypto, AI, GPU infrastructure, and decentralized compute. Moving almost all token supply to Solana gives the project a cleaner base for future network activity.
The remaining unmigrated supply is described as largely inactive cold storage, meaning the active market has mostly completed the transition.
Render’s migration to Solana was about performance.
A decentralized rendering network needs to coordinate jobs, payments, and participants efficiently. If transaction costs are high or settlement is slow, the user experience suffers. Solana’s low fees and fast confirmations make it attractive for networks that expect frequent interactions.
For Render, that matters because the project is not just a token. It is infrastructure for distributed GPU rendering.
As AI and graphics workloads grow, demand for compute infrastructure has become one of the most important themes in tech and crypto. Render’s pitch is that unused GPU capacity can be coordinated through a decentralized network.
That model needs a blockchain layer that can handle activity without creating too much friction.
Solana gives Render a faster settlement environment than Ethereum mainnet.
Token migrations can sound cosmetic, but they are often operationally important.
Moving from RNDR to native RENDER changes where the token lives, how it settles, and how users interact with the network. Exchanges, wallets, custodians, holders, and applications all need to support the transition.
A 98.4% migration rate suggests the process is nearly complete.
That reduces fragmentation between old and new token versions. It also gives the ecosystem more confidence that future integrations can focus on Solana-native RENDER rather than supporting a split supply across different formats.
The remaining inactive supply still matters, but it is less disruptive if most active holders and infrastructure have already migrated.
Render’s migration is also a win for Solana.
The network has worked to attract serious infrastructure projects, not just meme-token trading. Render gives Solana exposure to decentralized compute, GPU markets, AI workloads, and creator infrastructure.
That helps broaden Solana’s narrative.
A chain becomes more credible when it supports multiple types of activity: DeFi, payments, stablecoins, gaming, NFTs, AI infrastructure, and real applications. Render fits into the AI and compute side of that story.
For Solana, the question is whether projects like Render generate sustained transaction activity and user demand.
If they do, Solana’s role expands beyond trading and retail speculation. It becomes a settlement layer for more diverse applications.
The migration milestone is positive, but Render still has larger challenges.
Decentralized compute is a competitive market. Centralized cloud providers are powerful. Specialized GPU marketplaces are growing. AI infrastructure demand is huge, but users still care about reliability, pricing, performance, and ease of use.
Render needs to prove that its decentralized model can compete in that environment.
A smoother Solana-based settlement layer helps, but it does not solve every business question. The network still needs demand from creators, developers, AI users, and enterprise workloads.
Token migration is infrastructure. Adoption is the real test.
Still, completing nearly all of the migration removes a major transition risk. It gives Render a cleaner technical base and reduces uncertainty for holders and ecosystem partners.
For RENDER, the next phase is about proving that the Solana move improves the network’s utility.
If it does, the migration may be remembered as a meaningful step in connecting crypto rails with real compute demand.
This article is based on Render Foundation materials.
This article was written by the News Desk and edited by Samuel Rae.
This report is based on information released in official primary source disclosures at primary source documentation.
There’s something to be said about the reality of how helpful AI can be within running a business, and as someone who has been on the hunt to discover this for themselves, I speak from experience when I say that the path to AI-run success is not exactly what the internet makes it out to be.
A few months ago, as a business owner myself, I became consumed with the idea of building something truly special with AI assistance a real, legitimate business with actual revenue generation but I couldn’t find any real information on what the actual process of doing this looked like
Most articles written on the subject were either promotional pieces for SaaS tools or sensationalist opinion pieces focused on how the technology could replace humans in the workplace; no one actually seemed to want to discuss the day-to-day reality of building an AI-driven company and what it actually meant to run such an entity. Thus, I set out to answer these questions for myself, and in this piece, I’ll do my best to relay what I learned without pushing any particular tool or method as being somehow superior to the rest.
What Does it Mean to Build a Company with AI?
When I set out on this journey, I didn’t have a business plan. I had an idea, a hope, really that perhaps a single individual with the right tools could be able to achieve the same results as a small, mid-level team.
I had seen too many companies waste money and resources on tasks that a competent individual could complete and struggled to find much in the way of viable options for streamlining my own processes at the time, and so that’s what drove me to seek out information on how to build a company with AI assistance in the first place.
The truth is, I didn’t have a particularly strong grasp on what exactly it meant to build a company in this way at first — a majority of people don’t. What I failed to realise at the time was that to build a truly digitalised, AI-driven company meant to adopt a fully digitalised approach to every aspect of my own operations as well.
