Crypto World
What is proof of personhood? Verifying real humans in the AI age
As AI floods the internet with convincing fake humans, proving that a user is a real, unique person is becoming one of crypto’s hardest and most valuable problems. This guide explains what proof of personhood is, how the leading approaches work, and why the cure raises concerns of its own.
Summary
- Proof of personhood aims to verify that each real person can obtain only one identity while protecting their privacy.
- The technology has gained urgency as AI makes it easier to create convincing fake identities that can exploit voting, airdrops, and online platforms.
- Biometric systems, social trust networks, and zero knowledge identity methods offer different ways to verify unique humans, each with its own tradeoffs between privacy, security, and scalability.
Proof of personhood is a cryptographic mechanism that lets someone prove they are a real, unique human being, one person counted exactly once, without revealing who they actually are. That combination is what makes it both powerful and difficult: it must guarantee uniqueness, so that a single person cannot register as a thousand, while preserving anonymity, so that proving you are human does not force you to expose your identity. The problem it solves is old, but it has become urgent for a new reason. For most of the internet’s history, telling humans from machines was a minor nuisance handled by simple puzzles.
Now, with artificial intelligence able to generate text, images, voices, and entire online personas indistinguishable from a real person’s, the open internet faces a verification crisis: bots can flood platforms, manipulate votes, drain airdrops, and impersonate humans at a scale and quality never seen before. This guide explains what proof of personhood is, the attack it defends against, the main approaches to building it, the leading real-world example and its controversies, and why a technology meant to protect humanity raises hard questions of its own.
The reason this topic has moved to the center of crypto and beyond is that “one real human, counted once,” turns out to be a foundational requirement for a surprising range of things. Fair token airdrops depend on it, or a handful of people with thousands of fake accounts will scoop up everything meant for a community. Democratic voting and decentralized governance depend on it, or whoever can spin up the most identities wins.
Any system that distributes scarce resources to people, from community rewards to the long-discussed idea of a universal basic income, depends on being able to tell one person from a thousand sock puppets. And increasingly, the world of artificial intelligence depends on it, both to keep bots out of human spaces and, in a twist, to let trustworthy AI agents act on behalf of verified humans.
Proof of personhood sits at the intersection of cryptography, identity, and the defining technological anxiety of the moment, which is why it has become one of the most watched and most contested ideas in the field.
The sybil attack: the problem at the root
To understand proof of personhood, you first have to understand the attack it exists to stop, which is called a sybil attack. The name comes from a famous case study of a person with many personalities, and in computing it describes a single actor creating many fake identities to gain influence they should not have. On a network where one identity equals one vote, one share, or one claim, a sybil attacker who controls a thousand identities controls a thousand times the influence of an honest participant who has just one. Almost every open online system that tries to be fair, every vote, every giveaway, every reputation score, every “one person, one share” distribution, is vulnerable to someone who can cheaply manufacture identities.
Historically, sybil attacks were limited by the friction of creating convincing fake accounts at scale, and by crude defenses like puzzles meant to slow bots down. Artificial intelligence demolishes both limits. Modern systems can generate unlimited unique-looking personas, complete with plausible writing, profile photos, and behavior, and can solve the puzzles that once filtered them out.
The very technology that makes AI useful, its ability to produce human-like content, is what makes it the ultimate sybil weapon, capable of populating the internet with armies of fake humans cheaply and convincingly. This is the deeper reason proof of personhood has surged in importance: the old, informal defenses against sybil attacks have broken down precisely when the cost of mounting one has collapsed. If you cannot tell a real, unique human from a generated one, then every system that assumed it could is suddenly exposed, and rebuilding a reliable way to prove humanness becomes foundational infrastructure rather than a nice-to-have.
What a good proof-of-personhood system must achieve
Before looking at how anyone builds proof of personhood, it helps to define what success even requires, because the requirements pull against each other, and that tension shapes every design. A strong system needs to satisfy several properties at once. It must guarantee uniqueness, ensuring each real person can obtain exactly one verified identity and cannot register many. It must preserve privacy, so that proving you are a unique human does not force you to reveal your name, your face, or a linkable record of everything you do. It must resist attack, holding up against sophisticated adversaries, increasingly AI-powered, trying to fake or duplicate humanness. And ideally it must scale to billions of people across every country, language, and level of access, without excluding those who lack documents or technology.
The difficulty is that these goals are in tension. The strongest way to guarantee uniqueness is usually to collect something deeply personal and hard to fake, like a biometric, but collecting biometrics is exactly what threatens privacy and raises ethical alarms. The most privacy-preserving approaches, which avoid collecting sensitive data, often struggle to guarantee uniqueness or to resist a determined attacker.
Scaling to everyone on earth conflicts with the careful, high-assurance verification that strong uniqueness demands. Every proof-of-personhood design is, in effect, a particular set of compromises among uniqueness, privacy, security, and inclusivity, and there is no design that maximizes all four at once. Understanding a given system, therefore, means asking which of these properties it prioritizes and which it sacrifices, because that choice, more than any technical detail, determines what the system is good for and what it puts at risk.
The main approaches to proving humanness
There are several broad families of proof-of-personhood design, each making a different bet about how to balance those competing goals. The first and most discussed is biometric verification, which uses a physical trait of the human body, an iris, a face, that is hard to fake and naturally unique, to guarantee one person equals one identity. The bet here is that specialized hardware reading a unique biological signal is the only approach robust enough to resist an adversarial, AI-saturated environment, because you cannot generate a real human iris with a language model. The strength is powerful uniqueness; the cost is the privacy and ethical weight of collecting biometric data and the need for physical hardware and in-person enrollment.
A second family is the social-graph approach, which builds humanness through webs of trust: real people vouch for other real people, and the network of mutual verification makes it hard for a lone attacker to fake many identities, because each fake one would need real humans willing to vouch for it. This avoids collecting biometrics and leans on human relationships instead, but it can struggle to scale and to resist a well-resourced attacker who infiltrates the graph. A third family relies on credentials and accumulated signals, combining evidence like existing verified accounts, on-chain history, or government documents into a score or a passport that suggests a unique human without a single biometric gatekeeper.
This is flexible and privacy-conscious but generally offers softer guarantees of uniqueness than a biometric. A fourth, emerging family uses zero-knowledge identity techniques, proving facts about yourself, that you are an adult, that you are a unique holder of some credential, without revealing the underlying data, and increasingly leans on device-based passkeys and similar tools. Each family is a different answer to the same question, and the field has not settled on a winner, because each answer sacrifices something the others preserve.
