Menu
MCAP $2.90T ▼-2.16%
BTC $84,757 ▲+0.26%
ETH $2,688 ▲+0.93%
BNB $786.41 ▲+2.70%
XRP $1.490 ▲+0.81%
SOL $119.78 ▲+1.39%
DOGE $0.0929 ▲+0.92%
ADA $0.245 ▲+1.66%
TRX $0.3353 ▲+0.30%
LINK $14.13 ▲+3.27%
AVAX $11.10 ▲+4.05%
HYPE $89.55 ▲+2.93%
DOT $1.180 ▲+3.51%

AI Infrastructure

Every Crypto Coin Show story tagged AI Infrastructure, newest first, alongside the guides and interviews that give it context.

Guide: What is staking? How proof-of-stake yieldGuide: What is an Ethereum Layer 2?Newsletter
From the interview archive
Charles Hoskinson Has a Theory for AI Outage Affecting ChatGPT, Claude and Grok
AI News

Charles Hoskinson Has a Theory for AI Outage Affecting ChatGPT, Claude and Grok

, , , , , ,

A coordinated outage affecting ChatGPT, Claude, and Grok on September 3rd raises critical questions about the centralization risks embedded in enterprise AI infrastructure. For institutional investors in blockchain and decentralized systems, this incident underscores the strategic vulnerability of relying on consolidated compute resources and single points of failure in mission-critical AI services.

Three of the world’s most prominent artificial intelligence platforms experienced simultaneous service disruptions on Thursday morning, September 3rd, triggering outage reports that spiked from thousands to tens of thousands within a 90-minute window. Anthropic’s Claude reported capacity constraint errors beginning around 9:00 AM Eastern Time, followed by xAI’s Grok experiencing widespread unavailability, and finally OpenAI’s ChatGPT documenting elevated error rates across its services approximately 90 minutes later. Downdetector, the internet outage monitoring platform, recorded initial Claude and Grok complaints near 9:29 AM EDT, with ChatGPT reports jumping from approximately 5,000 to over 22,000 within ten minutes. While no cloud provider formally declared infrastructure failure and the companies offered limited public statements about root causes, the incident has sparked debate within blockchain and decentralized technology circles about the systemic risks posed by concentrated AI compute resources. Notably, Charles Hoskinson, founder of Cardano and co-founder of Ethereum, publicly suggested the outages bore hallmarks of a coordinated nation-state attack targeting the shared infrastructure dependencies that support these competing AI platforms.

The Cascading Failures Across Major AI Platforms

The sequence of events on Thursday morning painted a concerning picture of interdependent infrastructure vulnerabilities. Grok and Claude users began encountering service errors first, with Claude displaying a stark message: “Due to unexpected capacity constraints, Claude is unable to respond to your message. Try again soon.” This capacity constraint language distinguished Anthropic’s outage from technical failure narratives, suggesting the platform hit resource limits rather than experiencing infrastructure collapse. OpenAI subsequently acknowledged the incident, logging elevated errors across ChatGPT and Codex, its coding agent designed for software development workflows. Meanwhile, xAI initially declared no incident despite thousands of user reports indicating Grok’s model unavailability, a disconnect that raised questions about monitoring and incident communication protocols across the industry.

The geographic and temporal clustering of these failures defied conventional explanations rooted in isolated technical malfunctions. Nvidia hardware failures, even at scale, typically manifest unevenly across different physical locations and redundancy zones. The fact that three independent companies operating competing services experienced synchronized degradation within 90 minutes suggested either a shared upstream dependency, a coordinated infrastructure attack, or a cascading failure originating from common supplier systems. Reports indicated problems concentrated among U.S.-based users, and secondary services like Anthropic’s coding tools and the Cursor IDE similarly experienced disruptions. The damage proved uneven in scale, with Claude and Grok reaching approximately 1,500 complaints each on Downdetector while ChatGPT peaked above 35,000 reports, suggesting different service architectures and user base distributions but a common triggering event.

Google’s Gemini platform, operated by Google Cloud infrastructure and using proprietary TPUs rather than Nvidia processors, reported no significant incident despite hundreds of user-filed complaints. This divergence became analytically significant—Gemini experienced isolated reports while competitors faced widespread outages, suggesting Hoskinson’s observation about shared Nvidia dependency held technical merit. The exception that proved the rule pointed toward a specific technological common denominator affecting ChatGPT, Claude, and Grok but leaving Google’s differently-architected service relatively unscathed.

Infrastructure Centralization and the Nation-State Hypothesis

Charles Hoskinson’s public commentary on the incident explicitly named nation-state involvement as the most plausible explanation for synchronized outages across three competing AI platforms. “It looks like a national state brought down Claude, ChatGPT, and Grok,” Hoskinson stated, emphasizing that deliberate government action—rather than commercial sabotage, criminal hacking, or random technical failure—represented the most consistent interpretation of the available evidence. His analysis centered on the shared reliance these platforms place on Nvidia graphics processing units as core computational infrastructure. Unlike Google’s vertical integration around proprietary TPU development and deployment, OpenAI, Anthropic, and xAI all depend on acquiring, configuring, and maintaining Nvidia-based compute clusters, creating a unified technological chokepoint susceptible to coordinated disruption.

Investigation into physical infrastructure dependencies revealed a more complex picture of shared compute facilities. Reports indicated that SpaceX, which merged with xAI in February 2026, had rented Anthropic full compute capacity at Colossus 1, the Memphis-based supercomputing cluster built for Grok operations. This arrangement meant two of the three affected platforms—Grok and Claude—operated on interconnected or shared infrastructure within a single geographic location. A power issue, network failure, or deliberate disruption affecting the Memphis Colossus facility could theoretically cascade across both services simultaneously. ChatGPT’s independent infrastructure experienced separate but synchronized disruption, requiring explanation beyond shared physical plant failure. The pattern suggested either multiple coordinated strikes against distinct facilities or upstream attacks on shared suppliers serving all three platforms.

Hoskinson’s emphasis on Nvidia dependency as a potential attack vector carried geopolitical implications increasingly relevant to institutional investors and policymakers. The concentration of AI compute capability in specific hardware suppliers and the geographic clustering of data centers create systemic vulnerabilities analogous to critical infrastructure risks governments have historically targeted during conflicts or coercive campaigns. A coordinated disruption of Nvidia-based AI services could theoretically damage U.S. economic interests, degrade competitive advantage in AI development, or demonstrate technological vulnerabilities to domestic and international stakeholders. The incident, whether ultimately attributable to nation-state action or technical failure, exposed the fragility of centralized AI infrastructure governance.

Decentralization Implications and Institutional Investor Considerations

The Thursday outages immediately catalyzed discussion within blockchain communities about decentralized alternatives to concentrated AI infrastructure. Hoskinson himself referenced Midnight, Cardano’s privacy-focused sidechain, as representing a architectural approach to distributed AI compute that could mitigate single points of failure. The implicit argument positioned decentralized AI systems, built on blockchain infrastructure and distributed compute networks, as inherently more resilient to coordinated attacks and capacity constraints affecting monolithic platforms. For institutional investors evaluating exposure to blockchain infrastructure and decentralized technology platforms, the incident underscored a fundamental value proposition: architectural resilience through distribution and redundancy addresses legitimate risks materialized in real operational failure.

The implications extend beyond theoretical advantages into competitive and strategic positioning. Companies and institutions integrating AI services into mission-critical operations face demonstrated risks from concentrated infrastructure architectures. If future outages affect ChatGPT, Claude, or Grok—whether from technical failure, cyberattacks, or nation-state action—businesses relying exclusively on these services encounter operational interruptions with compounding costs. Decentralized AI networks, by contrast, distribute compute and service delivery across geographically dispersed, independently operated nodes with varied infrastructure suppliers and redundancy architectures. This model sacrifices some raw performance optimization for systemic resilience, a tradeoff increasingly valuable in risk-conscious enterprise environments. Institutional investors with exposure to blockchain platforms positioning themselves as decentralized AI infrastructure providers gained a concrete operational rationale for those platform evaluations.

The incident also raised questions about supply chain vulnerabilities and technology concentration risk that extend beyond immediate AI service availability. The reliance of three competing major AI platforms on Nvidia hardware, combined with the geographic clustering of compute facilities, creates technological and geopolitical risk vectors that governments, regulators, and institutional stakeholders now must formally consider. Strategic diversification of AI compute resources across multiple hardware suppliers, geographic regions, and infrastructure architectures represents a logical institutional response. Companies and investors evaluating long-term technology strategy may increasingly view decentralized alternatives and blockchain-based infrastructure not as experimental alternatives but as essential components of robust, resilient AI system architecture. The September 3rd outages provided a market-validated demonstration of centralized AI infrastructure’s vulnerability and decentralized approaches’ strategic value.

Nvidia plays “matchmaker” in Nordics as region becomes more attractive, connects GPU companies with data center operators
AI News

Nvidia plays “matchmaker” in Nordics as region becomes more attractive, connects GPU companies with data center operators

, , , , , , ,

AI NewsAugust 19, 2026·6 min read

Nvidia is actively introducing GPU holders to Nordic data center operators with available capacity, marking the chipmaker’s deepest push yet into the AI infrastructure supply chain beyond hardware sales. For institutional investors, this signals Nvidia’s shift toward controlling buildout partnerships, a structural advantage that could lock in demand and margin stability across the entire compute ecosystem.