In practice, this looked like research and validation became a continuous, daily process rather than an event that only happened occasionally — content writing became exponentially faster and far more experimental, as I could churn out ten different headlines for the same article in the same day rather than spending a week agonising over a single one, customer communications became far more immediate and efficient, and administrative tasks that previously bled into my entire day were consolidated into far more manageable chunks that took up, altogether, less than an hour.
It’s important to note, however, that none of this eliminated the need for judgement or critical thinking within my own operations — I simply replaced my own repetitive, time-wasting tasks with those that could be automated, which ultimately gets to the root of what I feel is the most important lesson within the entire experience in general.
The Lesson that Not Everyone Talks About
When it comes to actually learning how to build a company with AI assistance, I think the most important lesson to be learned is that AI doesn’t do the work for you it only removes the roadblocks that previously kept you from doing the work you needed to do at an acceptable pace.
My early experiments with AI were riddled with failure emails that I sent to prospective clients had been generated by AI and were immediately off-putting due to their robotic nature; market research conducted by my chatbots turned out to be wildly inaccurate when cross-referenced with actual data from other sources, etc. In the end, none of this was actually AI’s fault it simply reflected that people who are using these tools tend to make the same mistakes over and over again, and those mistakes are usually process-related. If you rush through the process of generating content, you’ll end up with content that sounds rushed. If you fail to fact-check your research, you’ll end up with research that’s full of easily avoidable errors.
So, Can You Actually Build a Company with AI Assistance?
Yes, but not in the way that most people seem to think
You can’t magically outsource your own judgement or decision-making to some magical algorithm, but what you can do is automate the drudgery of actually executing on an idea to let your brain focus on the more important tasks that require thought. The companies that have successfully utilised this method in practice have done so by building proper feedback loops into their processes and never putting out any work without first double-checking the AI’s work to ensure it meets their standards.
In essence, these companies know that the best way to use AI assistance is to treat any work that comes out of it as a first draft of something that will eventually need to be reviewed by a human being — this way, they’re able to maintain quality control throughout their operations while still enjoying the benefits of increased productivity and decreased overhead. It’s a delicate balance, but with the right approach, it’s entirely possible.
What Would You Say to Someone Considering This Approach?
If you’re reading this article, odds are you’re considering learning how to build a company with AI assistance one day, and so I urge you to think carefully about exactly what it is you hope to accomplish before investing too much time into learning the ins and outs of these tools.
Above all, I think it’s important that you realise that the end goal should always be a company that only requires your own brainpower to make decisions. Ideally, most of your daily tasks should be automated so that you only have to spend minimal amounts of your own time on them throughout the day.
Start small: take one repetitive task from your own operations and plug it into an AI-powered program before you try to scale up and automate your entire business at once and be prepared to spend some time troubleshooting before you begin seeing results. After all, the companies that will end up succeeding in this space aren’t the ones that claim to be “100% AI-run,” but rather the ones that use the technology as a tool to move faster than everyone else while keeping a close eye on what needs to be improved in their own operations.
This is the reality of learning how to build a company with AI assistance, and I hope that this piece serves not as some magical end-all-be-all guide, but rather an informative look at the actual process that people rarely ever seem to talk about, as well as an actionable starting point for anyone who wants to begin exploring the benefits for themselves.
I Tried Building a Company with Only AI was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.
I keep seeing the same debate pop up: is Claude smarter than Gemini, is Chat GPT still ahead, whatever. Honestly? Wrong question entirely. The stuff that’s actually going to matter is happening quietly, in places most people aren’t even looking.
I’ve been using these tools since they were basically novelties the kind of thing you showed your coworkers as a party trick. Ask around now and most people will tell you the future is “better answers” or “smarter writing.” That’s not really where this is going.
The bigger shift is in what these things fundamentally are, not how well they perform on some benchmark. Here’s my read on it, based on where the money and the engineering effort have actually been going.
Right now you type a question, you get an answer, that’s the whole interaction. That model has an expiration date on it.
The next phase is AI that actually does things instead of just describing them: books your flight, cleans up your spreadsheet, pushes a code fix. This isn’t a prediction; it’s already happening in early form. The labs have shipped versions of this that can browse the web, click through interfaces, run code.
What’s holding it back isn’t capability, it’s trust. Nobody wants software that deletes the wrong file or emails the wrong person by mistake. So a lot of what’s coming isn’t going to be flashier intelligence — it’s going to be boring stuff like permission systems, confirmation steps, undo buttons. The unglamorous plumbing that makes people comfortable handing over real responsibility.
Most AI still forgets you exist the second you close the tab. A few companies have bolted memory features on top, but it’s early.