The leading example: World and the Orb
The most prominent attempt to build proof of personhood at global scale is the project now called World, formerly Worldcoin, created by a company co-founded by the chief executive of a leading artificial intelligence lab alongside other founders, and launched in 2023. World made the boldest possible bet on the biometric approach, and examining it concretely shows both the promise and the problems of the whole field. Its centerpiece is a custom hardware device called the Orb, a polished sphere that scans a person’s iris.
The reasoning is that the iris is highly unique and extremely hard to forge, so an in-person iris scan is a strong way to guarantee that each verified human is counted exactly once, even against AI adversaries that can fake almost anything made of pixels but cannot fake a living eye on demand.
The privacy design is central to World’s pitch, because iris scanning sounds alarming and the project knows it. According to the project, when the Orb scans your iris it generates a unique cryptographic code, deletes the actual image after processing, and distributes only anonymized fragments of the code across a network to confirm you have not enrolled before.
The result is meant to be a credential, called a World ID, that proves you are a unique human without revealing your identity or storing your biometric image, with zero-knowledge techniques letting you later prove “I am a verified unique human” to an app without exposing anything else. The project reports a scale no other proof-of-personhood effort has reached, on the order of millions of people verified through Orbs and a widely used identity app, which is a meaningful achievement for a category that has historically struggled to grow. World is, in short, the clearest real-world test of whether the biometric approach can become global infrastructure, and its trajectory, successes and backlash alike, is where the abstract debate over proof of personhood becomes concrete.
The AI age and the pivot to verifying agents
What has thrust proof of personhood from a niche idea into a mainstream conversation is the arrival of capable artificial intelligence, and the relationship between the two is closer than it first appears. The same advances that make AI able to flood the internet with fake humans also make a reliable proof of humanness more valuable, because humanness is becoming the scarce, trustworthy thing in a sea of synthetic content. This is why a figure deeply associated with frontier AI is also behind the leading proof-of-personhood project: one venture helps create the problem of indistinguishable machine-generated humans, and the other proposes the verification layer to manage it. As AI-generated text, images, video, and behavior become impossible to tell from the real thing, a system that can certify “a unique human is behind this” turns into foundational infrastructure for trust online.
There is a striking twist in how the field is now evolving. Proof of personhood started as a way to keep bots out of human spaces, but it is increasingly being repurposed to let AI agents operate responsibly within human systems. As autonomous AI agents begin acting on people’s behalf, making purchases, sending messages, executing tasks, a new question arises: which human is this agent acting for, and is that human real and accountable? Proof-of-personhood projects have begun building tools that tie an AI agent to a verified human principal, so that an agent can prove it represents a genuine, unique person rather than running wild as an anonymous bot.
The leading project has also pivoted toward enterprise use, selling proof-of-humanity verification to companies, video platforms, and identity providers that want high assurance a user is real, while keeping the service free for the individuals being verified. The through-line is that AI did not just create demand for proving humans are human; it is reshaping proof of personhood into a layer that governs both humans and the machines acting for them.
Where proof of personhood actually gets used
It is easy to treat proof of personhood as an abstraction, so it helps to ground it in the concrete situations where a reliable proof of unique humanness changes what is possible. The most immediate is fair distribution. Crypto projects frequently give away tokens to early users through airdrops, and the entire premise, rewarding a broad community, collapses if a handful of people can each register thousands of identities and vacuum up the supply meant for many.
A proof-of-personhood gate, requiring each claimant to prove they are a unique human, restores the fairness the airdrop was supposed to deliver, and the same logic extends to any system handing scarce resources to people: community rewards, grants, promotional credits, or the long-discussed vision of a basic income distributed to verified individuals rather than to whoever runs the most bots.
A second arena is governance and voting. Decentralized organizations and online communities increasingly make decisions by vote, and a vote is only meaningful if each person counts once. Without proof of personhood, governance defaults to systems where influence is bought, whoever holds the most tokens or controls the most accounts decides, which concentrates power and invites manipulation.
A reliable proof of unique humanness opens the door to genuine one-person-one-vote systems online, a building block for fairer collective decision-making that has been technically out of reach. A third arena is the everyday integrity of online spaces: social platforms drowning in AI-generated accounts, review systems gamed by fake humans, and communities overrun by bots all need a way to certify that a participant is a real, unique person, and proof of personhood offers exactly that certification without forcing users to surrender their identities.
The newest and fastest-growing arena is the one created by autonomous AI. As software agents begin acting on people’s behalf, the question of which human stands behind a given agent becomes urgent, both to assign accountability and to keep anonymous bots from masquerading as authorized representatives.
Proof-of-personhood tools that bind an agent to a verified human principal let an agent prove it acts for a genuine, unique, accountable person, which is becoming a prerequisite for trusting agents with real tasks and real money. Enterprises are also adopting proof-of-humanity checks to defend high-value interactions, from video calls to account access, against deepfakes and impersonation.
Across all these cases, the common thread is the same: wherever a system needs to know that a participant is a real, unique human, and increasingly wherever it needs to know which human is behind a machine, proof of personhood is the missing layer that makes the guarantee possible. That breadth of application, spanning fairness, governance, online integrity, and the entire emerging world of AI agents, is why the idea has drawn so much attention despite its unresolved controversies.
The serious objections
A guide that only described the promise of proof of personhood would be misleading, because the field, and especially its biometric flagship, has drawn intense and substantive criticism that any honest reader should weigh. The first objection is the biometric honeypot problem. Building a system that scans the irises or faces of millions of people creates, by its nature, one of the largest collections of biometric data in the world, and even with deletion and anonymization, critics argue that such a database is an irresistible target and that the consequences of biometric data being compromised are uniquely severe, because you cannot change your eyes the way you change a password. The risk of normalizing mass biometric collection, and of who ultimately controls it, sits at the heart of the unease.
The second objection is centralization. A system built on specialized hardware that the project manufactures and controls creates a chokepoint: a single company decides who can verify, where the devices go, and how the system runs, which sits awkwardly with crypto’s ideals of decentralization and raises the prospect of a private entity becoming a gatekeeper of human identity online. The third objection is regulatory and ethical: the leading project has faced pushback, suspensions, and investigations from data-protection authorities in numerous countries worried about consent, privacy, and whether scanning eyes in exchange for tokens, sometimes in lower-income regions, is exploitative.