  • Nvidia reached out to at least one data center company to broker introductions between GPU holders and Nordic operators with available physical space.
  • Three Nordic facilities, atNorth, Bulk Data Centers, and Borealis Data Center, have earned Nvidia’s DGX-Ready Colocation certification for running heavy AI workloads.
  • The Nordic region is attracting massive new AI buildout: Stargate targeting 100,000 Nvidia GPUs by end of 2026, Verne/nScale planning 4,600 Blackwell Ultra GPUs.
  • 3 Nordic data center operators certified for Nvidia’s DGX-Ready program to date
  • 100,000 Nvidia GPUs targeted by Stargate project in Norway by end of 2026
  • 1.5B euros Pure DC committed to 110-megawatt Finnish campus deployment

Nvidia is stepping beyond its traditional role as chip supplier and moving upstream into the physical infrastructure layer of artificial intelligence deployment.

According to sources familiar with the matter, the company has begun introducing GPU customers, firms that have purchased or leased its processors, directly to Nordic data center operators that have capacity and the technical capability to house these systems.

The effort represents a calculated expansion of Nvidia’s control over the entire AI value chain, from silicon design through workload placement. Rather than leaving buildout decisions to market forces, Nvidia is acting as an intermediary to ensure that customers with cash and computational demand can locate the physical infrastructure needed to deploy their hardware efficiently.

Nvidia’s DGX-Ready Certification Becomes the North European Infrastructure Standard

The matchmaking infrastructure revolves around Nvidia’s DGX-Ready Colocation Data Center program, a vetting standard the company created to ensure facilities can reliably run its most demanding AI systems.

To qualify, a data center must demonstrate liquid cooling capability, pass Nvidia’s component validation testing, and provide ecosystem support robust enough to prevent deployment failures after customers install hardware on-site. Three Nordic operators have achieved this certification: atNorth, Bulk Data Centers, and Borealis Data Center.

All three certified facilities operate entirely on renewable energy sources, a differentiator that appeals to both environmental concerns and operational economics. The Nordic location itself provides a structural cooling advantage: ambient temperatures remain cold year-round, dramatically reducing the energy required to maintain optimal operating conditions for densely packed GPU systems.

This efficiency translates directly into lower total cost of ownership for customers running intensive AI workloads, making the region competitive against traditional data center hubs regardless of hardware costs.

Nvidia’s certification program serves a dual purpose: it creates a branded standard that reinforces the company’s technical authority over infrastructure decisions, while simultaneously limiting operator competition by establishing barriers to entry that smaller or under-capitalized facilities cannot easily overcome.

For institutional investors, this certification becomes a de facto prerequisite for competing in the high-density AI deployment market.

Nordic Power Economics and Renewable Energy Attracting Billions in New Capacity

The Nordic region’s infrastructure appeal extends well beyond Nvidia’s matchmaking efforts. The combination of abundant renewable power, cold climate, and available land has triggered a wave of major capital commitments from across the industry.

In July, Pure DC announced a 1.5 billion euro investment in a 110-megawatt data center campus in Finland, representing one of the largest single infrastructure bets in the region to date. Microsoft added capacity at an nScale site in Norway in April, signaling that major cloud operators view Nordic sites as essential to meeting AI infrastructure demand.

The Stargate project, backed by OpenAI, is targeting 100,000 Nvidia GPUs deployed in Norway by the end of 2026, a commitment that would make it one of the largest single GPU installations globally.

In Iceland, a joint venture between Verne and nScale is planning what it describes as Europe’s largest liquid-cooled AI GPU installation, aiming to deploy approximately 4,600 Nvidia Blackwell Ultra GPUs. Crusoe Energy also expanded its partnership with atNorth in August 2025 to grow the ICE02 site in Iceland, which will pair Nvidia’s DGX GB200 NVL72 systems with Blackwell architecture.

These projects collectively demonstrate that the Nordic region is no longer a niche alternative but a primary deployment hub for institutional AI infrastructure.

The economic logic is straightforward: geothermal power in Iceland, hydroelectric generation across Norway and Sweden, and naturally cold ambient temperatures reduce operational expenses by 30-50 percent relative to traditional data center markets in the continental United States or Western Europe.

For institutional investors managing large-scale AI infrastructure, this cost advantage compounds annually and directly flows to the bottom line.

Nvidia’s Supply Chain Control Extends Beyond Hardware Sales

Nvidia’s shift into matchmaking reveals a strategic calculation that extends far beyond selling chips. By introducing GPU customers to vetted data center operators, the company inserts itself into the infrastructure placement decision, a position that traditionally belonged to customers themselves or to infrastructure brokers and consultants.

This moves Nvidia one layer deeper into the AI supply chain than it has previously ventured, giving it visibility and influence over where compute capacity gets deployed and at what utilization rates.

One source told CNBC that Nvidia framed these introductions as part of the “value proposition to GPU customers,” implying that access to certified, cost-efficient deployment sites becomes a selling point attached to hardware purchases.

This bundling strategy, combining silicon with infrastructure matchmaking, raises barriers for competing chip designers and effectively locks in customer relationships beyond the point of sale.

The effort is not limited to the Nordic region. Nvidia has already attempted to link AI infrastructure firms with GPU holders in the United States and Asia, indicating this is a systematic strategy rather than a regional experiment.

For institutional investors evaluating Nvidia’s competitive moat, this infrastructure play matters because it shifts the company’s leverage from hardware scarcity alone to end-to-end solution control.

Certification Standards and Capacity Constraints Shape Near-Term Deployment Patterns

With only three Nordic operators currently holding Nvidia’s DGX-Ready certification, the supply of certified capacity is still constrained relative to demand. This creates a potential bottleneck: GPU customers may have the hardware and the capital, but they cannot deploy efficiently without access to a certified facility.

Nvidia’s matchmaking effort becomes essential infrastructure plumbing in this environment, and the company’s role as the connector between supply and demand grows more critical as deployment accelerates.

The certification process itself remains proprietary to Nvidia. The company controls the technical standards, the validation process, and the decision to grant or withhold certification from any given facility.

Unlike open industry standards, this creates an ongoing dependency: operators must maintain Nvidia’s goodwill to keep certification, and customers must navigate Nvidia-approved channels to find deployment options. This structural advantage compounds as the installed base of DGX-Ready facilities grows.

The timing of these matchmaking efforts suggests Nvidia is trying to prevent a supply-side constraint from limiting GPU sales. If certified capacity remains scarce, customers cannot deploy their hardware, and sales effectively stall.

By actively connecting supply and demand, Nvidia ensures that the limiting factor remains silicon availability, not infrastructure, a scenario that favors the chipmaker over customers trying to negotiate better hardware terms.

Watch for whether additional Nordic operators pursue DGX-Ready certification over the next 12 months, and whether Nvidia’s matchmaking efforts expand to other regions. The pace of new certification will reveal whether Nvidia intends to maintain tight control over capacity or gradually democratize access. Equally important: track whether certified facilities begin negotiating terms directly with customers, or whether they remain dependent on Nvidia introductions for deal flow. The answer will show whether Nvidia has built a durable structural advantage in infrastructure control or a temporary convenience for an immature market.

Get this in your inboxThe Crypto Coin Show newsletter covers the policy and market moves institutional crypto investors are pricing in.

Subscribe

Google Cloud eyes enterprise distribution in $100M deal with self-improving AI builder Mirendil
Institutional

Google Cloud eyes enterprise distribution in $100M deal with self-improving AI builder Mirendil

, , , , , ,

AI NewsAugust 6, 2026·5 min read

Mirendil, an AI lab founded by former Anthropic researchers, has secured a multi-year Google Cloud partnership valued at over $100 million to power its self-improving AI systems. For institutional investors and enterprise customers, the deal signals both a validation of recursive self-improvement as a near-term commercial technology and Google’s strategic pivot toward infrastructure plays that lock in hardware adoption through software partnerships.

  • Mirendil raised $200 million at $1 billion valuation in June; Google deal represents roughly 50% of seed capital committed to single infrastructure agreement
  • Partnership grants access to Google TPUs and Nvidia GPUs with managed training clusters, emphasizing hardware flexibility and workload optimization across chips
  • Self-improving AI already showing measurable commercial traction: Anthropic reports 80% of code merged in May 2026 written by Claude models, up from single digits in early 2025
  • $100M+ Google Cloud commitment as percentage of Mirendil’s entire seed funding round
  • 80% Anthropic code merged by AI models as of May 2026, versus prior single-digit baseline
  • $1B Mirendil valuation at seed stage, one of largest AI startup seed rounds on record

Mirendil, the AI research lab founded by former Anthropic researchers, has committed to a multi-year partnership with Google Cloud worth more than $100 million, securing computational infrastructure for its self-improving AI models.

CEO Behnam Neyshabur confirmed the deal, which provides access to Google’s Tensor Processing Units (TPUs) and Nvidia graphics processors alongside managed training clusters.

The agreement represents a substantial portion of the company’s recently closed seed funding and underscores growing institutional confidence in recursive self-improvement as a near-term technology capable of automating research workflows across medicine, biology, chemistry, and materials science.

Mirendil commits half its seed capital to single Google Cloud infrastructure deal

The $100 million Google commitment amounts to roughly 50 percent of Mirendil’s $200 million seed round, which closed in late June at a $1 billion valuation. That round, led by Andreessen Horowitz with participation from Kleiner Perkins and Nvidia, represented one of the largest seed financings for an AI startup to date.