What’s coming is assistants that actually track your ongoing projects and how you write and what you keep running into problems with — without you re-explaining your whole situation every single time. That’s genuinely useful. It also raises uncomfortable questions about data retention and consent. My guess is the tools that win here won’t just remember more — they’ll let you actually see what’s stored and delete it, rather than just saying “trust us.”
“It can look at pictures now” used to be a headline feature. Soon that’ll just be table stakes. Voice, video, live camera feeds — these are going to merge into one conversation rather than sitting in separate menus you have to hunt for.
Point your phone at something broken, get spoken help back instead of typing out three paragraphs describing the problem. This stuff already exists in rough form. What’s actually improving is speed and reliability, not whether it’s possible at all.
There’s a whole separate competition happening that has nothing to do with which model tops the leaderboard. It’s about which company can get something genuinely useful running on your phone without needing a data center behind it.
On-device matters because it’s faster, it’s private, and it’s cheaper to run. Expect a split forming — giant models for heavy lifting, small efficient ones baked directly into your phone for everyday tasks.
Most assistants sound pretty interchangeable right now — competent, a little bland. That’s going to change. Some will stay blunt and no-nonsense. Others will lean warm, or get tuned specifically for law or medicine or teaching.
This matters more than it sounds like it should, because tone is tied directly to trust, and trust is what decides whether someone actually uses this thing for something that matters health, money, their kid’s homework.
This is the part that gets ignored in most of these takes. Governments in the US, EU, and across Asia are actively writing the rules right now around transparency, copyright, data use. These aren’t theoretical debates. They decide what actually ships.
Expect more labeling on AI-generated content, clearer ways to opt out of training data, tighter restrictions around healthcare and hiring and anything involving kids. The companies that get ahead of this instead of fighting it are probably going to end up with an advantage that outlasts a few missed product launches.
Benchmark scores make for good headlines. They don’t decide who actually wins long-term. What decides that is whether people trust a tool enough to hand it something real.
That trust gets built through consistency and honesty about limitations and through how a company handles it when something breaks. An assistant that says “I’m not sure” when it isn’t sure will probably earn more loyalty over years than one that scores a point higher on some test nobody outside a research lab has heard of.
Not some dramatic leap forward. More like a slow accumulation of smaller changes tools that remember more, act more on their own, run faster locally, and get shaped as much by regulators as by engineers. What you’re using today is a rough draft, not a finished product.
The real race isn’t about who has the smartest model. It’s about who builds something boring enough, reliable enough, that you stop noticing you’re even using it.
Curious what you think — five years from now, do these feel more like tools to you, or more like teammates? Drop your take below.
Everyone’s Asking the Wrong Question About AI Chatbots was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.
Intelligence does not have to be artificial.
The goal of this field was always to reproduce what a brain does. Somewhere along the way “artificial” stopped meaning inspired by the real thing and started meaning nothing like it — enormous, power-hungry, and opaque.
You’re tired of AI launches and IPOs? So am I. Every week there’s a bigger model, a longer context window, another benchmark nobody outside the lab can reproduce — and underneath it, the same machine doing the same thing a little faster and a lot more expensively. I mean, just looking at my emails these days is making me nauseous. I do not even check my social media anymore, and even less the stock market.
But, instead of complaining and be satisfied with the status quo, I decided to look at the problem from a different angle.
The main problems everybody knows without knowing it…
The cost problem isn’t separate from the design. It falls out of four choices that the field made early and never really revisited.
1- It reasons in the dark. Which makes hallucination or fake generation very hard to catch, yet to fix. Hidden states are well, hidden.
2- Scale is not intelligence. The reflex has been to make the model bigger and hope understanding shows up (it never will, the bigger the model, the more “links” it can do between concept and give the illusion of understanding). Scale = $$$$$$$$$$$$$$$.
3- Biology as the last of their concern. The brain runs on about twenty watts, and that number is a challenge, not a footnote. While we cannot make an AI that works on 20watts we can definately reduce the amount of energy consumption.
4- The root of it is profit. Not science. Even OpenAI leader is confirming it by saying that AI will eventually be sold like electricity and water — by companies like OpenAI. Article link: https://www.businessinsider.com/sam-altman-ai-utility-electricity-water-openai-2026-3
The goal of this field was always to reproduce what a brain does. Somewhere along the way “artificial” stopped meaning inspired by the real thing and started meaning nothing like it — enormous, power-hungry, and opaque.
I think we need to take the biology seriously instead of metaphorically: real neural mechanisms, a memory that consolidates the way a hippocampus does, a neurochemistry that actually modulates behaviour, learning that happens as the system runs rather than only in an offline training run. Those are design constraints, not decoration. And will lead to the “second generation” of AI.
a very convenient one if what you need is a reason to keep raising money.