A fourth, more technical critique questions whether a crypto token needs to be attached to identity verification at all, suggesting the financial layer may be unnecessary to the core function. And a fifth points out that large platforms or governments could build competing verification systems with less controversy, or that softer software-only methods might prove good enough, leaving the biometric approach burdened by risks its rivals avoid. None of these objections proves the technology is bad, but together they explain why proof of personhood, despite solving a real and growing problem, remains genuinely contested.
Why it matters and where it goes
Stepping back, proof of personhood is one of those rare ideas whose importance is rising in lockstep with the technology that makes it necessary, and that is the clearest way to understand its trajectory. The case for it is straightforward and getting stronger: as AI erases the line between human and machine online, almost every system that assumed it could tell the difference, fair distribution, honest voting, bot-free communities, accountable AI agents, needs a new foundation, and a reliable way to prove unique humanness is that foundation. The demand is real, it is growing, and it is not going away, which is why serious people and serious money keep flowing toward the problem even after years of difficulty and controversy.
The open question is not whether proof of personhood matters but which approach, if any, will earn enough trust to become a genuine standard. The biometric path offers the strongest uniqueness guarantees and the most scale so far, but carries the heaviest privacy, centralization, and regulatory baggage. The social-graph, credential, and zero-knowledge paths avoid some of that baggage but offer softer guarantees or struggle to scale. It is entirely possible that no single system wins, and that the future is a patchwork of methods suited to different contexts, a biometric proof for the highest-assurance needs, lighter software proofs for everyday ones.
It is also possible that the privacy concerns prove decisive and the world rejects mass biometric identity altogether, pushing the field toward less invasive designs. What seems certain is that the underlying need, proving a real, unique human in a world full of convincing fakes, is now permanent, and that how society chooses to meet it, and who it trusts to run the infrastructure, will be one of the defining questions where crypto, artificial intelligence, and identity collide. Proof of personhood is the attempt to answer it, and the answer is still being written.
Frequently Asked Questions
What is proof of personhood in simple terms?
Proof of personhood is a way to prove you are a real, unique human, counted exactly once, without revealing who you are. It has to do two things at the same time: guarantee uniqueness, so one person cannot create many identities, and preserve privacy, so proving you are human does not expose your name or identity. It matters because, as AI makes fake humans cheap and convincing, many online systems, fair giveaways, honest voting, bot-free communities, can only work if they can reliably tell one real person from a thousand fakes.
What is a sybil attack?
A sybil attack is when a single actor creates many fake identities to gain influence they should not have. On a system where one identity equals one vote or one share, someone controlling a thousand fake identities has a thousand times the honest influence. Almost every open online system that tries to be fair is vulnerable to it. Sybil attacks used to be limited by the friction of making convincing fake accounts, but AI removes that limit by generating unlimited realistic personas, which is why defending against sybil attacks now requires proving real, unique humanness.
How does the iris-scanning approach work?
The leading biometric project uses a device called the Orb to scan a person’s iris, because the iris is highly unique and very hard to fake, even by AI. According to the project, the Orb generates a unique cryptographic code from the scan, deletes the actual image after processing, and distributes only anonymized fragments to confirm the person has not enrolled before. The result is a credential proving you are a unique human without revealing your identity, and zero-knowledge techniques let you later prove “I am a verified unique human” to an app without exposing anything else about yourself.
What are the alternatives to biometric verification?
Several. Social-graph systems build humanness through webs of trust, where real people vouch for other real people, avoiding biometrics but struggling to scale. Credential-based systems combine signals like verified accounts, on-chain history, or documents into a score suggesting a unique human, offering flexibility but softer uniqueness guarantees. Zero-knowledge identity methods prove facts about you, such as being a unique credential holder, without revealing the data, and increasingly use device-based passkeys. Each approach makes a different trade-off among uniqueness, privacy, security, and scale, and the field has not settled on a single winner.
Why is proof of personhood controversial?
Mainly because the strongest approach, biometrics, raises serious concerns. Collecting iris or face data from millions creates a large biometric database that critics see as a honeypot, made worse because you cannot change your biometrics like a password. Building it on hardware one company controls creates centralization and gatekeeping worries that clash with crypto’s ideals. The leading project has faced regulatory pushback and suspensions in many countries over privacy and consent, and some argue that verifying people in lower-income regions for tokens is exploitative. Others question whether a token is needed at all, or whether less invasive methods would suffice.
How does proof of personhood relate to AI?
Closely, in two directions. First, AI created the urgency: as it makes fake humans cheap and convincing, proving real humanness becomes valuable precisely because humanity is becoming the scarce, trustworthy thing online. Second, the field is evolving from keeping bots out to governing the AI agents now acting on people’s behalf. New tools tie an AI agent to a verified human principal, so an agent can prove it represents a genuine, accountable person instead of running as an anonymous bot. So proof of personhood is becoming a layer that verifies both humans and the machines acting for them.
This article is educational information, not investment or identity-security advice. Proof-of-personhood projects, their scale, and their regulatory status change quickly, and details reflect reporting available as of June 25, 2026. Consider the privacy and security implications carefully, and verify current information from primary sources before enrolling in or relying on any identity system.
Telegram founder Pavel Durov says Russian authorities have designated him a “terrorist” after he refused government demands for mass surveillance and censorship on the messaging platform. Durov posted his response publicly in a Telegram message published a day after Russia announced new criminal charges against him.
In that post, Durov claimed Russia also blocked him from “publishing information on the Internet,” adding that authorities appear to have “got confused about who can ban whom from the Internet.”
Key takeaways
- Durov’s latest statement follows Russia’s announcement of charges, after Russia’s security service accused him of facilitating terrorist activity.
- The Russian allegations center on Telegram’s alleged failure to remove channels tied to terrorist groups and Ukrainian intelligence services.
- Russia’s case builds on a criminal probe reportedly launched in February over alleged non-compliance with Russian rules on extremist and terrorism-related content.
- Durov is also facing separate legal scrutiny in France, and regulators in Australia have reportedly initiated court proceedings over terrorism-related content.
Russia escalates case with “terrorist” label
Russia’s Federal Security Service (FSB) accused Durov of facilitating terrorist activity, according to earlier coverage from Cointelegraph. The claim is that Telegram did not remove certain channels used by terrorist organizations as well as channels linked to Ukrainian intelligence services.