By allocating such a substantial portion of early capital to a single infrastructure partner, Mirendil’s founders are signaling both the computational intensity of their research agenda and confidence in Google’s ability to deliver the horsepower required for recursive self-improvement at scale.

The concentration of spend reflects the economics of frontier AI development, where training and inference costs dominate operational budgets. For Mirendil, the Google arrangement locks in multi-year access to heterogeneous compute, both TPUs and Nvidia GPUs, at a time when capacity remains constrained and pricing volatile.

Cofounder Harsh Mehta emphasized that the company’s models are architected to distribute workloads efficiently across different chip types, lowering costs not only for Mirendil’s internal research but for enterprise customers licensing its systems.

Self-improving AI moves from research concept to measurable productivity metric

Recursive self-improvement refers to AI systems that iteratively refine and redeploy themselves, becoming progressively more capable without constant human retraining or intervention. Mirendil’s stated ambition is to build models capable of performing the full range of work conducted by a frontier research laboratory, from hypothesis generation through experimentation and iteration.

CEO Neyshabur, a former Anthropic research scientist and co-inventor of the Sharpness-Aware Minimization optimization algorithm, articulated the vision plainly: “You can have a self-improving AI where you can point a problem at it, and it keeps getting better with time.” His example was Alzheimer’s research, a domain where self-improving systems could autonomously pursue open-ended scientific goals without constant human guidance.

Anthropic has already demonstrated measurable traction with this model. As of May 2026, more than 80 percent of code merged into Anthropic’s systems was written by its Claude models, compared to single-digit percentages in early 2025.

The company acknowledged that this metric likely overstates true productivity gains, yet confirmed that the acceleration is genuine: engineers are now merging approximately eight times as much code per day as they were a year prior.

This data point carries weight for institutional investors evaluating whether self-improvement remains theoretical or has entered the realm of practical enterprise deployment.

Mirendil’s cofounder Harsh Mehta contributed to Anthropic’s internal automation efforts, giving the company direct operational experience with self-improving workflows. That institutional knowledge, combined with access to Anthropic’s research legacy, positions Mirendil to compress the timeline from prototype to production deployment.

The Google partnership accelerates that path by removing infrastructure barriers.

Google positions self-improving AI as hardware lock-in and enterprise upsell opportunity

Google’s $100 million commitment is not purely philanthropic. For the cloud giant, Mirendil represents a strategic customer whose software layer can drive adoption of Google’s own silicon. Neyshabur noted that Mirendil’s workload optimization capabilities help customers extract more performance from Google’s chips, providing Google with a competitive pitch against rival cloud providers.

In a market where raw chip performance is increasingly commoditized, software that can orchestrate heterogeneous compute more efficiently becomes a differentiator worth funding directly.

Amin Vahdat, Google’s senior vice president and chief technologist for AI and infrastructure, framed the partnership in systemic terms. Progress in AI infrastructure, he indicated, is no longer driven solely by chip-level performance but by the ability to orchestrate entire systems, balancing compute, memory, networking, and scheduling across complex workloads.

Mirendil’s software architecture, tested against both TPUs and Nvidia GPUs, offers Google a vehicle to demonstrate that thesis to enterprise customers. When those customers adopt Mirendil’s models, they simultaneously adopt Google Cloud as the optimal deployment platform.

The deal also grants Google early exposure to recursive self-improvement at scale. Should Mirendil’s research succeed, Google gains both a strategic partner and a template for understanding how self-improving systems operate within its cloud infrastructure.

That knowledge compounds in value as self-improving AI moves from research labs into commercial deployments across pharmaceuticals, materials science, and research institutions.

Institutional investors face new metrics for evaluating self-improving AI runway and unit economics

The Mirendil-Google deal establishes a practical pricing baseline for self-improving AI infrastructure partnerships. At $100 million over multiple years, the deal implies annualized commitment in the tens of millions of dollars for a company still in early-stage product development.

Investors evaluating AI infrastructure plays must now assess whether early-stage AI firms can sustain multi-year infrastructure commitments while simultaneously building enterprise sales organizations and navigating regulatory scrutiny.

Anthropic’s disclosed metrics on code generation and engineer productivity suggest that self-improving systems do accelerate output measurably, even as the company cautioned against overstating gains.

For institutional investors, the critical question is whether Mirendil’s models will achieve similar or superior productivity metrics in domains like drug discovery or materials science, where economic value per unit of acceleration is orders of magnitude higher than software development.

A self-improving system that accelerates therapeutic discovery by even 10 percent could justify substantially larger infrastructure spending.

The heterogeneous compute architecture, spanning both Google TPUs and Nvidia GPUs, also signals a broader industry shift. Rather than betting on a single chip manufacturer, Mirendil’s design assumes a multi-vendor landscape and optimizes for portability.

This approach reduces lock-in risk for enterprise customers and suggests that the AI infrastructure market is moving toward abstraction layers and scheduler-level optimization, not raw silicon dominance.

Watch for Mirendil’s first disclosed customer deployments and performance benchmarks in pharmaceuticals or materials science over the next 12-18 months. Those metrics will determine whether Mirendil’s self-improving architecture translates competitive advantage in research workflows into enterprise revenue sufficient to justify its $1 billion valuation and the implied $50+ million annual spend on Google infrastructure. Google’s willingness to expand or renew the commitment at higher levels would signal internal validation of recursive self-improvement’s commercial viability; conversely, any public shift to alternative compute partners would suggest the technology faces unforeseen deployment challenges.

Get this in your inboxThe Crypto Coin Show newsletter covers the policy and market moves institutional crypto investors are pricing in.

Subscribe

Up 439%, Then Margin-Called: Did Leopold Aschenbrenner’s Situational Awareness Actually Blow Up?
Blockchain

Up 439%, Then Margin-Called: Did Leopold Aschenbrenner’s Situational Awareness Actually Blow Up?

, , , , , ,

Agentic FinanceJuly 30, 2026·5 min read

Leopold Aschenbrenner’s AI-focused hedge fund Situational Awareness collapsed in July 2024 after a 439% gain, forced to liquidate all stock holdings via margin call when semiconductor bets soured. Ken Griffin’s Citadel acquired the entire portfolio in a single trade, and Citadel’s subsequent Bitcoin miner holdings suggest the crypto sector may have played a larger role in the fund’s strategy than initial disclosures revealed.

  • Situational Awareness grew to $45 billion by early July 2024, up 450% year-to-date, before margin calls forced complete liquidation
  • Citadel purchased all of Situational Awareness’ disclosed stock positions in a single transaction after other major funds declined
  • Citadel’s Bitcoin miner holdings comprised approximately one-quarter of its last reported stock portfolio, signaling undisclosed crypto exposure in the original fund
  • 439% Six-month gain for Situational Awareness before margin call forced liquidation
  • $45B Fund size at peak in early July 2024, before complete unwinding
  • 25% Bitcoin miner allocation in Citadel’s reported holdings after acquisition

Leopold Aschenbrenner’s Situational Awareness hedge fund collapsed in late July 2024, liquidating its entire public stock portfolio through forced margin calls despite generating a 439% return over six months.

The fund, which had swelled to $45 billion by the start of July, faced mounting losses as its concentrated bets on artificial intelligence-related semiconductors and leveraged short positions on software stocks backfired during a sector-wide selloff.

Bank of America, Goldman Sachs and JPMorgan Chase issued margin calls after the Nasdaq-100 fell 10% from its early June peak, forcing Aschenbrenner to unwind every position. Ken Griffin’s Citadel hedge fund acquired the entire liquidated portfolio in a single transaction, while Millennium Management and Jane Street declined to bid.

OpenAI Superalignment researcher pivots to AGI-driven hedge fund thesis

Aschenbrenner, who was hired by OpenAI to work on its Superalignment team in 2023 before being let go in April 2024, framed his departure as a result of raising AI safety concerns internally. In June 2024, he published an influential essay series titled Situational Awareness that became the intellectual foundation for the hedge fund.

His central thesis was blunt: “AGI by 2027 is strikingly plausible,” he wrote, referring to artificial general intelligence that would match human capability across most cognitive tasks.

The essay did more than forecast AGI timelines; it identified a specific bottleneck that Aschenbrenner believed markets had mispriced. Rather than chips themselves, he argued, the true constraint would be physical infrastructure: power contracts, electrical transformers, and energy supply chains. This analysis directly informed his fund’s initial positioning.

Situational Awareness’ first stock disclosure, covering December 2024, listed six holdings totaling $254.8 million, every one concentrated in power generation or semiconductor manufacturing firms, with zero exposure to crypto assets in that filing.

The thesis attracted significant capital rapidly. The fund’s stated strategy was to capitalize on what Aschenbrenner viewed as a structural supply crunch in electricity and computing hardware.

However, the concentrated bets on memory chip makers such as SK Hynix, combined with leveraged short positions in software firms like Adobe, left the fund acutely vulnerable to rotation trades and sector reversals.

July market reversal exposes leverage and margin call cascade

July proved catastrophic for the fund’s positioning. A broad selloff across AI-related semiconductor stocks, which fell more than 30% in a two-week span, directly targeted Situational Awareness’ largest long positions. SK Hynix, a memory chip manufacturer central to the fund’s thesis, suffered particularly sharp declines.

Simultaneously, Aschenbrenner’s short positions in software stocks moved sharply against him as software equities outperformed during the broader rotation.

The fund’s use of borrowed money to amplify its stock positions transformed a painful month into an existential one. By leveraging its capital through margin loans from major Wall Street brokers, Situational Awareness had been able to control positions far larger than its underlying equity cushion. When collateral values dropped, the mathematics of margin became unforgiving.

