The Wild West of Crypto Assets and the AI Infrastructure Frenzy: Last Week's Market Summary
The reason for the simultaneous sharp drop in cryptocurrency at the start of the week was a chain reaction of forced liquidations in the absence of news, with the highly leveraged market structure reinforcing a "Wild West" appearance. Meanwhile, the main player in the stock market is AI infrastructure investment. Massive contracts and capital circulation are occurring one after another, spreading to power, semiconductors, networks, and electrical components. This article covers major players extensively and organizes "what is happening / where the opportunities and risks lie" using facts and figures.
1. Crypto Market: Leverage and "Off-the-Books" Risks
1-1. Impact of Overseas Exchanges
Bitcoin and Ether are seeing a notable backdrop of high leverage in overseas derivatives markets. Binance offers up to 125x leverage on futures, creating a structure where forced liquidations can easily chain together in a short period (US-based firms have more restrained leverage due to regulatory compliance).
Binance's corporate registration is confirmed in the Cayman Islands, and its massive global trading volume also amplifies market impact.
1-2. The Essence of Stablecoins
"Practical use in payments" is handled by low-volatility stablecoins, but the largest player, Tether (USDT), is registered in the BVI and based in Hong Kong, and debates regarding the transparency of its reserves continue. Structural risks should always be monitored.
2. AI Infrastructure: Mega-Contracts and "Circulating Capital"
2-1. CoreWeave × OpenAI / NVIDIA
CoreWeave has signed its third contract of the year with OpenAI, worth up to $6.5 billion, bringing the annual total to $22.4 billion. It also signed a $6.3 billion capacity reservation agreement (with a buy-out clause for unused capacity) with NVIDIA. While this "circular" relationship, which includes mutual investment and mutual ordering, is a growth driver, it requires an assessment of its soundness.
2-2. Oracle × OpenAI: Unprecedented Scale
Oracle reported a Remaining Performance Obligation (RPO) of $455 billion this term, up 359% from the previous year. Reports suggest OpenAI contracted for $300 billion in compute over about five years, which is seen as the primary driver of the RPO surge. The scale supports growth expectations, but if the financing environment deteriorates, fulfillment risks could materialize.
2-3. Demand Indicated by "Token Explosion"
Microsoft explained that it processed 100 trillion tokens in the quarter (a 5x increase year-over-year). The surge in AI workloads is demanding the total mobilization of new data centers, GPUs, networks, and power.
3. Semiconductors, Networks, and Components: The Map of Winners
NVIDIA (GPUs), Broadcom (custom chips/networks), Arista Networks (AI networking), Amphenol (high-speed connectors/cabling), Eaton (power infrastructure), and Vertiv (high-density cooling) are benefiting from the "necessities" of AI data centers. Arista saw Q2 2025 revenue up 30%, and Amphenol accelerated with Q2 2025 revenue up 57%. Eaton/Vertiv are expanding their product lines for AI power burst detection and high-density cooling.
4. Energy: The "Bottleneck" of Power and a Two-Front Strategy
Power demand for AI data centers is a long-term constraint. GE Vernova and Siemens Energy are the two giants of gas turbines, and CEO Ketchum of U.S.-based NextEra suggested that the construction cost of gas-fired power plants has more than tripled from $785/kW in 2022 to as much as $2,400/kW. A 'realistic mix' of renewables, gas, and nuclear power is becoming essential.
At the same time, Quanta Services is increasing its presence in the construction and maintenance of transmission, distribution, and substations, while Vertiv serves as a recipient of capital investment for cooling and power, and Eaton does so for switchgear and similar equipment.
5. Mega-platforms and Software: Redistribution of Hegemony
Hyperscalers (Microsoft, Google, Amazon, Meta, Oracle) are increasing their AI capacity by using a combination of massive internal funds and external financing. Given the history where 'dull but solid' payment infrastructures like Visa/Mastercard have maintained their top positions, companies that control data and customer touchpoints are likely to be the long-term winners. In software, it is a stage where Palantir and MongoDB are leveraging data foundations, and existing giants (Salesforce/Adobe, etc.) are rushing to revamp their features to dispel concerns that their moats are thinning due to AI (individual stocks will see mixed results based on earnings and product progress).
Summary
The 'Wild West'-like volatility of crypto and the unprecedented mobilization of capital for AI infrastructure are happening simultaneously. While examining the breadth of the industry from semiconductors → networks → electrical construction → power across various stocks, constantly checking the three risk axes—leverage structure, concentration of counterparties in ultra-large contracts, and power bottlenecks—is the shortest path to identifying the 'companies that will survive' in the next wave.
