This morning, all three major U.S. stock indexes closed lower, with Marvell falling more than 10% and NVIDIA down 4.57%.

Meta is testing robots to maintain the data centers that support its artificial intelligence (AI) systems, with applications including replacing network cables, rebooting servers, moving racks, and conducting equipment inspections. The robots currently being tested by Meta come from Watney Robotics, Kinova, and ABB. These robots still face challenges such as slow operation, limited battery life, difficulty with visual detection, and struggles navigating dense cables and complex obstacles, requiring human supervision and assistance. However, one Meta employee estimated that if cable-replacement robots become fully mature, up to 80% of tasks in some roles could be automated by machines. This figure is the employee’s personal estimate, not Meta’s official forecast. As AI infrastructure continues to expand, Meta is confronting two key issues: a shortage of data center technicians, and efforts to use robots to reduce operational and maintenance costs. Meta noted that the U.S. is experiencing the largest infrastructure construction boom since World War II, and the company is facing a severe shortage of skilled workers, hence “more workers are needed, not fewer.” But employees worry that robots could reduce demand for experienced technicians, shifting remaining work to lower-paid staff who would carry out tasks based on AI-generated instructions. Currently, Meta’s robots cannot independently complete complex repairs, but their further development may redefine whether the expansion of AI data centers will create more jobs or accelerate automation of technical positions.

After Fed Chair Walsh delivered a hawkish signal at the Jackson Hole Annual Meeting, Bitcoin rallied then pulled back, falling as low as $76,877 on Friday—sharply retreating from its overnight high of $81,455—and closing at $77,557, down 3.39% on the day. Earlier this week, Bitcoin had notched double-digit gains, but the $81,000–$82,500 resistance zone capped its upside again. Walsh noted that U.S. inflation is falling too slowly, and the Federal Reserve "has more work to do" before hitting its 2% inflation target, prompting markets to sharply raise bets on a September rate hike. CME FedWatch data shows the probability of a September rate hike rose to 55.7% from 35.4% the prior day. The hawkish shock also triggered mass liquidations of leveraged positions in the crypto market, with roughly $481 million in total liquidations over the past 24 hours, including more than $360 million in long liquidations. However, the market’s long-term bullish sentiment has not reversed significantly. Prediction market data shows traders currently assign a 77% probability that Bitcoin will hit its next major target of $84,000, and a 23% chance it will fall to $55,000; Friday’s pullback has not altered these odds for now. On the fundamental front, U.S. spot Bitcoin ETFs have posted net inflows for 8 consecutive trading days as of Wednesday, drawing roughly $2.8 billion in total—marking the longest such inflow streak since April. Technically, Bitcoin’s RSI stands at around 69.7, not yet in the extreme overbought territory that triggered prior pullbacks. Should it fall further, the $73,670–$75,157 zone will act as a key defense level for bulls, while reclaiming the $81,000–$82,500 range is critical to opening up new highs. In the short term, Walsh’s downplaying of forward guidance means markets lack a clear policy path ahead of the next FOMC meeting, and Bitcoin is likely to remain highly volatile amid inflation data and shifting rate expectations.

Bullish announced a $100 million stablecoin debt financing facility for GPU-focused lending protocol USD.AI to support its loan business targeting high-performance computing assets. Bullish noted that AI infrastructure financing has emerged as a key segment of the private credit market, and this funding will help USD.AI expand its financing scale for AI computing power assets. Additionally, Bullish will facilitate the listing of multiple trading pairs for sUSDai, the token issued by USD.AI, on its own exchanges alongside a corresponding market-making program, to boost secondary market liquidity and price discovery for GPU debt assets. David Choi, CEO of Permian Labs (USD.AI’s developer), stated that the $100 million financing and institutional-grade market infrastructure will enable USD.AI to further provide financing for AI infrastructure development and advance the growth of the computing power-collateralized credit market.

Beating AI News Flash: Alternative asset management firm Blue Owl Capital announced that a fund it manages has led a $2.4 billion computing power equipment financing round for IREN. The funds will be used to support IREN’s procurement of NVIDIA AI computing infrastructure, including Blackwell Ultra GPUs, for its Mackenzie data center campus in British Columbia, Canada. The financing comprises $1.2 billion in senior secured term loans and $1.2 billion in senior secured notes, with disbursements to be made in installments aligned with hardware delivery and deployment progress. IREN currently holds over 5GW of global data center development reserves, and the latest funding will further expand its AI Cloud infrastructure to meet demands for AI training and inference. Nico Caprez, NVIDIA’s Vice President of Global AI Infrastructure Growth, noted that CUDA enables NVIDIA’s AI infrastructure to serve diverse customers and workloads while continuously increasing in value over time, making it an investable infrastructure asset class. Blue Owl added that this transaction further validates the model of conducting large-scale AI infrastructure financing backed by GPU devices.

Qualcomm plans to launch a long-term robotics investment initiative in Japan, centered around its newly established Qualcomm Japan Robotics Center, to partner with Japanese robotics, industrial automation, research, and tech firms in building an open physical AI ecosystem. Qualcomm stated that the initiative aims to drive technological innovation and industrial upgrading in Japan’s robotics, physical AI, and industrial automation sectors, while aligning with Japan’s policy priorities of addressing labor shortages, enhancing industrial competitiveness, and expanding AI and robotics applications. The Qualcomm Japan Robotics Center will handle functions including application R&D, industrial ecosystem collaboration, talent cultivation, and robotics innovation. Participating entities include Japanese companies and research institutions such as Fanuc, Kawasaki Heavy Industries, Toyota, Sony, Panasonic, Fujitsu, NEC, NTT DOCOMO, KDDI, and Preferred Networks. The investment initiative covers the full lifecycle of robotics, from technical R&D to commercial deployment and global expansion, with four core pillars: establishing a robotics R&D center, expanding the industrial ecosystem, driving global commercialization for Japanese robotics firms, and supporting the launch of demonstrative physical AI projects. Qualcomm said it aims to help Japanese robotics startups, developers, and research institutions accelerate innovation by providing chip platforms, technical support, development resources, and a global network of customers and partners, while advancing Japanese robotics solutions from the domestic market to global markets.

SVM Layer 1 network Fogo announced that the Fogo Foundation was recently hacked by unknown attackers, resulting in approximately 400 million FOGO tokens being transferred to the attacker’s address. The foundation stated it immediately notified relevant trading platforms and is actively coordinating with law enforcement agencies and forensics experts. To date, the incident has not impacted the Fogo blockchain itself, and the network continues to operate normally. The foundation added that it will disclose additional information in a timely manner.