Durov’s rebuttal came through his own Telegram account the day after Russia made the allegations public. In addition to responding to the charges, he framed the dispute as a conflict over Telegram’s stance toward government demands, particularly around surveillance and content restrictions.
The message also suggests a broader disagreement about control of online speech and information access—Durov saying Russian authorities barred him from publishing information on the internet, while disputing the legitimacy of who can impose such restrictions.
February investigation tied to alleged content non-removal
Russia’s current push does not appear out of nowhere. Cointelegraph previously reported that a criminal investigation in February followed regulatory accusations that Telegram left nearly 155,000 channels, chats, and bots accessible despite Telegram’s alleged violations of Russian laws relating to extremist material, terrorism, drug trafficking, and other illicit categories.
That earlier context matters because it indicates that the case is tied to a longer-running compliance argument—how Telegram moderates content and how it responds to removal requests. For investors and platform users watching the regulatory risk landscape, these developments reinforce that messaging platforms can face escalating consequences when regulators argue they are not acting quickly or thoroughly enough to meet local legal standards.
At the same time, much remains uncertain: public allegations do not automatically translate into immediate changes on the ground, and the specifics of what was or wasn’t removed—and under what procedural thresholds—are not detailed in the statements referenced here.
Legal pressure extends beyond Russia
Durov’s legal situation has been international. Cointelegraph reported that he was arrested in France in August 2024 and has remained the subject of a judicial investigation over allegations that Telegram facilitated criminal activity, including by allegedly failing to moderate illegal content adequately and respond to law enforcement requests.
According to earlier coverage, Durov has denied wrongdoing, arguing that French authorities did not follow due process when seeking information from Telegram. His arrest was followed by a TON Community-backed campaign that collected more than 9 million signatures on an open letter urging French authorities to release him.
There have also been reported changes in travel conditions: French authorities initially allowed Durov to return temporarily to Dubai in March 2025 before lifting travel restrictions entirely later that year, as noted by Cointelegraph. The implication is that while legal pressure can be intense, authorities may adjust constraints over time depending on the status of the proceedings.
Meanwhile, Cointelegraph also reported that Australia has moved toward court proceedings, with regulators alleging Telegram failed to remove terrorism-related content. The platform’s exposure in multiple jurisdictions highlights a recurring regulatory theme worldwide: governments increasingly expect major communication platforms to take meaningful action against content they deem illegal, even when the platforms frame enforcement as complex, global, and constrained by legal standards.
Privacy, surveillance, and the EU debate
Beyond the courtroom, Durov has publicly positioned himself around digital privacy and free speech. In April, he warned that a proposed European Union age-verification app could open the door to broader online surveillance, according to Cointelegraph. The argument—tying age verification to a wider data-collection trajectory—signals how Durov views regulatory measures as potentially expanding monitoring rather than simply improving safety.
In the same period, Cointelegraph reported that Durov blamed alleged tax data leaks for an increase in crypto-related kidnappings in France. He said Telegram would leave the country rather than grant authorities access to users’ private messages, tying the debate back to the line between lawful access and what he frames as unacceptable intrusion into private communications.
Taken together with the Russia and France allegations, these public remarks suggest that Durov is consistent in his framing: that enforcement actions aimed at removing content or complying with requests can also become mechanisms for surveillance or compelled access to private data.
What to watch next is how each jurisdiction’s process unfolds—whether courts compel specific operational changes, whether regulators escalate further, and how Telegram responds in practice. For users, the key question is whether enforcement will lead to tangible moderation shifts; for the broader industry, it will be whether these cases set precedents for how governments can pressure encryption-adjacent communication platforms to cooperate with investigations.
Google is reportedly backing a proposed $15 billion financing package for an Anthropic-linked data center in Texas, extending an AI infrastructure boom that has also lifted Bitcoin miners expanding into high-performance computing.
Summary
- Nexus is in advanced talks to raise $15 billion for its Hubbard, Texas, data-center campus.
- Google would guarantee parts of Anthropic’s leases and receive an estimated 20% project stake.
- The campus includes an on-site natural-gas plant capable of generating 1.6 gigawatts.
- AI-focused mining stocks rallied Thursday, with IREN, Hut 8 and CleanSpark gaining more than 20%.
Google and Anthropic financing terms
A group of banks led by Morgan Stanley is discussing a financing package for Nexus Data Centers’ planned campus in Hubbard, Texas, according to a Wall Street Journal report cited by Reuters.
The proposed package includes a $14 billion bridge loan and a revolving credit facility. The report did not disclose the size of the revolving facility or identify the other lenders involved in the talks.
Google has reportedly agreed to guarantee billions of dollars in lease and power-payment obligations if Anthropic defaults. Its backing would apply to four data-center leases signed by the AI developer and related agreements to purchase electricity from an on-site power plant.
The guarantees remain limited to the minimum amount lenders require to complete the financing, according to the report. In exchange, Google is expected to receive an equity stake of around 20% in the combined data-center and power project.
Neither Nexus nor the banks have announced a completed transaction.
Texas campus includes a 1.6 GW power plant
Nexus plans to pair the data-center campus with a natural-gas-fired power plant capable of producing 1.6 GW of electricity. The on-site facility would allow the project to secure a large power supply without relying entirely on additional capacity from the Texas grid.
Access to electricity has become one of the main constraints on new US data centers. AI facilities require large and continuous power supplies, while new transmission lines, generation assets and grid connections can take years to approve and construct.
The Nexus structure resembles other AI projects using financially stronger technology companies to support the obligations of private AI developers. Google’s guarantee would improve the project’s credit profile and could help Nexus obtain financing on better terms than Anthropic could secure independently.
Anthropic plans to use tensor processing units co-designed by Google and Broadcom at the site. The chips would be financed separately under a vendor agreement between Anthropic and Broadcom, according to the Wall Street Journal.
Google’s participation would therefore cover several parts of the project: chips, credit guarantees and an expected equity position. The arrangement also gives Google a larger role in Anthropic’s infrastructure even as the AI developer buys processors from competing suppliers.
Bitcoin miners secure similar AI agreements
Nexus is competing for capital and AI tenants during a month marked by several multibillion-dollar agreements involving Bitcoin mining companies.
TeraWulf signed a 20-year lease with Anthropic on July 6 for its Justified Data campus in Hawesville, Kentucky. The contract covers about 401 MW of critical IT capacity and is expected to generate approximately $19 billion in revenue over its initial term.