While I have been plain, here’s where I don’t stand: AGI. The industry’s favourite three letters do a lot of quiet work — a general, human-beating machine, forever a few years and a few hundred billion away. It’s a wonderful story — or a frightening one, depending on where you stand — and a very convenient one if what you need is a reason to keep raising money. It’s a poor description of what these systems actually are, and a worse goal to organise a field around.
It’s a poor description of what these systems actually are, and a worse goal to organise a field around.
And the way today’s models are built won’t get there — not for lack of ambition, but for reasons you can put numbers on. Large language models improve along a scaling curve, and that curve has a shape: the returns diminish. Each new increment of capability takes not a little more compute but multiples more; the graph everyone cites bends the wrong way, flattening as the bill climbs. Every training run costs more than the last and buys less than the last one did. That isn’t a detail better engineering erases. It’s the shape of the method itself.
Every training run costs more than the last and buys less than the last one did. That isn’t a detail better engineering erases. It’s the shape of the method itself.
Now set that against a hard limit: power is finite. You can’t answer a curve of exponentially rising cost with an infinite supply of energy, because there isn’t one. A method whose only real lever is “make it bigger” runs into a wall that isn’t philosophical — it’s thermodynamic. Somewhere on that curve the next run stops being affordable, then stops being physically possible, long before it stops being merely better at text.
You don’t get a different kind of thing by making the same thing bigger
And that’s the deeper point: what scales here is fluency, not understanding. A model trained to predict the next word learns the statistics of language extraordinarily well. It doesn’t thereby acquire a grounded model of the world, a cause it can reason about, or a memory it can update — and no amount of the same training conjures those out of more of the same text. You don’t get a different kind of thing by making the same thing bigger. You get a costlier version of the same thing. A transformer is, underneath, a very good text generator; scale it and you get a better text generator — not a mind that understands, and not consciousness quietly emerging from the weights. Fluency is not comprehension, and no quantity of the first ever becomes the second. Something like general intelligence, if it’s reachable at all, will come from a different design — grounded, able to reason step by step, able to learn as it runs.
The point of this work was never to conjure a god
The point of this work was never to conjure a god. It was to build something genuinely useful — that reasons, remembers, and helps — and to run it on hardware people can actually afford. Intelligence doesn’t have to be general to be worth having, and it certainly doesn’t have to be a superbeing to earn its keep. Chasing AGI is how you end up with the bill on the other pages. Building something useful, efficient, and yours is how you don’t.
What a discovery is for, and how it gets used, stays a human call — the machine widens what we can see; the judgment is still ours.
None of this means the tools are useless — the opposite. An AI can read across billions of documents and surface a link between two of them in seconds, connections no person would ever stumble on alone. That is a genuinely powerful research instrument, and we build with it every day. But it won’t know what to do with what it finds unless someone told it beforehand what to look for and why. Finding is not deciding. What a discovery is for, and how it gets used, stays a human call — the machine widens what we can see; the judgment is still ours.
If one ever goes autonomous and causes genuine harm, it will be because a person somewhere pointed it that way —
Some people will tell you AI is the real long-term danger. We’d put it the other way around: the danger is us. A model does what it is built and instructed to do. If one ever goes autonomous and causes genuine harm, it will be because a person somewhere pointed it that way — wrote the objective, wired it to something it should never have touched, or pulled out the guardrails that other people had put there in the first place. Even the runaway story needs a human at the start of it: someone to build it, aim it, and take it off the leash. Even if it escapes, a human had to set it loose or dare it to.
That isn’t a reason to be careless — it’s the opposite. It means the responsibility is ours and stays ours, which is exactly why we should keep the reasoning legible and the controls somewhere a person can see them. A tool you can read is a tool you can hold to account. That matters far more than pretending the machine has a will of its own.
Now time for a little shameless self-promotion ;) I built Grillcheese Research Laboratory exactly to study, learn and solve those problems and share how to do it with as much people as possible. I invite you to check the link to our website if you are curious. https://grillcheeseai.com
Let me know in the comment what you think and if you have more ideas / different views / links.
Thanks for reading and have a wonderful day!
Yours, Nick
Beyond A.I. was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.
ChangXin Memory Technologies, better known as CXMT, is preparing for one of the most closely watched semiconductor listings of 2026.
At the same time that China’s public markets are establishing an official price for the company’s shares, on-chain traders have begun forming a separate view of what CXMT could be worth after its listing.
The result is an unusual experiment in global price discovery.