Lenders demanded additional capital to maintain the loans, a standard margin call that the fund could not or chose not to meet.

CNBC reported that Bank of America, Goldman Sachs and JPMorgan Chase all issued margin calls simultaneously. Faced with the choice of raising emergency capital or liquidating positions at steep losses, Aschenbrenner chose total unwinding. Citadel, the multi-strategy hedge fund run by Griffin, moved to acquire every disclosed holding in a single transaction.

The speed and size of the transaction signaled confidence in the underlying portfolio quality, even as it underscored Situational Awareness’ inability to meet its obligations. Bloomberg reported that Millennium Management and Jane Street, two other major hedge funds, had reviewed and declined to bid on the positions.

Bitcoin miners in Citadel’s portfolio signal undisclosed crypto positioning

The Situational Awareness collapse generated significant institutional attention within crypto markets, not because of its stated semiconductor thesis, but because of what happened after Citadel took control.

Citadel’s subsequent holdings disclosures showed that Bitcoin miners represented approximately one-quarter of its reported stock portfolio, a far larger allocation than typical for the diversified hedge fund.

This Bitcoin miner concentration was not present in Situational Awareness’ initial December 2024 filing, which contained zero crypto-related holdings. The emergence of major miner positions in Citadel’s portfolio after the acquisition suggested two possibilities.

Either Citadel had independently initiated a Bitcoin miner strategy after acquiring Aschenbrenner’s book, or the original Situational Awareness fund had held significant undisclosed crypto exposure, potentially through private positions, derivatives, or alternative vehicles not captured in its quarterly stock filings.

The timing and scale of Citadel’s miner holdings warrant scrutiny among institutional investors tracking crypto market structure. Bitcoin miners serve as a crucial bridge between crypto adoption rates and traditional energy and semiconductor markets, making them strategically relevant to institutional portfolios betting on AGI infrastructure theses.

If Aschenbrenner had indeed held undisclosed miner exposure, it would reframe the fund’s failure not as a simple leverage miscalculation, but as a mispriced exposure across multiple asset classes simultaneously.

Citadel’s acquisition and potential undisclosed crypto leverage

Citadel’s willingness to acquire Situational Awareness’ entire position in a single trade, despite the fund’s leverage collapse, indicated confidence that the underlying long book retained value even at forced liquidation prices.

The transaction size and speed suggested Citadel was taking a strategic view of infrastructure plays across semiconductors, power generation, and potentially mining operations. The subsequent prominence of Bitcoin miners in Citadel’s disclosed holdings added a layer of complexity that raises questions about how much crypto exposure had been embedded in the original fund.

For institutional crypto investors, the Situational Awareness implosion serves as a cautionary example of how leverage amplifies thesis risk, particularly in concentrated portfolios. However, the Bitcoin miner concentration that emerged in Citadel’s book after the acquisition suggests that AGI infrastructure theses may naturally pull investors toward crypto-adjacent asset classes.

This connection between semiconductor supply constraints, electricity demand, and Bitcoin mining efficiency may not be incidental but structural.

The key outstanding question for institutional investors tracking this episode is whether the Bitcoin miner positions now held by Citadel represent a standalone conviction play or a continuation of Aschenbrenner’s original infrastructure thesis. Citadel has not publicly disclosed the timing or motivation behind its miner acquisitions, leaving open whether they were part of the original Situational Awareness portfolio, an immediate add after taking control, or a subsequent strategic reposition. Any commentary from Citadel management on how it views Bitcoin mining within its broader AI infrastructure thesis could clarify whether the Situational Awareness collapse represents a temporary dislocation in a sound structural trade, or whether fundamental thesis errors triggered the margin call.

Get this in your inboxThe Crypto Coin Show newsletter covers the policy and market moves institutional crypto investors are pricing in.

Subscribe

The brutal $346M math behind Galaxy’s high-stakes race to build CoreWeave’s Texas AI mega-center
Blockchain

The brutal $346M math behind Galaxy’s high-stakes race to build CoreWeave’s Texas AI mega-center

, , , , ,

FundraisingJuly 26, 2026·5 min read

Galaxy Digital is financing a $3.507 billion AI data-center build in Texas with debt carrying a 9.875% coupon, roughly $346 million in annual cash interest, that begins flowing to creditors in 2027 while principal repayment waits until construction ends. For institutional investors, the structure reveals both the capital intensity of competing in AI infrastructure and the execution risk now embedded in one of crypto’s largest firms outside pure trading or custody.

  • Galaxy subsidiary priced $3.507 billion in senior secured notes at 9.875% coupon on July 23, closing July 28.
  • Annual cash interest of $346.3 million begins February 1, 2027, while principal amortization defers until project completion.
  • CoreWeave committed 260 MW of critical IT load for Phase II with deliveries expected in first half of 2027.
  • $3.507B Total debt financing for CoreWeave Texas data-center phases versus prior Bitcoin-mining sector deals
  • 9.875% Annual coupon rate on Galaxy’s senior secured notes, reflecting current credit risk premium
  • 400 MW Target utility capacity across eight data halls, representing scale of AI infrastructure buildout

Galaxy Digital’s financing for CoreWeave’s Helios campus in Texas marks a structural watershed for crypto-native capital allocation: interest payments begin immediately upon note issuance in 2027, but the company will not repay principal until the data center is built and operational.

The $346.3 million annual interest bill, paid semiannually starting February 1, 2027, represents a fixed cash drain on Galaxy regardless of CoreWeave’s progress or performance. The 9.875% coupon reflects the credit risk and duration that creditors are pricing into a construction project that may not generate revenue for two to three years after initial completion.

The deal structure splits cash flows into two distinct phases. Interest accrues and pays in cash on a fixed schedule. Principal amortizes at 4 percent of the original $3.507 billion annually, roughly $140.28 million per year, but only after the Helios campus reaches operational status.

That deferred repayment schedule places Galaxy’s construction timeline at the center of its debt service obligations. The senior secured notes mature August 1, 2031, giving Galaxy approximately five years from issuance to repay the full balance.

Creditors hold first-priority claims on nearly all project assets and Galaxy’s equity stake in the issuer, but the liens do not extend to Galaxy Digital’s broader corporate assets.

CoreWeave’s 260 MW commitment anchors Phase II revenue but leaves delivery timing as the critical variable

CoreWeave committed to approximately 260 MW of critical IT load for Phase II of the Helios campus in April 2025, with terms substantially similar to its previously announced 15-year, 133 MW Phase I agreement. That cumulative 393 MW of committed load, spread across two phases, gives the project a defined customer base and revenue runway.

Galaxy reported on July 6 that Phase I had been completed on schedule, positioning Phase II deliveries to begin in the first half of 2027. The timing alignment matters because CoreWeave’s rental payments begin flowing only when Galaxy delivers colocation capacity.

The data center is designed for 400 megawatts of utility capacity and 260 MW of critical IT capacity across eight data halls at the Helios campus. Galaxy’s previous AI infrastructure deals, particularly the Phase I agreement, establish a template: long-term leases with predictable revenue but operational execution risk.

The spread between utility power available (400 MW) and committed critical IT load (260 MW) suggests either additional capacity for future customer commitments or built-in redundancy and cooling overhead typical of hyperscale data centers.

Institutional investors tracking Galaxy’s debt coverage ratios should note that CoreWeave revenue does not flow until Phase II capacity is delivered. The company must service $346 million in annual interest starting in less than a year while still completing construction.

That timing compression raises the stakes on Galaxy’s project execution and CoreWeave’s ability to occupy and deploy hardware in the first half of 2027 as planned.

9.875% coupon reflects data-center leverage and competitive pressure in AI infrastructure financing

The coupon on Galaxy’s notes sits materially higher than U.S. Treasury yields and even investment-grade corporate debt, pricing in both the leverage of a construction-phase project and the capital intensity of competing in AI infrastructure. For comparison, Galaxy’s previous corporate-level debt and traditional infrastructure financing typically trade at lower yields.

The 9.875% rate reflects creditor concern that construction delays, CoreWeave’s inability to secure additional customers, or shifts in AI workload demand could compress project economics.

The note structure itself, with interest paid in cash upfront and principal deferred, transfers construction risk partially back to Galaxy. Creditors are effectively saying: we will collect cash interest regardless of progress, but we are pushing principal repayment downstream until you prove the asset generates revenue.

Galaxy must therefore fund operating expenses, debt service, and working capital from existing revenue sources, debt refinancing, or additional equity, all while building a $3.5 billion facility.

The senior secured status means creditors rank ahead of equity holders but take a second lien to whatever collateral CoreWeave (or other future customers) provides under their lease agreements.

First Phase I completion in July opens window for Phase II delivery but creates execution dependency

Galaxy’s announcement on July 6 that Phase I had been completed on schedule represents the first major operational proof point. Phase I’s on-time delivery suggests the contractor, supply chain, and power infrastructure can meet timelines.

However, Phase II carries additional complexity: it adds a second building with four new data halls and must scale power infrastructure from 133 MW to 260 MW of critical load. The first half of 2027 delivery target means Phase II must advance through final build-out, testing, and CoreWeave customer hardware deployment over the next 18 months.

If Phase II slips even moderately, into late 2027 or 2028, Galaxy’s interest payments continue but CoreWeave revenue remains minimal. Each quarter of delay compresses the window between project completion and the 2031 maturity date, reducing time for the company to service debt from operations.

