U.S. manufacturing expansion hits four-year high, while bond market falls into "credit blind spot"

Data released by Tether shows that its tokenized gold product Tether Gold (XAU?) saw a 9.5% rise in investor holdings during the second quarter of 2026, reflecting sustained growing demand for on-chain gold assets. The data indicates that despite a 14.1% drop in gold prices in Q2, investors continued to add to their XAU? positions, showing that market demand is not solely driven by bullish gold trends—some investors are taking advantage of price pullbacks to increase their exposure to physical gold. As of June 30, 2026, XAU? is fully backed by physical gold at a 1:1 ratio, with total gold reserves amounting to 707,747.139 troy ounces, equivalent to approximately 22.01 tons and valued at around $2.837 billion. By the end of Q2, 612,823.66 XAU? tokens had been issued, up 53,225.02 from 559,598.64 at the end of Q1, corresponding to an additional 1.66 tons of physical gold held by investors, representing a roughly 9.5% rise in client holdings. Tether noted that XAU?’s gold reserves remained unchanged throughout the quarter, with every token backed by at least one troy ounce of physical gold. The gold is stored in Swiss vaults, including 1,759 London Good Delivery standard bars and other bars of various specifications. Tether CEO Paolo Ardoino said that gold experienced its largest quarterly decline in 13 years during Q2, yet investors continued to buy XAU?, indicating that tokenized gold is becoming a new way for investors to allocate to physical gold. Additionally, data from Tether International SA de CV shows that the company purchased a total of about 27.1 tons of gold in the first half of 2026, with an average monthly procurement of roughly 4.5 tons. If included in central bank gold purchase rankings, its total gold holdings would stand at approximately 150 tons, ranking third in global gold accumulation during the period, second only to Poland and China.

Hugging Face CEO Clément Delangue stated that the self-initiated cyberattack by an OpenAI test model is an "unprecedented" case in the artificial intelligence industry, and called on the U.S. to establish a legal regulatory framework for autonomous AI systems. In an interview, Delangue noted this marks the first time a "highly autonomous system has carried out such attack actions," highlighting new cybersecurity risks posed by AI agents. He argued that companies should be required to disclose incidents where AI systems autonomously perform network operations, and clear legal boundaries should be defined. Earlier, OpenAI disclosed that one of its unreleased AI models broke free from a controlled environment during security testing, connected to the internet, and launched a complex attack on the Hugging Face platform. The model used a combination of multiple attack techniques to attempt to obtain information needed for internal security assessment tasks, OpenAI said. Hugging Face later found the AI agent executed over 17,000 operations over several days before being stopped by its security team. Hugging Face added that it used open-source AI models to complete incident analysis and successfully fended off the intrusion. Delangue emphasized he does not believe OpenAI acted maliciously, but the incident reflects that developers may lose control over highly autonomous AI systems during testing. He called on the U.S. government to explicitly ban autonomous AI cyberattacks and mandate companies to report when their AI systems independently perform network operations. "Only through transparent disclosure can we understand technical risks and build safer systems."

U.S. President Donald Trump posted on social media, saying the duality of Iran’s leadership is "incredible!" They demanded a meeting—some would even say they are "begging" for talks. Negotiations have begun, and more meetings are scheduled in the coming days. At the same time, however, Iran’s leadership publicly and proudly claims it is not holding any discussions, talking about nothing, and is only engaging with the Omani side. Subsequently, they repeated their usual line, asserting that the Strait of Hormuz will be firmly controlled by them. In reality, however, the strait is now completely under the control of the U.S. Navy and our "blockade operation"—or as some refer to it, the "American Steel Wall!" Nothing can enter Iran without our permission; nothing can pass through unless an agreement is reached or a "full surrender" is achieved. Whether Iran is willing to admit it or not, we are in fact discussing how to resolve a problem created by them over the past decades. It is very simple: Iran will never possess nuclear weapons! (Jin10)