One market is selling regulated equity through a formal IPO on Shanghai’s STAR Market. The other is trading a perpetual contract linked to expectations surrounding the company before its public debut.
They are not the same asset. They do not provide the same rights. But together, they reveal how traditional financial events are increasingly becoming tradable on-chain narratives.
With Ave.ai integrating Hyperliquid perpetual markets, users can now spot emerging contracts such as the on-chain CXMT perp alongside crypto assets, tokenized market opportunities, stock-related contracts and other real-world trading themes.
This is not simply another market listing. It reflects a much larger shift in how traders discover and price global assets.
CXMT is one of China’s most important semiconductor companies and a leading domestic producer of dynamic random-access memory, or DRAM.
DRAM is a critical component in computers, smartphones, data centers and AI infrastructure. The global market has historically been dominated by Samsung Electronics, SK Hynix and Micron, making CXMT’s rapid development strategically significant for China’s semiconductor ambitions.
The company priced its Shanghai STAR Market IPO at 8.66 yuan per share. CXMT is expected to raise approximately 57.9 billion yuan, or US$8.5 billion, by selling nearly 6.7 billion shares. The offering implies a post-listing valuation of about 579 billion yuan, or US$85.2 billion.
If the overallotment option is fully exercised, the offering could raise as much as approximately US$9.8 billion. The deal is positioned to become the largest A-share IPO completed by a Chinese semiconductor company.
The scale of the offering reflects more than investor demand for another technology stock. CXMT sits at the intersection of several major themes:
Reuters has described CXMT as China’s DRAM champion, while the company’s listing is expected to rank among Asia’s largest share sales of 2026.
But before the company’s shares begin trading publicly, a separate market has already started expressing an opinion.
A Hyperliquid HIP-3 ticker representing CXMT was reportedly acquired for 500 HYPE, with plans to introduce a CXMT pre-IPO perpetual market.
This means crypto-native traders do not necessarily need to wait for the official Shanghai listing before taking a position on market expectations surrounding CXMT.
However, the distinction is critical:
The on-chain CXMT perpetual is not CXMT stock.
Buying CXMT shares through the Shanghai IPO gives an investor formal ownership in the publicly listed company, subject to the rules, eligibility requirements and settlement structure of China’s securities market.
Trading a CXMT pre-IPO perpetual gives the trader exposure to a derivatives contract whose price reflects market expectations. It does not provide equity ownership, shareholder voting rights, dividend rights or access to the official IPO allocation.
Reports indicate that the CXMT HIP-3 ticker was acquired for 500 HYPE and prepared for launch in a pre-IPO market segment.
The difference can be summarized simply:
These markets should not be treated as substitutes. They represent two different forms of price discovery.
CXMT’s official IPO price of 8.66 yuan was established through a regulated offering process involving the issuer, underwriters, institutional demand and exchange requirements.
The on-chain market works differently.
Perpetual traders continuously submit bids and asks based on their expectations of CXMT’s future value. Their decisions may incorporate the IPO price, expected first-day performance, comparable-company valuations, semiconductor demand, AI-related sentiment and short-term speculation.
One market asks:
What price should CXMT use to issue its shares?
The other asks:
Where might the market value CXMT once trading begins?
That distinction makes pre-IPO perpetual markets especially interesting — but also especially risky.
There may be limited liquidity, uncertain reference prices, rapidly changing settlement expectations and large gaps between bids and asks. A quoted perpetual price cannot automatically be translated into a reliable corporate valuation.
For example, reports of large CXMT bids on Hyperliquid generated theoretical valuation comparisons far above the official IPO valuation. But those figures were based on pre-IPO derivative orders rather than completed equity transactions, and should not be interpreted as definitive market capitalization.
In other words, the on-chain market can be informative without necessarily being accurate.
It captures expectations, positioning and speculation in real time. It does not replace formal valuation work.
The emergence of CXMT on Hyperliquid is possible through HIP-3, Hyperliquid’s framework for builder-deployed perpetual markets.
HIP-3 allows qualified deployers to create and operate new perpetual markets. The deployer is responsible for defining the market, selecting the oracle structure, establishing contract specifications, setting leverage limits and managing settlement when required.
This model expands the range of assets that can potentially become tradable on-chain.
Historically, crypto perpetual markets concentrated on digital assets such as Bitcoin, Ethereum and major altcoins. Builder-deployed markets make it possible to explore contracts connected to a wider universe:
Hyperliquid currently presents itself as a fully on-chain, non-custodial venue supporting hundreds of spot and perpetual markets across crypto and other asset categories.
CXMT demonstrates what happens when permissionless market creation meets a major global IPO.