CoreWeave’s own capital constraints and customer readiness introduce a second variable: the company may own the right to deploy capacity, but its ability to monetize it depends on customer hardware availability and integration timelines.

Institutional credit investors should track Galaxy’s quarterly construction progress reports and any CoreWeave disclosures about Phase II customer commitments or delays. The August 1, 2031 maturity date is now a hard constraint; Galaxy cannot simply extend or refinance at maturity if the project has not generated expected cash flow by then.

Galaxy’s next material test arrives in the first half of 2027, when CoreWeave Phase II deliveries are scheduled to begin and the first principal amortization payments ($140.28 million) fall due on an unknown timeline after project completion. Watch for Galaxy’s Q4 2026 and Q1 2027 earnings calls for any revision to the Phase II delivery schedule, announcements of additional CoreWeave capacity commitments, or disclosure of refinancing discussions with creditors. The company must also navigate rising competition: CleanSpark and Hut 8 have already secured multibillion-dollar AI data-center lease commitments, and CoreWeave itself has announced $20 billion in total funding, suggesting the customer is pursuing multiple build partners. Galaxy’s ability to deliver Phase II on time and retain CoreWeave’s full 260 MW commitment will determine whether the 9.875% coupon represents fair compensation or an early warning signal.

Get this in your inboxThe Crypto Coin Show newsletter covers the policy and market moves institutional crypto investors are pricing in.

Subscribe

The AI compute gap: Enterprises are buying infrastructure faster than they can measure what it costs
Blockchain

The AI compute gap: Enterprises are buying infrastructure faster than they can measure what it costs

, , , , , ,

AI NewsJuly 16, 2026·6 min read

Enterprise AI infrastructure spending is accelerating far faster than organizations can track or control its costs, with 83% of deployed GPUs running at half capacity or less and fewer than half of firms able to measure their actual compute expenses. This efficiency gap represents a critical blind spot for institutional investors evaluating AI infrastructure plays and cloud compute providers, as it signals both massive margin upside for vendors who can deliver transparency and substantial financial risk for enterprises locked into underutilized, costly deployments.

  • 83 percent of enterprises report GPU utilization at 50 percent or less, indicating severe stranded capacity in current deployments
  • 64 percent plan to switch or add infrastructure providers within twelve months, with 38 percent evaluating new vendors within the next quarter
  • Only 21 percent of surveyed enterprises run AI in production at scale, yet 45 percent plan to evaluate AI-specialized cloud providers in the coming year
  • 44% of enterprises that can rigorously track AI compute costs
  • 64% planning infrastructure provider switches or additions within twelve months
  • 21% running AI workloads in production at meaningful scale today

A survey of 107 enterprises with more than 100 employees reveals a widening chasm between the speed at which organizations are purchasing AI infrastructure and their ability to understand what it costs them.

While spending intentions are rising sharply, particularly toward specialized AI compute and inference layers that most firms do not yet operate, actual deployments are running cold, utilization metrics are poor, and cost visibility is rare. The result is what researchers call a “compute gap”: aggressive capital allocation meeting inadequate financial controls and measurement infrastructure.

This dynamic creates both opportunity and risk for institutional investors positioning around AI infrastructure, cloud vendors, and specialized compute providers.

GPU utilization stalled at 50 percent signals massive waste in enterprise deployments

The most striking finding from the research is the chronic underutilization of GPU infrastructure already in place. Across surveyed enterprises, 83 percent report GPU utilization rates of 50 percent or less, meaning the majority of deployed accelerators are sitting idle or underloaded most of the time.

For capital-intensive hardware purchases, utilization at that level translates directly into stranded costs: an enterprise running a $500,000 GPU cluster at 50 percent utilization is effectively paying $1 million per unit of productive compute.

The problem extends beyond simple overcapacity. Many organizations have procured hardware based on peak demand forecasts or vendor sales projections without the operational discipline or workload consolidation strategies needed to keep utilization high.

Containerization, workload scheduling, and dynamic resource allocation, standard practices in hyperscaler infrastructure, remain inconsistently deployed across enterprise environments. The result is that each percentage point of utilization improvement represents significant cost recovery and margin expansion for vendors or savings for buyers who can operationalize it.

Critically, fewer than half of all surveyed firms, just 44 percent, can rigorously track what their AI compute actually costs. Without unit economics visibility, enterprises cannot identify which workloads are efficient, which should be migrated, consolidated, or retired, or where to invest next. This measurement gap compounds the utilization problem: firms cannot optimize what they cannot see.

Majority of enterprises plan to switch cloud providers within a year despite immaturity

Perhaps more revealing than underutilization is the high velocity of planned infrastructure churn. A clear majority of enterprises, 64 percent, intend to switch or add an infrastructure provider within the next twelve months. Even more aggressive, 38 percent are evaluating new vendors within the next quarter.

For a foundational category this important, such high churn intent is unusual and signals that incumbent providers are failing to deliver on integration, cost management, or performance expectations.

The decision criteria for these switches are instructive and diverge sharply from early-stage crypto market behavior. When enterprises evaluate infrastructure, only 8 percent cite headline token price or cost-per-million-tokens as the primary deciding factor. Instead, 41 percent prioritize integration with the existing stack, and 35 percent focus on total cost of ownership.

This pattern reflects a mature buyer perspective: enterprises are comparing platforms on operational fit and lifecycle economics, not on raw commodity pricing. Vendors whose tools integrate cleanly with Kubernetes, existing data pipelines, monitoring systems, and identity infrastructure will win deals that cheaper, isolated alternatives cannot.

The churn also reflects incomplete satisfaction with current deployments. Many enterprises launched AI initiatives on hyperscaler platforms or through model provider APIs because those were the most accessible entry points.

As workloads mature, organizations are discovering that those initial choices trade flexibility and cost control for convenience, making them candidates for replacement with purpose-built infrastructure or specialized cloud providers.

45 percent of enterprises plan to evaluate AI-specialized clouds almost none use today

The single largest area enterprises plan to evaluate over the next twelve months is AI-specialized cloud providers, 45 percent of respondents flagged this as a priority, even though most of these same firms do not currently operate on such platforms.

This planning intention outpaces actual deployment by a wide margin: only 21 percent of surveyed enterprises run AI in production at scale today, yet they are already budgeting for infrastructure most have never used.

This mismatch between purchase intent and production maturity creates a critical window for specialized compute providers. Organizations are actively shopping for alternatives to hyperscaler AI offerings and model provider APIs. They are prepared to incur the switching costs and operational friction associated with migrating workloads to new platforms.

The constraint is that switching decisions will be made on integration, cost transparency, and total cost of ownership, not on raw performance or price per token.

The frontier technical challenge that will shape these purchasing decisions is barely registering on enterprise radar. As inference workloads scale, the bottleneck will shift from compute (GPUs) to memory bandwidth. Roughly one in five enterprises either are unaware of this constraint or have not begun to address it.

This gap between where the technical challenge lies and where enterprise procurement is focused suggests that many firms will invest capital in infrastructure optimized for compute when their actual constraint will be memory bandwidth, potentially leaving them with expensive, wrong-sized deployments once inference scales.

Cost accounting blindness leaves enterprises vulnerable to vendor lock-in and margin erosion

The inability of 56 percent of enterprises to rigorously track compute costs is not merely a financial control problem, it is a structural vulnerability that affects both the enterprises and the vendors serving them. When a buyer cannot measure what infrastructure costs, it cannot optimize purchasing, consolidate workloads, or make informed switching decisions.

Conversely, vendors selling into blind customers face margin pressure and customer dissatisfaction once those customers eventually gain visibility and discover they have been paying more than necessary.

This dynamic creates a near-term opportunity for infrastructure vendors and cloud providers who can offer automated cost tracking, utilization analysis, and optimization recommendations. A vendor that helps an enterprise shift from 50 percent to 75 percent GPU utilization generates immediate value and deepens switching costs through operational integration.

By contrast, vendors relying on pricing opacity or customer inertia are building unstable customer relationships that will likely rupture once measurement infrastructure arrives.

For institutional investors evaluating AI infrastructure and cloud compute companies, the compute gap signals both a market sizing opportunity and a vendor quality screen. The gap is large, billions in annual stranded capacity and untracked costs, but capturing it requires selling solutions that improve operational visibility and integration, not just cheaper hardware or lower per-token pricing.

Vendors with strong product-market fit on integration and observability will win the next wave of enterprise infrastructure spending; pure price competitors will face pressure from both customers gaining cost visibility and from market saturation as more providers enter the AI compute space.

The next critical milestone is whether enterprises actually execute on their switching plans in the coming quarter. With 38 percent of firms planning to evaluate new infrastructure providers in the next three months, the second and third quarters of 2026 will reveal whether this churn intent translates into active procurement and actual migration, or whether organizational friction and switching costs prove higher than current surveys suggest. The vendors that can demonstrate cost reduction and seamless integration in pilot deployments during this window will likely capture disproportionate share of the enterprise AI infrastructure market over the next eighteen months.

Get this in your inboxThe Crypto Coin Show newsletter covers the policy and market moves institutional crypto investors are pricing in.

Subscribe

Bitcoin Reacts As Fed Minutes Reveal Split on Rate Hikes
Bitcoin

Bitcoin Reacts As Fed Minutes Reveal Split on Rate Hikes

, , , , , , ,

BitcoinJuly 8, 2026·6 min read

The Federal Reserve’s June meeting minutes reveal internal disagreement over whether rate hikes remain necessary, a split outcome that directly influences institutional crypto portfolios through expectations of continued monetary tightness. Bitcoin’s immediate 2.7% decline signals that digital asset markets remain acutely sensitive to Fed rate-path uncertainty, making the July 28-29 FOMC decision a critical catalyst for institutional positioning.