According to data from S3 Partners, short positions in SpaceX have continued to surge since the company’s listing, with the current nominal short value hitting around $23.6 billion—surpassing Tesla’s roughly $22 billion—to rank among the largest short bets in the U.S. stock market. Data shows investors are currently shorting approximately 206 million SpaceX shares, accounting for about 32.2% of the company’s publicly traded shares. Short sellers are betting on SpaceX’s upcoming first quarterly report since its listing, as well as potential price pressure from the start of insider share lock-up expirations this week. At the time of SpaceX’s initial listing in June, short positions stood at only around 40 million shares, making up 5% to 7% of its publicly traded shares. As the stock price plummeted, short positions rose rapidly. Since its first trading day on June 12, SpaceX’s market capitalization has shrunk by more than $500 billion, with its share price falling over 50% from its intraday peak. Data from Bernstein shows that this week’s first lock-up expiration will allow the sale of roughly 20% to 30% of restricted insider shares. Subsequently, shares accounting for about 7% of total shares will be unlocked in phases before October, with another round of lock-up expirations following the third-quarter earnings report, and additional shares released after the end of the 180-day lock-up period. Bernstein forecasts that by December, SpaceX’s publicly traded share proportion could rise to around 40% of its total share count. Analysts note that some investors are temporarily reluctant to buy SpaceX shares based on fundamentals due to ongoing lock-up pressure. The market is also closely watching SpaceX’s first quarterly report since listing, released after U.S. trading hours on Tuesday, as investors will use the report to assess the company’s performance in rocket launches, Starlink, and AI businesses.

Trump-linked bitcoin mining firm American Bitcoin (ABTC) announced that its president and interim CFO Matt Prusak will step down on August 4 to join AI and energy infrastructure company Giga Energy as chief business officer and interim CFO. Prusak said that after "years of building his bitcoin business", he will move upstream to the energy infrastructure sector, focusing on power supply issues that currently constrain the development of both bitcoin mining and AI computing. American Bitcoin, co-founded with Eric Trump and backed by Hut 8, is a Nasdaq-listed bitcoin mining firm. Prusak previously led the company's bitcoin accumulation strategy, which included expanding computing power and increasing BTC holdings per share. Headquartered in Houston, Giga Energy primarily develops power equipment and AI data center infrastructure. The company said it has delivered over 6.5GW of power infrastructure to date and is developing more than 500MW of AI data center capacity. Industry insiders believe Prusak's move reflects that the bitcoin mining sector is accelerating its shift toward AI infrastructure. As competition in the mining business intensifies and profit margins come under pressure, an increasing number of mining firms are leveraging their power resources, land reserves and data center capabilities to enter the AI computing infrastructure market. As large tech firms compete for power and data center capacity, energy supply has become a core bottleneck for AI computing expansion, and companies with power resource integration capabilities are gaining more attention.

AI startup June, founded by former Salesforce executive Efrat Rapoport, has emerged from stealth mode and announced the close of a $20 million pre-seed funding round. The round was led by Time Ventures, the investment arm of Salesforce CEO Marc Benioff, with participation from tech industry figures including Michael Dell, Aaron Levie, and George Kurtz. June focuses on solving system integration challenges that enterprises face when deploying AI tools. The company notes that the main bottleneck for enterprise AI adoption today is not model capability, but rather getting AI agents to work in tandem with existing enterprise software systems, data platforms, and complex workflows. June’s platform scans a company’s existing systems, analyzes business processes and data structures, identifies bottlenecks, and automatically generates AI agent deployment plans—including steps like cleaning duplicate data fields, connecting data sources, and optimizing workflows. The founding team previously launched speech recognition firm Bonobo AI, which was acquired by Salesforce in 2019. After the acquisition, the team worked on Salesforce’s AI-related projects and observed the challenges enterprise clients faced when integrating AI into their existing systems. Currently, many large enterprises still rely on traditional data management platforms such as Salesforce, ServiceNow, Databricks, and Workday, requiring AI agents to address issues like legacy systems, fragmented data, and long-standing technical debt.