The market can begin forming expectations before traditional public trading officially starts.
The challenge for on-chain traders is no longer simply gaining access to more markets.
It is discovering the right market at the right time.
New contracts frequently appear across different protocols, chains, interfaces and market operators. Traders may need to move between social media, analytics dashboards, block explorers, wallets and decentralized exchanges before they can even understand what is available.
Ave.ai is addressing this fragmentation by integrating Hyperliquid perpetual trading into its broader on-chain platform.
Ave Wallet Pro’s iOS perpetual DEX integration allows users to access Hyperliquid market data, manage assets and interact with perpetual markets through a mobile on-chain trading experience.
For users following CXMT, this means the emerging on-chain perpetual can be discovered within the same ecosystem they already use to explore other trading opportunities.
Through Ave.ai, traders can increasingly move across multiple market categories:
Ave.ai’s main platform already combines real-time blockchain data, wallet monitoring, smart-money tools, price alerts, copy trading and trading interfaces. It reports integrations across more than 130 blockchains and 300 decentralized exchanges.
Adding Hyperliquid perps expands that model beyond traditional crypto-token discovery.
Users can now spot an emerging market such as the CXMT perpetual without treating stock narratives, on-chain derivatives and crypto trading as completely separate worlds.
Ave.ai has historically been strongly associated with meme-coin discovery, on-chain analytics and early token opportunities.
Its expansion into stock-related perps, ETFs and pre-IPO markets may appear to be a change in direction.
A better interpretation is that the definition of an “on-chain asset” is expanding.
Stocks are becoming tokenized. Commodity and equity indices are appearing as perpetual contracts. ETFs are entering blockchain-based trading environments. Private-company expectations are becoming tradable through pre-IPO derivatives.
As more traditional assets move on-chain, the infrastructure originally built for crypto discovery becomes relevant to a much broader financial market.
Ave.ai is therefore not abandoning its original positioning. It is extending the same core capabilities — discovery, analysis and execution — to new asset categories.
The progression is increasingly clear:
Meme coins → Multi-chain assets → Crypto perps → Stock perps → ETFs → Pre-IPO markets
What connects these categories is not their legal structure. It is their growing availability through on-chain infrastructure.
Ave.ai’s role is to make those fragmented opportunities easier to discover and access through one integrated entry point.
CXMT is especially significant because it combines three powerful market narratives.
The growth of AI infrastructure has increased demand for memory chips across servers, data centers and advanced computing systems.
CXMT represents China’s effort to build a stronger domestic memory-chip industry and reduce reliance on foreign suppliers.
The Hyperliquid contract gives crypto-native traders a way to express a view on a major Chinese IPO before the underlying shares begin public trading.
This creates a market that may attract several different groups:
For Ave.ai users, CXMT is not only another ticker. It is an example of how globally important financial events are becoming visible within on-chain trading platforms.
Pre-IPO perpetuals involve substantial uncertainty. Before interacting with a CXMT-linked contract, traders should examine several factors carefully.
Confirm what the contract represents, how its index or oracle is calculated, and what happens when the underlying shares begin trading.
Understand whether the contract continues after the IPO, transitions to a different reference price or settles under specific conditions.
A visible price does not guarantee that a large position can be opened or closed near that level.
Perpetual positions may generate recurring funding payments. Holding costs can become significant when positioning becomes highly one-sided.
Pre-IPO contracts can experience extreme volatility. High leverage may result in liquidation even when the trader’s longer-term thesis is ultimately correct.
The perpetual contract may trade at a substantial premium or discount to the official IPO price. There is no guarantee that the two prices will converge immediately.
Availability may vary depending on a user’s location, platform eligibility and applicable regulations.
The most important part of the CXMT story is not that another perpetual contract has been launched.
It is that an IPO taking place on Shanghai’s STAR Market is simultaneously becoming an on-chain trading event.
Stocks, ETFs, commodities and pre-IPO expectations were once almost entirely confined to traditional financial infrastructure. Today, their price exposure is increasingly being represented through blockchain-based markets.
This transition will not eliminate traditional exchanges. Nor will perpetual contracts replace regulated equities.
Instead, the financial market is developing an additional layer of price discovery — one that operates globally, continuously and on-chain.
Traditional markets establish ownership.
On-chain derivatives establish exposure.
Traditional IPOs allocate shares.
Pre-IPO perpetuals aggregate expectations.
The two systems may coexist, interact and sometimes disagree.
That disagreement is exactly what makes them valuable to watch.
CXMT offers a preview of what the next generation of on-chain trading could look like.