  • Nine of 19 Fed officials now project at least one rate hike before end of 2026, reversing prior expectations of no hikes this year.
  • Core inflation remains at 3.3% to 3.4%, well above the Fed’s 2% target, driven partly by AI infrastructure spending surge.
  • Bitcoin fell 2.7% in 24 hours following the minutes release, underscoring crypto markets’ direct exposure to rate-path revisions.
  • 3.50%-3.75% Federal funds rate held steady by unanimous FOMC vote in June meeting
  • 3.3%-3.4% Core inflation running above Fed’s 2% target through May 2024
  • 9 of 19 Fed officials now penciling in at least one rate hike before end of 2026

The Federal Reserve released minutes from its June 16-17 policy meeting on July 8, exposing a committee fractured on the case for further monetary tightening.

While all 12 voting members unanimously held the federal funds rate at 3.50% to 3.75%, the published record revealed that a minority of participants had argued for a rate increase even in June, and nine of the 19 policymakers now project at least one hike before the close of 2026.

This marks a notable reversal from earlier projections that anticipated no hikes this cycle, signaling a recalibration of near-term rate expectations at a moment when institutional crypto investors are already parsing every word from Chair Kevin Warsh, who took the helm of the central bank only weeks before this meeting.

Minority Fed Officials Pressed for June Rate Hike Despite Committee Consensus

The minutes disclosed that a small but vocal group of participants had advocated for tightening at the June meeting itself, even as the broader consensus held firm.

These officials pointed to persistent inflation risks stemming from three sources: tariff pass-through into consumer prices, geopolitical energy shocks in the Middle East, and an unexpected surge in artificial intelligence infrastructure spending that has turbocharged demand for semiconductors, data centers, and electricity.

The fact that some policymakers reached for the hawkish case in June, only to acquiesce to the hold, suggests a committee closer to the tightening threshold than headline statements typically convey.

Core inflation has remained stubbornly above target, running at 3.3% in April and an estimated 3.4% in May, against the Fed’s 2% goal. Fed staff subsequently raised inflation forecasts for 2026 and 2027, reflecting their assessment that tariff dynamics, supply-side shocks, and AI-driven capex would continue to push prices higher in the near term.

Several participants acknowledged that AI investment could eventually deliver productivity gains and lower costs, but only after a lag of several years; the immediate effect remains upward pressure on demand and prices.

Warsh’s own description of the debate was notably candid. At his post-meeting press conference, the chair stated, “We had a good family fight on it for a couple of days, and we ended up, I think, in a better place,” signaling that internal disagreement was both real and, in his view, productive.

Warsh’s Refusal to Project His Own Rate Path Adds to Policy Uncertainty

A detail that attracted particular attention from markets was Warsh’s decision not to submit his own interest-rate projection to the Summary of Economic Projections, the quarterly dot plot that guides investor expectations.

His silence on his personal rate view introduced an interpretive gap: markets could not directly discern whether the new chair leaned toward the hawkish minority or the dovish mainstream. This ambiguity proved costly for risk assets, particularly cryptocurrencies, which have historically underperformed during phases of rising rate-hike probability.

The timing of Warsh’s elevation to the Fed chair role compounds the significance of this information void. Unlike his predecessor, Powell, whose communication patterns and preferences are well-established in market memory, Warsh operates with minimal track record as a central bank communicator.

His first FOMC meeting and first post-meeting press conference thus became a blank slate onto which investors projected their own assumptions. The market’s immediate reaction to the minutes preview, which flagged Warsh’s missing projection as a key uncertainty, suggested that institutional traders had been positioning for clarity that never arrived.

The nine officials now forecasting at least one hike by end of 2026 represent a meaningful shift in modal expectations within the Fed itself, even though no rate increase was delivered in June.

Bitcoin Slides 2.7% as Crypto Markets Reprice Rate-Hike Odds

Bitcoin traded near $62,240 following the minutes release, representing a 2.7% decline over the preceding 24 hours. The move occurred against a backdrop in which Bitcoin had recently rebounded toward $64,000 on the strength of bullish spot ETF inflows, suggesting that the mood swing from the Fed minutes was sufficient to reverse near-term momentum.

Options activity had turned call-heavy in the days before the release, indicating that traders had been positioning for a dovish outcome; the split revealed by the minutes punished that bet.

The episode underscores a persistent structural sensitivity of digital asset markets to Federal Reserve communication and rate-path expectations. During periods of rising rate hikes or extended monetary tightness, crypto valuations face headwinds due to higher opportunity costs for yield and reduced risk-asset demand.

This relationship held even during the 2023-2024 period in which rate cuts were eventually delivered; any surprise hawkish signal from the Fed immediately triggered sell-offs in Bitcoin and other risk assets.

For institutional portfolios that carry significant crypto allocations, the Fed minutes thus function as a direct input to trading and positioning decisions, much as they do for equities or bonds.

The broader sensitivity of crypto to Fed communications has its roots in the asset class’s behavior during the monetary expansion of 2020-2021 and the subsequent tightening cycle of 2022-2023.

Bitcoin and Ethereum rallied sharply during the zero-rate era and declined sharply during the hiking cycle, establishing a pattern that remains embedded in market behavior even as rate cuts and potential easing loom.

Inflation, AI Capex, and the Path to July 28-29 FOMC Decision

The July 28-29 FOMC meeting looms as the next critical date for rate-path determination. Between now and that decision, two major data releases will inform the committee’s deliberations: the June and July employment reports and the June and July inflation readings.

The June jobs report had already been published before the minutes release; July jobs and inflation data will arrive in early August, just before the July 28-29 meeting concludes.

If core inflation remains elevated or re-accelerates, the case for at least one hike in the second half of 2024 or early 2025 would strengthen, bolstering the position of the nine officials who now project tightening. Conversely, if inflation rolls over and employment cools, the median committee view would likely shift back toward a patient, on-hold stance.

The role of AI infrastructure spending in the inflation picture adds an unusual twist: Fed staff and several policymakers have flagged AI investment as a source of near-term price pressure, yet none can say with confidence how long that effect will persist or whether it might eventually be overwhelmed by productivity gains.

For institutional crypto investors, the key question is whether Warsh’s next communication, his prepared remarks ahead of the July 28-29 meeting or the post-meeting press conference itself, will clarify his own rate bias.

Warsh’s silence on his dot-plot projection and his somewhat cryptic description of the Fed’s internal debate have left markets searching for signals of his leanings. The crypto sell-off following the minutes release suggests that institutional traders are braced for a more

Get this in your inboxThe Crypto Coin Show newsletter covers the policy and market moves institutional crypto investors are pricing in.

Subscribe

Galaxy Digital completes Phase I of Helios AI data center, delivers 133 MW of power to CoreWeave
Blockchain

Galaxy Digital completes Phase I of Helios AI data center, delivers 133 MW of power to CoreWeave

, , , , ,

AI NewsJuly 7, 2026·5 min read

Galaxy Digital has completed Phase I of its Helios AI data center in West Texas, delivering 133 megawatts of compute capacity to CoreWeave under a 15-year lease, a watershed moment for the company’s transition from cryptocurrency mining to infrastructure-as-a-service for artificial intelligence. The project stayed on budget and timeline, positioning Galaxy to generate over $1 billion in average annual revenue while reducing its earnings volatility to crypto market swings.

  • Phase I power becomes operational at Helios, supplying 133 MW to CoreWeave with billing starting Q2 2026
  • Galaxy invested $350 million in equity and secured $1.4 billion debt financing at 80% loan-to-cost ratio
  • Phase II construction underway in greenfield stage; data hall deliveries expected mid-2027 with full campus reaching 526 MW
  • 133 MW of critical IT capacity delivered to CoreWeave under fifteen-year lease agreement
  • $1.4B debt facility closed to finance remaining construction costs beyond equity commitment
  • 526 MW total CoreWeave allocation across all three phases once complete through 2027

Galaxy Digital has crossed a critical threshold in its strategic pivot from cryptocurrency mining operations to institutional AI infrastructure provision.

The company announced that Phase I power at its Helios data center campus in West Texas is now fully operational, making available 133 megawatts of compute capacity to CoreWeave, a leading provider of cloud infrastructure for generative AI workloads. Under the terms of a 15-year lease agreement, CoreWeave will begin billing Galaxy for the power in the second quarter of 2026.

This completion marks the physical culmination of a transformation that began in 2022, when Galaxy acquired the site, originally a top-tier 180-megawatt Bitcoin mining facility, for $65 million, then systematically repurposed it for AI infrastructure rather than cryptocurrency operations.

The scope of Galaxy’s capital commitment underscores the company’s conviction in the long-term profitability of the shift. The company deployed $350 million of its own equity into Phase I, with the remainder of construction costs financed through a $1.4 billion debt facility structured at an 80% loan-to-cost ratio.

Reaching financial close on that debt facility, combined with meeting the project’s timeline and budget targets, represents a significant validation of both Galaxy’s operational execution and the underlying demand for high-density power infrastructure tailored to AI workloads.

The Helios campus, which spans more than 2,200 acres, sits on approved power capacity of 1.63 gigawatts with a path to expand to 3.6 gigawatts.

The project’s full completion will dwarf today’s initial deployment.