CleanSpark followed with a $6.6 billion lease for its Sandersville campus in Georgia. The 20-year agreement covers 175 MW and could rise to $11.6 billion if the unidentified technology customer exercises its extension options.
Hut 8 signed a second 352 MW lease at its Beacon Point campus in Texas on July 20. The $9.8 billion contract brought the property’s total base-term lease value to $19.6 billion. Renewal options could increase the figure to $50.2 billion.
IREN announced $2.8 billion in multi-year AI cloud contracts on the same day. It raised its year-end annualized AI cloud revenue target above $4 billion, with about 85% of the target under contract.
Core Scientific then signed an agreement giving AMD access to up to 2.5 GW of data-center capacity. The initial phase will provide 500 MW of AI-ready infrastructure in 2027, although the companies did not disclose the contract’s value.
These companies accumulated power agreements and industrial sites for Bitcoin mining before AI demand raised the potential value of those assets. Their existing grid access can shorten development timelines, although AI facilities require more advanced cooling, networking and backup systems than mining operations.
AI-focused mining stocks rally
Shares of several miners with AI infrastructure exposure rose sharply on July 30, although the broader rally cannot be attributed solely to the Nexus report.
IREN closed approximately 30.7% higher at $38.26, while Hut 8 advanced 22.7% to $108.27. CleanSpark rose 21.1%, Core Scientific added 20.4% and TeraWulf gained 18.1%.
The moves extended an earlier market response to July’s AI contracts. Hut 8 had risen as much as 17% when it announced its second Beacon Point lease, while IREN climbed as much as 19% following its cloud-contract update.
Alphabet shares moved in the opposite direction Thursday, falling about 0.9% to $333.66. The decline came as investors continued assessing the cost of Google’s AI infrastructure expansion rather than signaling a clear response to the uncompleted Nexus financing.
What comes next for the Nexus project
The immediate catalyst will be whether Morgan Stanley and the other banks finalize the $15 billion package and disclose its maturity, interest costs and guarantee structure.
Confirmation of Google’s expected 20% stake would also clarify how much direct exposure the company is taking to Anthropic’s infrastructure obligations. Until financing closes, the loan, guarantees and equity arrangement remain proposed terms reported by people familiar with the discussions.
For US-listed miners, completion would provide another valuation benchmark for large power-backed AI campuses. It would also add a conventional data-center developer to the competition for the same tenants, lenders, chips and power supplies now driving miners’ expansion beyond Bitcoin.
The Bitcoin miner is using part of its treasury and a BTC-backed credit facility to fund an AI campus tied to a potential multi-billion-dollar infrastructure contract.
Coinbase shares fell in extended trading after the crypto exchange reported a second-quarter loss and lower revenue, overshadowing record market share and growth across stablecoins and derivatives.
Summary
- Coinbase reported a loss of $1.36 per share, reversing earnings of $5.14 a year earlier.
- Quarterly revenue fell to about $1.2 billion from $1.5 billion in the prior-year period.
- Coinbase captured a record 10.3% of global crypto trading volume during the quarter.
- COIN faced immediate support near $152, with the next downside level around $140.
Coinbase revenue falls as quarterly loss returns
Coinbase generated approximately $1.2 billion in second-quarter revenue, down 20% from $1.5 billion during the same period last year. The company posted a loss of $1.36 per share, compared with earnings of $5.14 per share a year earlier.
Shares initially closed regular trading at $163.58, up 2.18% for the session. However, the earnings report reversed that gain, sending the stock lower in after-hours trading.
The chart showed an extended-market price near $152, implying a decline of about 7% from the regular close. Earlier after-hours readings placed the drop closer to 5%, suggesting the stock remained volatile as investors assessed the report.
Lower revenue and the return to a quarterly loss weighed on sentiment despite several operating improvements. Coinbase also completed its 14th consecutive quarter of positive adjusted EBITDA and reduced its forecast for full-year adjusted expenses.
Coinbase reaches record 10.3% trading share
Weak financial results contrasted with Coinbase’s expanding presence in the global crypto market. Its share of worldwide crypto trading volume rose to a record 10.3% from 9.1% in the first quarter.
That marked the third consecutive quarter in which the US exchange increased its market share. The gain came even as overall crypto market volume declined by double digits.
Derivatives activity remained close to the record level reached during the previous quarter. Revenue and contracts tied to event markets increased 106% quarter over quarter, pushing the business above a $100 million annualized revenue rate.
Coinbase also continued reducing its reliance on Bitcoin spot trading fees. Revenue excluding Bitcoin spot activity accounted for 88% of net revenue, reflecting a broader shift toward subscriptions, stablecoins, payments and financial infrastructure.
Subscription and services revenue reached $555 million, compared with just $6 million in the second quarter of 2020. The segment generated 48% of net revenue, up from 29% in the fourth quarter of 2024.
USDC and Base activity support diversification
Stablecoins provided another area of growth. Average USDC balances held across Coinbase products reached a record $20 billion, representing more than 30% of the stablecoin’s circulating supply at quarter-end.
Coinbase reported that USDC and its partner stablecoins accounted for 79% of the more than $37 trillion in stablecoin transaction volume recorded during the year. Stablecoin volume on Base, the exchange’s Layer 2 network, increased sevenfold from a year earlier.
The figures show how Coinbase is expanding beyond transaction fees tied directly to crypto price cycles. This diversification could provide more recurring revenue, although the quarterly loss shows that growth in newer business lines has not fully offset weaker overall conditions.
For US investors, the results offer mixed signals. Coinbase remains a major publicly traded proxy for the domestic crypto industry, but its earnings continue to reflect trading activity, digital asset prices and regulatory conditions.
The company also reported gains from using artificial intelligence in its engineering work. Code changes processed per engineer increased 2.2 times year over year, while integration test coverage across core services rose 2.5 times over six months.
COIN price tests $152 support after earnings
COIN’s after-hours decline pushed the stock below several closely watched technical levels. The regular-session close of $163.58 sat just above the 20-day simple moving average at $162.97 but below the 50-day average at $165.29.
A move toward $152 would place the stock at immediate chart support. Failure to hold that level could expose the late-June low near $140.
On the upside, COIN must reclaim the 50-day average before testing the 100-day SMA near $178.43. The 200-day average at $213.49 remains a larger long-term resistance level.