A trader may begin by monitoring a semiconductor IPO, compare its formal offering price with an on-chain perpetual market, examine real-time positioning and then act through a connected trading interface.
With Hyperliquid perpetuals integrated into Ave.ai, users can spot CXMT and other emerging on-chain markets alongside the broader crypto ecosystem.
The opportunity is no longer limited to discovering the next meme coin.
It increasingly includes discovering how the next stock, ETF, commodity or pre-IPO event is being priced on-chain.
As traditional financial assets move onto blockchain infrastructure, platforms that unify discovery, data and execution will become increasingly important.
CXMT may be one of the first major Chinese IPOs to receive meaningful on-chain price discovery before its public debut.
It is unlikely to be the last.
CXMT Is Heading to IPO— But On-Chain Traders Are Already Pricing It was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.
AI can act, but cannot bear responsibility
An AI agent may select a counterparty, negotiate terms, interact with a smart contract and authorise payment. Yet it is not generally recognised as a legal person, therefore its outputs need to be attributed to a human being or organisation. The UNCITRAL Model Law on Automated Contracting, adopted in 2024, supports contracts formed or performed through automated systems, including AI and machine-to-machine transactions. It establishes rules for attributing automated outputs and addressing unexpected outcomes without requiring the system to possess legal personality. And the emerging direction is clear: autonomous execution does not remove human or corporate accountability.
Rome’s architecture of delegated commerce
Roman law distinguished between people who were legally independent (“sui iuris”) and those subject to another’s authority (“alieni iuris”). The “paterfamilias” was the legally independent head of the household and principal holder of its property. He was not a ‘beneficial owner’ in the modern legal sense but can be compared cautiously with a principal asset owner, trustee, company or family office. Nevertheless, commerce required others to manage farms, ships and businesses and so the peculium was a fund placed under another person’s practical administration whilst remaining connected to the principal. The Roman jurist Gaius, Institutes, Book IV, sections 69 to 74, explained that liability depended on the authority granted; where the principal expressly ordered a transaction or appointed someone to operate a business or ship, liability could extend beyond the peculium. In other circumstances, recovery might be limited by reference to that fund. Justinian’s Institutes, Book IV, Title VII later restated this graduated approach and, in today’s climate, the resulting lesson is clear:
The greater the authority given to an AI agent, the greater the potential exposure of the principal behind it.
Four questions for AI transactions
In the case of wallets, a separate wallet does not itself determine authority or liability; asset segregation, attribution and recourse remain distinct questions.
What modern cases tell us
In the case of Quoine Pte Ltd v B2C2 Ltd, algorithms entered cryptocurrency trades after a platform failure activated a fallback price. The Singapore Court of Appeal treated the deterministic programs as mechanisms selected by their human operators, rather than inventing a separate legal mind for the software. The case suggests that using an automated system does not necessarily allow its deployer to disown a resulting contract, with these limits of unchecked automation having been exposed by US global financial services firm, Knight Capital. In 2012, faulty software sent more than four million erroneous orders in forty-five minutes, producing losses exceeding $460 million. Unsurprisingly, the SEC found inadequate safeguards, testing and supervisory controls and imposed a $12 million penalty. The lesson is that an AI peculium needs more than a capped wallet — it requires transaction limits, cumulative exposure controls, approved counterparties, price tolerances and an effective suspension mechanism. Another example can be seen in the case of Moffatt v Air Canada, where a tribunal held the airline responsible after its chatbot gave a customer inaccurate information about bereavement fares. These decisions are not universally binding but illustrates that a business cannot assume its AI interface is legally separate from the organisation deploying it. Meanwhile, the Ooki DAO litigation has provided a related warning — a US court held that a decentralised organisation could be sued as an unincorporated association and treated as a person under the Commodity Exchange Act. Similarly, the SEC’s 2017 DAO Report emphasised that regulatory treatment depends on economic reality, not technological terminology. A wallet, smart contract, DAO or SPV may segregate operations but it cannot automatically override securities law, sanctions obligations, consumer protection or fiduciary duties.
Why England and Wales could lead
The Law Commission has concluded that the law of England and Wales can generally support smart legal contracts without wholesale statutory reform. It also identified areas requiring further attention, including deeds, jurisdiction, interpretation and remedies. The Property (Digital Assets etc) Act 2025 has further confirmed that digital or electronic assets are not prevented from being objects of personal property rights merely because they fall outside the traditional categories of things in possession and things in action. That improves certainty over digital property but it does not determine who is responsible when an AI transfers it. The commercial opportunity is to combine existing contract, property, trust, company and financial-services law with a technically enforceable AI mandate.