Galaxy Executive Projects Over $1 Billion in Annual Revenue From CoreWeave Deal

Mike Novogratz, Galaxy’s founder and CEO, framed the Phase I completion as confirmation that the company can execute at hyperscale in infrastructure development, a capability required to win long-term contracts with institutional AI platforms.

In recent commentary, Novogratz emphasized that demand for high-density, AI-ready power represents a structural economic shift rather than a temporary cycle. He further noted that Galaxy now operates as an approximately even mix of data infrastructure revenue and digital asset holdings, reducing the company’s dependence on cryptocurrency market sentiment to drive earnings.

The demand for high-density, AI-ready power is not a cycle; it is a structural shift, and Galaxy is built to meet it.

Mike Novogratz, Founder and CEO, Galaxy Digital

Galaxy expects the full Helios-CoreWeave relationship to generate more than $1 billion in average annual revenue once all three phases reach completion. That figure carries significant weight for a company that reported a $216 million net loss in the first quarter of 2025, driven substantially by declines in cryptocurrency prices.

By diversifying revenue into long-term, contracted AI infrastructure leases, Galaxy aims to decouple its financial performance from the volatility of digital asset markets and establish a stable, recurring revenue base comparable to traditional data center operators.

Phase II Greenfield Construction Targets Mid-2027 Data Hall Deliveries

With Phase I now operational, Galaxy’s engineering and construction teams have shifted focus to Phase II, which remains in greenfield stage with structural work currently underway. The company expects to begin delivering data halls in that phase by mid-2027.

Once Phase II and Phase III are complete, CoreWeave will control 526 megawatts of critical IT load across the entire Helios campus, exhausting the site’s 800 megawatts of approved gross power allocation.

That sequential buildout reflects both the engineering and financial reality of hyperscale data center development: phasing construction, power delivery, and lease activation allows Galaxy to stage capital deployment and CoreWeave to activate capacity as its customer demand warrants.

The timeline carries weight for institutional investors monitoring Galaxy’s ability to convert capital expenditure into revenue generation. Each phase of completion reduces execution risk and validates the assumptions embedded in the original pro forma financial models.

Phase II’s mid-2027 data hall delivery date, if met, would continue a pattern of on-time, on-budget execution that began with Phase I, a track record increasingly important as institutional infrastructure investors scrutinize management teams’ ability to deliver large, complex projects under inflationary construction cost environments and volatile interest rate conditions.

The campus’s total approved power footprint of 1.63 gigawatts allows for staged expansion without additional regulatory hurdles.

Galaxy Introduces Institutional OTC Prediction Markets as Alternative Revenue Stream

Alongside the Helios announcement, Galaxy expanded its institutional client offerings by launching an over-the-counter prediction markets service operated through its Global Markets trading desk. The service targets hedge funds, family offices, and other institutional investors seeking to trade prediction markets at scale while maintaining enhanced privacy protections.

The offering focuses exclusively on non-sports prediction markets listed on platforms including Kalshi and Polymarket, covering categories ranging from political outcomes to economic data releases.

The move extends Galaxy’s institutional client relationships beyond infrastructure into markets and volatility trading.

This addition reflects Galaxy’s broader strategy to monetize multiple revenue lines across its digital assets and infrastructure divisions. While the prediction markets service generates smaller dollar volumes compared to the data center leasing arrangements, it reinforces Galaxy’s positioning as an infrastructure and market access provider to institutional crypto and derivatives traders.

For institutional investors evaluating Galaxy’s earnings diversification, the expanded client offerings signal management’s intent to deepen relationships across multiple product categories rather than rely solely on the Helios-CoreWeave transaction to drive revenue stability.

Galaxy’s next critical milestone arrives in Q2 2026, when Phase I billing to CoreWeave commences and the company begins recognizing recurring infrastructure revenue at scale. Investors will watch whether Phase II data hall deliveries begin on schedule in mid-2027, as any delays would signal execution challenges and compress the timeline to profitability; simultaneously, CoreWeave’s actual consumption of the 133 MW delivered in Phase I will indicate real demand dynamics versus contracted capacity, providing the first empirical test of Galaxy’s thesis that AI infrastructure represents a durable, non-cyclical revenue source.

Get this in your inboxThe Crypto Coin Show newsletter covers the policy and market moves institutional crypto investors are pricing in.

Subscribe

Why’s Terawulf stock surging today?
Equities

Why’s Terawulf stock surging today?

, , , , , ,

AI NewsJuly 6, 2026·5 min read

TeraWulf locked in $19 billion in contracted revenue over two decades from AI infrastructure leader Anthropic, fundamentally shifting its business model from volatile Bitcoin mining to predictable, long-term AI data center leasing. The deal validates the company’s infrastructure pivot while addressing institutional investor concerns about revenue stability and competitive positioning in the high-margin AI compute sector.

  • TeraWulf signed a 20-year lease with Anthropic valued at approximately $19 billion in total revenue
  • NASDAQ:WULF shares surged 17% to $24.14, reflecting 117% year-to-date gains since January 2025
  • First Anthropic services begin H2 2027; full 401-megawatt capacity operational by early 2028 at Hawesville facility
  • $19B Total contracted revenue from Anthropic over the full 20-year lease term
  • 117% Year-to-date stock appreciation since the start of 2025 through announcement day
  • 401MW Critical IT capacity at Hawesville campus when fully operational in early 2028

TeraWulf announced Monday that it has secured a two-decade lease agreement with Anthropic, the maker of the Claude family of large language models, to operate a purpose-built AI infrastructure campus at its Justified Data site in Hawesville, Kentucky.

The deal is expected to generate approximately $19 billion in total revenue for TeraWulf and represents a watershed moment for the company’s transformation from Bitcoin mining into the higher-margin, more stable AI data center market. The announcement drove TeraWulf’s stock up 17 percent to $24.14 by late Monday morning, extending its year-to-date rally to 117 percent.

Anthropic Campus Provides $19 Billion Revenue Foundation Over Two Decades

The Hawesville facility will be developed in stages, with Anthropic’s first services commencing in the second half of 2027 and the full 401 megawatts of critical IT capacity becoming operational in early 2028. That capacity level positions the site to support large-scale AI model training workloads, addressing the intense computational demands of frontier large language models.

For institutional investors monitoring data center supply and demand dynamics, the deal demonstrates that Anthropic, now valued at roughly $40 billion in private markets, is willing to commit to long-term, fixed-capacity arrangements at scale.

The contract’s 20-year term is the critical element here. Bitcoin mining, TeraWulf’s original business, generates volatile returns tied directly to cryptocurrency prices and mining difficulty, making revenue projections unreliable.

By contrast, a fixed lease with a major AI model developer locks in predictable cash flows across two decades, substantially reducing business risk and improving the company’s ability to service debt and fund future expansion.

Institutional equity analysts have flagged revenue visibility as a key valuation concern for data center operators; this deal substantially addresses that concern for TeraWulf.

The choice of Anthropic as an anchor tenant carries additional weight. The company competes directly with OpenAI and other large language model developers and has secured commitments from Google, Salesforce, and other major enterprise customers.

By betting on Anthropic’s continued prominence in AI over the next 20 years, TeraWulf is implicitly wagering on a specific vendor’s survival and competitiveness, a real risk, but one mitigated by Anthropic’s current market position and funding depth.

TeraWulf Exits Texas Joint Venture to Control Full Ownership and Operations

On the same day, TeraWulf announced the sale of its 50.1 percent stake in an AI data center joint venture located in Abernathy, Texas, to an investor group led by Fluidstack, its co-investor in the project. While the company did not disclose the transaction’s financial terms, it stated that it realized a premium over the approximately $450 million it had invested in the venture.

The sale represents a deliberate shift in capital deployment strategy, prioritizing sites where TeraWulf maintains complete operational control and direct customer relationships over shared ownership structures.

This move reveals management’s conviction that owning infrastructure outright, rather than sharing it with joint venture partners, is the superior long-term position. Joint ventures introduce governance friction, complicate customer negotiations, and limit pricing power.

By exiting Abernathy and redploying the sale proceeds into wholly owned platforms, TeraWulf gains the flexibility to negotiate customer terms more aggressively and avoid disputes with partners over capacity allocation or pricing.

For institutional investors, it signals management clarity about the competitive dynamics of AI infrastructure and confidence in TeraWulf’s ability to build owned assets at scale.

The pairing of the two transactions reveals a coherent capital strategy. TeraWulf is simultaneously locking in long-term, predictable revenue (the Anthropic lease) while shedding partnerships that dilute ownership and control.

In the company’s own language, the moves “enhance TeraWulf’s long-term revenue visibility, strengthen its financial position, and further align the Company’s capital with infrastructure platforms where it maintains direct ownership, customer relationships, and operational control.”

Bitcoin Mining Revenues Overtaken by AI Data Center Transition in Q1 2025

The Anthropic deal caps a dramatic business model pivot. In Q1 2025, TeraWulf generated $21 million in revenue from high-performance computing hosting services while deriving roughly $13 million from Bitcoin mining operations, the first quarter in which AI data center revenues exceeded mining revenues.

The company’s original footprint included Bitcoin mining facilities in New York and Pennsylvania; those assets are now being repurposed or complemented by AI infrastructure builds designed to accommodate far larger, longer-term customers with different technical requirements.