The average directional index stood at 10.11 before the earnings reaction. A reading this low indicates weak trend strength, matching the stock’s recent sideways movement around the $160 region. The post-earnings gap may provide a stronger directional signal if trading volume remains elevated during the next regular session.
Grok AI predicts a massive breakout for Ethereum, and this price prediction sets the bar unusually high. The bull case runs to $6,000 to $8,000, with a stretch scenario reaching $10,000 to $12,000 by the end of 2026, up from roughly $1,890 today.
Accelerating spot ETH ETF inflows anchor the case. These are described as already flipping positive, with BlackRock’s ETHA taking the leadership position and cumulative net inflows exceeding $11 billion.
Expanding staked ETH ETF products are named as a second driver, unlocking yield for institutions in a way that turns simple price exposure into something closer to an income-bearing asset. That structural shift did not exist in any prior Ethereum cycle.
Post Pectra and Fusaka scaling adds real technical weight to the case. PeerDAS is expected to deliver a multi-fold increase in blob capacity, making Layer 2 networks meaningfully cheaper to operate, while the Glamsterdam upgrade in the second half of 2026 is set to boost Layer 1 throughput through ePBS and parallel execution.
Ethereum’s dominant share of stablecoins and tokenized real-world assets, estimated at tens of billions and still growing, forms the usage backbone underpinning the technical upgrades. Rising staking lockups are tightening liquid supply, a potential ETH-to-BTC ratio recovery is floated as a further tailwind, and broader institutional and RWA adoption is framed as solidifying Ethereum’s position as the premier settlement layer.
The bear case is treated as a real possibility rather than a footnote. Stalled ETF flows, Layer 2 competition or fee compression limiting how much value accrues to the base layer, upgrade delays, regulatory setbacks, or macro tightening could all keep Ethereum trading between $2,200 and $4,000 instead.
Ethereum Price Prediction: ETH Just Fell Back Below The Level It Spent Weeks Trying To Hold
Price closed at $1,877.71, down 2.20%, in a session ranging between $1,871.84 and $1,932.72. That red day breaks a run of gains that had briefly pushed Ethereum back above the $1,900 mark.
Zoom out, and the broader trend since September 2025 has been a long, uneven decline. Ethereum peaked near $4,950 that month, then broke down hard through January, gapping from above $3,000 to under $2,200 in a matter of weeks.
Since that crash, price has made two separate recovery attempts, one in April that stalled near $2,450, and another in June that also topped out near the same level before rolling back into a sharp flush to $1,540. The climb since that June low pushed price above $1,900 for the first time since the flush, but today’s drop pulls it right back under that mark.
Support sits at $1,850, then the June low near $1,540 if this pullback deepens. Resistance stacks at $1,930, then $2,200, then the heavier ceiling near $2,450 that has already rejected two separate rally attempts this year.
Momentum here has cooled sharply after weeks of steady gains, with today’s decline erasing some of the recent progress. For Grok’s bull case to gain any real traction, Ethereum first needs to reclaim $1,930 and then clear $2,450, the exact level that has stopped this chart twice before, rather than losing ground the way it did today.
Here is what Grok AI Predicts For LiquidChain’s Near Future
Every cycle has one moment when standing still costs more than moving. That moment is now.
Bitcoin, Ethereum, and XRP are all trapped under the same resistance they have tested for weeks. The macro catalyst is always one data point away. The institutional wave always lands next quarter. Large cap traders waiting for a breakout are lined up behind a decision that belongs entirely to someone else.
Grok AI has flagged what experienced cycle traders already understand instinctively. Capital that vanishes as statistical noise at Bitcoin’s scale can completely reshape the price of a small, undiscovered project. The asymmetry here is not complex. It exists in the gap between what something is genuinely worth and what the market currently believes it is worth. That gap collapses the instant it gets noticed. Right now, it remains fully open.
Cross-chain fragmentation has quietly extracted value from every DeFi participant since the first bridge went live. Bitcoin, Ethereum, and Solana were built independently, with no shared infrastructure and zero intention of communicating with each other. Every transaction crossing those boundaries pays for that decision in fees, failed execution, and slippage extracted before settlement even completes. Bridges were never the fix. They became a business built on top of the unsolved problem.
LiquidChain eliminates that business model entirely. All 3 networks merge into a single execution layer. One deployment reaches everything at once. Zero cross-chain tax on any interaction, anywhere.
Grok AI has flagged it as a coin worth watching. The presale sits at $0.01454 with just over $860,000 raised.
Execution is unproven. Adoption remains an open question. Established assets offer a smoother climb toward a ceiling the whole market can already see. LiquidChain is the entry point that disappears the moment the market catches on.
The post Elon Musk Grok AI Predicts Ethereum Will Hit This Price by End of 2026 appeared first on Cryptonews.
Elon Musk says 90% of Earth has a financial reason to move to Spain. He made the claim on Thursday, as thousands of people crossed from Morocco into the Spanish city of Ceuta.
It is an argument he has made many times before. Until now, he aimed it at the United States.
What Happened in Ceuta
Ceuta is a small Spanish city on the coast of North Africa. About 85,000 people live there. It sits right on the border with Morocco.
Thousands crossed over on Thursday. Most swam around the Tarajal seawall. At least nine people died.
Follow us on X to get the latest news as it happens
Rachid Sbihi runs the union for Ceuta’s border police. He described “absolute chaos” and said the border had “totally collapsed,” according to the Associated Press.
Local leader Juan Jesús Vivas asked Madrid to declare a national emergency and send the army. The Interior Ministry said no. It sent troops and extra police anyway. Spain and Morocco then agreed to speed up returns.
Prime Minister Pedro Sánchez said Spain was mobilizing all necessary resources. In January, his government gave legal status to about half a million undocumented workers.
“The Government of Spain is fully committed to providing an immediate response to the situation in Ceuta…working with Moroccan and international authorities, and preparing the necessary measures to restore normalcy as soon as possible… This is the moment to build solutions, with responsibility and cooperation,” wrote Sanchez.
Musk attacked that plan at the time. Sánchez told him “Mars can wait.” The clash fits a pattern in Musk’s European political commentary.
Where the ‘90% of Earth’ Line Comes From
The idea is simple. Musk says a country becomes a magnet when its benefits beat what most of the world lives on. That magnet then grows big enough to break the budget.