Building a modern peculium protocol
A modern AI peculium should be a legal and technical control framework where it would identify the principal and define the AI’s objectives, permitted assets, counterparties, jurisdictions and transaction types in a digitally signed mandate. Capital could be placed in a segregated wallet or account and smart-contract permissions would impose per-transaction and cumulative limits. Borrowing, pledging assets, using an unapproved protocol or exceeding a threshold would require human authorisation and instructions, data sources, decisions and transactions would be logged so the agent’s conduct could be reconstructed. Lawyers, trustees, directors, compliance officers or regulated custodians could validate authority, approve exceptional actions, preserve evidence and activate emergency suspension and insurance could then be priced against a measurable mandate and maximum exposure. Furthermore, ring-fencing would still have limits as it could not automatically exclude claims arising from fraud, negligence, sanctions breaches, regulatory violations, fiduciary misconduct or express authorisation by the principal. This all echoes Rome where liability depended not only on the assets allocated, but also on what was ordered, who benefited and how much authority had been granted.
The EU AI Act requires proportionate human oversight for high-risk systems, including the ability for authorised people to intervene or stop systems that are not operating as intended. The UK’s principles-based framework emphasises safety, transparency, accountability, governance and redress; both approaches point toward controlled autonomy rather than artificial personhood.
Autonomy without unaccountability
Roman law did not solve AI governance two thousand years in advance. It did, however, recognise that commerce could be delegated without leaving authority and liability undefined. AI agents do not need fictional personhood to contract and move value — they need intelligible mandates, restricted access to assets, transparent records, effective human control and credible recourse. Jurisdictions that build this architecture first could provide the trusted infrastructure through which autonomous commerce, machine-to-machine payments and AI-managed wealth operate at scale. Rome’s enduring lesson is that delegation becomes commercially useful only when authority, assets and accountability have clearly defined boundaries.
Rome’s 2000-year-old answer to AI liability: give the agent a budget, not legal personhood was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.
Bitcoin Magazine
Bitcoin Miner Hut 8 Shares Jump on $9.8 Billion AI Data Center Deal
Bitcoin miner Hut 8’s shares rose Monday after the Toronto Stock Exchange- and Nasdaq-listed firm said it had signed a second 15-year lease worth $9.8 billion for its AI data center.
Hut 8 shares peaked as high as $106 a pop before dropping to around $101. They closed Monday up over 10%.
The deal will see the Toronto-based firm’s Beacon Point campus in Texas data center cover 352 megawatts of IT capacity. The tenant using the data center’s will have its capacity doubled to 704 MW.
Hut 8 added that the campus has a base-term contract value of $19.6 billion over 15 years, rising to as much as $50.2 billion if renewal options are exercised.
Asher Genoot, CEO of Hut 8, said: “The real test of our power-first approach is what our partners are willing to commit against it. Our tenant at Beacon Point chose to double its footprint at the site, the strongest validation an asset can receive.”
Hut 8 last year signed a deal with American Data Centers Inc., a company backed by President Donald Trump’s sons Eric and Donald Jr., to contribute its Bitcoin mining equipment and help debut their American Bitcoin mining firm.
Hut 8 is one of a number of top publicly listed miners that have started directing resources to providing the infrastructure for high-powered computing.
The company in December secured a Google-backed partnership with Anthropic and Fluidstack to build up to 2.3 gigawatts of AI data center capacity in the U.S.
JUST IN: #Bitcoin mining company Hut 8 just announced it partnered with Google for financial backing on a 15-year lease.
— Bitcoin Magazine (@BitcoinMagazine) December 17, 2025
Bullishpic.twitter.com/NQN9JmW0ob
A number of Bitcoin miners have already gone all-in on the industry as minting the biggest digital coin by market cap becomes harder and demand for AI compute surges.
As the price Bitcoin has dipped, it has become harder for Bitcoin miners to make ends meet.
Nasdaq-listed Bitfarms last year announced that it would wind down mining operations to focus on high-performance computing.
Instead of dropping mining operations completely, a number of Bitcoin miners have instead marketed themselves as “compute” or “digital infrastructure” companies while switching between minting digital coins and providing compute for AI — depending on which is more profitable.
Top miners Terawulf, IREN, and Cipher Mining all last year signed multi-year HPC contracts with Alphabet Inc.’s Google and Microsoft.
Both the crypto mining and HPC industries require huge amounts of energy and data centers — but the move isn’t always easy: AI data centres require more expertise than Bitcoin mining.
This post Bitcoin Miner Hut 8 Shares Jump on $9.8 Billion AI Data Center Deal first appeared on Bitcoin Magazine and is written by Mathew Di Salvo.
Login