The transition has carried steep near-term costs. In Q1 2025, TeraWulf reported net losses of $427.63 million as it invested heavily in buildout and development. The company ended the quarter with approximately $3.1 billion in cash and an equivalent amount in long-term debt, creating a leveraged position that makes long-term revenue certainty essential to financial sustainability.

Anthropic’s contract directly addresses that imperative by providing decades of predictable cash inflow.

TeraWulf is not alone in this transition. Numerous Bitcoin mining operators, including Core Scientific, Hut 8, and others, have pivoted toward offering power and space to AI firms as a way to stabilize revenues and escape the cyclicality of cryptocurrency price swings.

Data Center Demand Surge Underpins AI Infrastructure Expansion Across Industry

TeraWulf’s ability to attract Anthropic and other AI customers reflects a structural undersupply of high-performance data center capacity. The International Energy Agency has projected that data center electricity consumption will nearly double from current levels to approximately 945 terawatt-hours annually, driven primarily by AI model training and inference workloads.

That growth trajectory supports the economics of long-term leasing arrangements and provides confidence that demand will not suddenly evaporate.

The scarcity of available compute capacity is acute. Major AI labs including OpenAI, Google DeepMind, Meta, and Anthropic have all publicly signaled that infrastructure availability is a constraint on their ability to scale research and commercialize new models.

For TeraWulf, the timing of the Anthropic deal also reflects favorable financing conditions. The company is less dependent on volatile equity markets to fund expansion if it can use contracted revenues to secure debt financing.

Banks and institutional debt investors are more willing to lend against predictable, long-term lease revenues than against speculative mining operations, further reducing TeraWulf’s cost of capital and competitive disadvantage relative to larger data center operators like Equinix or CoreWeave.

Phased Delivery Timeline Extends Revenue Recognition Through 2028 and Beyond

The Hawesville facility’s staged rollout, first services in H2 2027 and full capacity by early 2028, means TeraWulf will begin recognizing Anthropic revenues in roughly two years. That timeline provides critical breathing room for the company to manage its leverage and reduce reported losses as AI services revenue ramps.

It also creates quarterly guidance visibility for analyst consensus, a key metric that institutional equity investors use to assess management’s execution risk.

The phased approach

Get this in your inboxThe Crypto Coin Show newsletter covers the policy and market moves institutional crypto investors are pricing in.

Subscribe

Meta’s AI Data Center Caught Infecting Town Water Supply With Deadly Bacteria
Blockchain

Meta’s AI Data Center Caught Infecting Town Water Supply With Deadly Bacteria

, , , , , ,

AI NewsJuly 6, 2026·5 min read

Meta’s Cheyenne, Wyoming data center triggered a citywide ban on fill-and-flush discharge procedures after discharging wastewater containing a rare, multidrug-resistant bacterium linked to a 31.3% mortality rate. The incident exposes institutional crypto and AI investors to localized regulatory backlash that can shut down entire data center campuses regardless of individual operator compliance.

  • Meta-affiliated Goat Systems LLC discharged wastewater containing Cupriavidus gilardii, a multidrug-resistant pathogen linked to ten deaths including three immunocompromised children.
  • Cheyenne Board of Public Utilities revoked fill-and-flush discharge privileges for all data center campuses connected to municipal water, not just Meta’s facility.
  • The bacterium has a 31.3% mortality rate across 32 known infections since 2009, making it unusually dangerous despite extremely rare human infection.
  • 31.3% Mortality rate of Cupriavidus infections across documented cases since 2009.
  • 800,000 Square feet of Meta’s under-construction Cheyenne data center campus facility.
  • 10 Total deaths linked to Cupriavidus gilardii infections on medical record.

The Cheyenne Board of Public Utilities has imposed an industry-wide moratorium on fill-and-flush discharge procedures at all data center campuses connected to municipal water infrastructure, following the discovery that Meta-affiliated Goat Systems LLC discharged industrial wastewater containing Cupriavidus gilardii into local pipes.

The bacterium, a rare multidrug-resistant pathogen with no standard testing protocol in municipal water systems, was identified during routine fecal contamination screening and traced directly to Meta’s 800,000 square-foot campus currently under construction.

The regulatory response punishes not merely the operator responsible for the contamination, but every data center operator in Cheyenne’s jurisdiction, signaling how quickly localized environmental incidents can cascade into blanket infrastructure restrictions that threaten facility operations across an entire market.

Cheyenne officials discovered deadly bacterium in Meta wastewater during routine contamination testing

Frank Strong, the Cheyenne Board of Public Utilities engineering and water resource division manager, confirmed that municipal staff identified Cupriavidus gilardii during standard testing for fecal pathogens, a detection that surprised utilities personnel unfamiliar with the organism’s presence in industrial wastewater.

The bacterium’s identification required extended investigation; Strong noted that utilities “had to go through quite a process to figure out what it was,” underscoring that municipal labs do not routinely screen for this particular pathogen. The discovery emerged only because routine monitoring happened to flag an anomaly that prompted deeper analysis.

Cupriavidus gilardii carries clinical significance disproportionate to its rarity in human cases. Medical literature documents ten deaths attributed to the bacterium, including three fatalities among immunocompromised children, with a documented mortality rate of 31.3% across 32 confirmed infections dating back to 2009.

Though human infection remains extremely rare, the pathogen’s multidrug-resistant profile and high case fatality rate elevate its public health implications substantially above those of common waterborne contaminants.

Critically, no residents have contracted the infection from the discharge event, as the contamination was confined to municipal wastewater infrastructure rather than potable drinking water supplies.

Meta halted discharge immediately after utilities identified the contamination source

Meta and its construction contractor Fortis responded by ceasing all industrial wastewater discharge into municipal systems and shifting to offsite hauling once the contamination was confirmed.

A Meta spokesperson stated: “When the board shared that it found a substance in the city’s wastewater, not public drinking water, Fortis immediately stopped discharging industrial wastewater and began hauling it offsite.” The company characterized its posture as collaborative remediation, though it acknowledged uncertainty regarding the precise origin of the pathogen within its facility.

However, Meta’s operational response proved insufficient to prevent regulatory escalation. The Cheyenne Board of Public Utilities did not restrict the offending operator alone; instead, it enacted a sweeping prohibition on fill-and-flush discharge procedures across all data center campuses connected to municipal water infrastructure.

This blanket approach means that compliant operators face operational constraints imposed by a single facility’s noncompliance, creating negative externalities that extend beyond the responsible party.

Goat Systems LLC was found in “significant noncompliance” with Cheyenne’s industrial waste regulations following months-long investigation, establishing clear regulatory violation independent of whether the facility remains operational.

Cheyenne’s industry-wide ban signals growing municipal resistance to data center water practices

The Cheyenne incident reflects escalating public and regulatory scrutiny of data center environmental footprints across the United States.

Fill-and-flush procedures, in which facilities flood cooling systems with water before initial startup, are standard industry practice, yet they concentrate significant volumes of untreated or partially treated wastewater into municipal systems during construction phases.

The procedure creates concentrated discharge events that strain municipal infrastructure designed for gradual, baseline flows rather than episodic industrial loading.

Institutional investors with exposure to data center operators, whether crypto mining firms, AI infrastructure companies, or traditional hyperscale facilities, now face a replicable regulatory model. One contamination event sufficient to trigger public health concern can generate blanket restrictions affecting all operators in a jurisdiction, regardless of their individual compliance records.

This dynamic introduces regulatory tail risk: even operators with robust environmental protocols face operational constraints and capital redeployment costs when municipal authorities respond to incidents originating elsewhere in the same market.

The Cheyenne precedent also illustrates how environmental incidents can accelerate from facility-specific enforcement to policy-level restrictions within months.

The investigation timeline, from initial contamination detection through regulatory response, compressed permit-level adjustments into a compressed window, suggesting that future similar incidents may trigger faster, broader policy changes as municipal officials preempt rather than respond to public health concerns.

Meta’s facility remains under construction as regulatory constraints tighten

Meta’s Cheyenne campus was still under construction at the time of the discharge, meaning the fill-and-flush procedure occurred during the commissioning phase rather than routine operations.

This timing is significant for understanding institutional implications: data center construction projects generate the highest-intensity water discharge events, concentrating environmental risk during periods when facilities are least operationally mature and contractors maintain primary control over industrial processes.

The exact source of Cupriavidus within Meta’s facility remains unknown, complicating future prevention strategies and suggesting the pathogen may have originated from construction materials, water sourcing, or handling practices rather than facility equipment.

The unresolved contamination source creates operational uncertainty for Meta and replicates a problem across the industry: construction contractors managing fill-and-flush procedures lack standardized protocols for pathogen detection and prevention, and municipal authorities have limited tools to require them prior to incidents occurring.

Institutional investors funding data center construction projects must now account for the possibility that standard commissioning procedures can trigger regulatory responses affecting entire regional markets, regardless of facility design quality or operator intentions.

Watch for whether other Wyoming municipalities adopt similar blanket fill-and-flush restrictions, and whether the Wyoming Department of Environmental Quality issues statewide guidance on industrial wastewater discharge from data center construction projects. Meta has stated it is “working with” Fortis to resolve the contamination issue, but has not disclosed whether the facility’s construction timeline will be delayed or whether the company will pursue alternative water sourcing for commissioning procedures. The Cheyenne Board of Public Utilities decision remains in effect, and no timeline has been announced for reconsidering the industry-wide ban or establishing compliance pathways for resumed fill-and-flush operations.

Get this in your inboxThe Crypto Coin Show newsletter covers the policy and market moves institutional crypto investors are pricing in.

Subscribe

Load More