He made the same point in April about America. Free taxpayer money can beat the living standards of 90% of Earth, he wrote. That gives 90% of Earth a reason to move there.
A 2024 version used smaller numbers. America holds 4% of the world’s people. A shift of just 1% would crush essential services, he said. For Spain, he put the number at 7 billion.
Spain’s entire budget will be destroyed by illegal migrants. It’s basic math: if Spain offers free stuff to migrants that is above 90% of the living standard of Earth, they create a forcing function for 90% of Earth to move to Spain, which is around 7 billion people!” Musk explained.
Researchers see it differently. They find that jobs, distance, language, and family already living abroad matter more than benefits. That gap is why the line spreads fast, and why it gets attacked just as fast.
The post Elon Musk Says 90% of Earth Would Move to Spain: Is He Right? appeared first on BeInCrypto.
China Shipping containers are seen at the port of Oakland, as trade tensions continued over U.S. tariffs with China, in Oakland, California, U.S., May 12, 2025.
Carlos Barria | Reuters
BEIJING — One of the Chinese economy’s few growth drivers lost steam in July, according to the latest survey of businesses by China Beige Book.
“U.S.-bound shipments fell outright for the first time in several months,” the U.S.-based research firm said Friday. The findings are based on a survey of 1,436 Chinese businesses between July 20 and 28.
The last time China’s exports to the U.S. fell was in March, when they plunged more than 26% from a year ago, in line with the general trend of double-digit declines since trade tension escalated in April 2025, according to official data accessed via Wind Information.
Shipments to the U.S. rose by 14% in June, helping overall exports surge by 27% — the most in nearly five years. The growth came as businesses frontloaded shipments ahead of expected higher U.S. tariffs later in the summer. The rapid development of data centers to power AI has also driven demand for China-made parts.
The latest China Beige Book study found factory activity decelerated in July, with manufacturing seeing the worst performance in employment as all sectors surveyed saw job growth worsen from a year ago.
Retail sales also fell in July from the prior month and the year-ago period, the report said, noting travel and restaurants “saw a sharp on-year downturn.”
China’s top policymakers on Thursday emphasized the need to expand domestic demand and international trade cooperation, according to a state media readout. The statement underscored Beijing’s priority of achieving technological “breakthroughs.”
Trade data for July is due out Aug. 7, while retail sales and investment figures are expected on Aug. 17.
Bitcoin (BTC) and Ethereum (ETH) look set to finish July ahead of most major asset classes, with the former adding over 7% and the latter gaining almost 20% in the last 30 days.
The performance adds to a month of recovery for the two largest cryptocurrencies after a difficult first half of 2026, although historical data suggests August has been a much tougher month for BTC.
Bitcoin and Ethereum Lead July Returns
Data from CoinGlass at the time of writing showed that Ethereum had gained 19.5% during the month while Bitcoin had risen 7.37%. Meanwhile, a comparison by analyst Ash Crypto across major markets showed chip stocks fell 22% in the same period, with the Nasdaq 100 and the Russell 2000 slipping by 9% and 3%, respectively.
The S&P 500 also fell, but its decline was much smaller than that of its counterparts, at about 1%. Silver dropped by 2.64%, but gold was little changed, adding just 0.38% to its value over 30 days.
What makes the gains by the cryptocurrencies noteworthy is that before July, they had endured a rough 2026. CoinGlass data shows BTC fell more than 10% in January, as it continued a red run that had started in October 2025. That sequence continued into February, when the OG crypto lost almost 15%, before reprieves in March and April. May registered a -3.41% return and June recorded the worst drop of the year so far when the asset lost over 20% of its worth.
Ethereum’s first two quarterly performances were just as bad, with Q1 returns at -21.26% and those for Q2 at -25.28%.
Recall that BTC started July trading near $58,000 but gradually climbed the chart, hitting a monthly high near $67,000 last week before price action started cooling somewhat. It was pretty much the same with ETH, as CoinGecko data shows it kicking off the month near $1,500 and eventually ending up very close to $2,000 as July drew to a close.
At the time of writing, the world’s second-largest cryptocurrency was changing hands just above $1,900, having shed about 1% in the last seven days. However, despite the good monthly run, it’s still more than 50% lower than where it was a year ago and about 61% away from its August 2025 all-time high. Bitcoin, on its part, has settled near $64,000, which is almost half of its own ATH, after shrugging off the slight volatility that came with yesterday’s decision by the Fed to keep interest rates unchanged.
August Record Keeps Traders Cautious
While July brought relief for crypto investors, CoinGlass data points to a recurring seasonal pattern. Every August since 2022 has ended with Bitcoin posting a monthly loss, including declines of 6.49% in 2025, 8.6% in 2024, 11.29% in 2023 and 13.88% in 2022.
That backdrop has kept analysts divided on what comes next, with Ali Martinez forecasting that Bitcoin’s bear market could last until October, while traders Pepesso and Crypto Lens expect another move lower before a broader recovery begins in 2027.
The post Bitcoin, Ethereum Outperform Markets in July as Chip Stocks Plunge 22% appeared first on CryptoPotato.
Then the regulator stepped in. On June 22, the CFTC opened a request for comment on extending perpetual contracts to physically-delivered crude oil; 67 questions on reference prices, liquidity, position limits, and customer protection. And when the CME tried to self-certify its 24/7 oil contract in July, the CFTC stayed it, blocking the fast track and forcing a full review first.
This story – the market sprinting toward round-the-clock leveraged access to everything, and the rule-writers trying to decide how fast is safe – will keep repeating as the U.S. works to onshore derivatives flow and exchanges push for a level playing field with their offshore counterparts. When the biggest U.S. derivatives exchange is shrinking oil contracts for 24/7 retail access, and the U.S. derivatives regulator is drafting the rules for perpetual oil, you can stop debating whether the model won. It won.
So the interesting conversation isn’t “are perps spreading.” It’s three sharper questions: which asset classes get perpetuals next, where the leverage actually concentrates, and what breaks along the way.
On what’s next, follow the friction. Perps are most valuable precisely where the traditional market is most annoying, where it closes at night, gates you by geography, demands accreditation, or settles at a crawl. That’s why commodities, pre-IPO equities, and hard-to-reach foreign stocks got perpetuals first: enormous latent demand, hopelessly constrained access. The same logic points straight at private credit, carbon, freight, and the long tail of real-world assets coming onchain. Anything with a reference price and a frustrated audience is a candidate. The underlying almost doesn’t matter; the demand to trade it freely does.
MoonPay has launched PayBox, a noncustodial payment vault that lets users prepare and execute crypto transactions or online purchases through ChatGPT and Claude.
Summary
- PayBox connects with ChatGPT and Claude through custom connectors and natural-language commands.
- Users can buy, swap, bridge or deposit crypto and complete travel, dining and retail purchases.
- Passkey approvals and user-defined spending limits control what connected AI assistants can execute.
- MoonPay supports Solana and several EVM networks, including Ethereum, Base, Arbitrum and Polygon.
MoonPay PayBox turns AI conversations into transactions
PayBox allows users to connect a payment vault to ChatGPT or Anthropic’s Claude and describe a transaction in plain language. The AI assistant can then research available options, prepare the transaction, and execute it under permissions set by the user.
Supported crypto actions include buying digital assets with fiat currency, swapping tokens, moving assets between blockchains, and depositing funds into decentralized finance protocols. PayBox can also complete commercial transactions such as booking flights, reserving restaurant tables and purchasing goods from online retailers.
The launch extends AI assistants beyond research and transaction preparation by giving them limited authority to act on a user’s behalf. However, that authority depends on the security settings attached to the PayBox account.
MoonPay said the product is live through the PayBox website. Users must connect it to a supported AI platform through a custom connector before issuing payment instructions.
Passkeys and spending rules limit AI access
PayBox offers two authorization models. Under the “Always Ask” setting, every transaction requires the user to approve the action with a passkey. The approval applies to only one transaction and expires after use, preventing the AI from applying it to a different payment.
The “Autonomous” model allows the connected assistant to operate within spending limits and other rules chosen by the user. This option removes the need to approve each eligible transaction separately, but it does not give the assistant unrestricted control over the vault.
Any change to the permission model or transaction rules requires another passkey authorization. Users can therefore define how much the assistant can spend and the types of actions it can perform before enabling autonomous execution.
PayBox supports both crypto wallets and payment cards. For wallet transactions, private keys are divided using multiparty computation and stored across secure hardware environments. MoonPay said neither it nor the connected AI assistant can independently reconstruct the complete key or authorize an asset transfer.
Card payments use Visa’s agentic commerce protocol, allowing the assistant to complete approved purchases without receiving or storing the underlying card number.
PayBox supports Solana and major EVM networks
MoonPay has added support for Solana and several Ethereum Virtual Machine-compatible networks. The initial list includes Ethereum, Hyperliquid, Tempo, Base, Robinhood Chain, Arbitrum and Polygon.
The vault also integrates with x402, an open payment standard designed for services that accept transactions initiated by AI agents. MoonPay said its first x402 integrations cover travel bookings, restaurant reservations and purchases from major online retailers.
The network coverage lets users carry out several steps through one conversation. An assistant could, for example, help a user acquire an asset, bridge it to another blockchain and deposit it into a supported DeFi protocol, provided every step falls within the account’s permissions.
The product relies on security technology developed by Sodot, a key-management company MoonPay acquired earlier in 2026. MoonPay said Sodot’s infrastructure secures more than $50 billion in assets across over 10 million wallets.
MoonPay expands from institutional trading into AI payments
PayBox follows MoonPay’s recent expansion into tokenized financial products. As crypto.news reported in June, Franklin Templeton added its BENJI tokenized money market fund to MoonPay Trade.
That integration allows institutional users to exchange USDC, USDT and other stablecoins for BENJI through MoonPay’s on-chain trading platform. It also gives BENJI holders access to stablecoin liquidity and supports uses such as treasury management, portfolio rebalancing, collateral and liquidity provision.
PayBox targets a different part of the market by connecting consumer and crypto payments directly with conversational AI. MoonPay describes the vault as noncustodial because users retain control of their assets and neither MoonPay nor the AI provider can move funds alone.
For US users, PayBox’s use of Visa’s agentic commerce framework could make AI-assisted card payments more practical, while passkey controls may help address concerns over unauthorized purchases. Access to individual crypto assets, DeFi protocols, and payment services may still depend on location and the rules applied by each provider.
Samsung Says It’s Launching Galaxy S26 FE and Tab S12 Later This Year
15 Minutes Story That will Change Your FINANCIAL LIFE !!
Boy George pulls out of Jesus Christ Superstar just days before West End stint amid controversy following the release of pro-Israel song We Will Dance Again
-
Fashion6 days agoWeekend Open Thread: Brooks Brothers
-
Sports4 days agoCommonwealth Games boxing: Jadumani Singh seals dominant 5-0 win over Pakistan’s Sumama Rehman to enter quarter-finals | Commonwealth Games News
-
Business1 day agoWhy Trees Belong on the Risk Register
-
Tech4 days agoIntel is reversing course and bringing hyper-threading back to its server chips
-
Crypto World5 days agoRipple bought a bank in pieces. The $4 billion audit
-
Politics4 days agoLuke Littler dismantles Gerwyn Price to retain title in Blackpool
-
Entertainment6 days agoA New Post-Apocalyptic Gundam Anime Series Blasts Into SDCC
-
Politics3 days agoThe Part of the Electric Transition Nobody Wants to Discuss
-
News Videos4 days agoBITCOIN JUST ENTERED THIS CRITICAL ZONE…
-
Fashion7 days ago16 Dresses for the High Summer Event
-
Sports7 days ago2026 3M Open leaderboard: Scottie Scheffler finds putter in Round 1, sits three back
-
Business2 days agoMajor shareholder moves on Canyon
-
Crypto World5 days agoXRP Ledger adds $2.6B as RWA inflows rank second
-
Politics4 days agoSpain sweeps the board at 2026 World Cup with individual awards
-
News Videos12 hours agoBitcoin Enters the 3rd Stage of the Bear Market
-
Entertainment2 days ago‘Stargate’ Creator’s New Sci-Fi Series Returns for Season 3 Tomorrow
-
Tech6 days agoAnthropic launches Claude Opus 5, a cheaper AI model for coding, agents and enterprise workflows
-
Crypto World2 days agoKraken Enables Retail Access to Jersey Mike’s IPO via Tokenized Shares
-
Entertainment5 days agoSara Gilson Killed By Husband After Viral “Pedophile” TikTok Video
-
News Videos2 days agoClaude: Build Financial Dashboards in Minutes (2026)
You must be logged in to post a comment Login