TechFlow Selected published a new edition of “TechFlow Intelligence” at 11:45:32 on June 18, 2026. The column said its AI Agent had completed a daily patrol across more than 200 global information sources, spanning crypto, AI and technology, and filtered out 99% of the noise to keep the signals readers need. This edition was led by two headline items: Solana-based on-chain stock trading reached a record level, and OpenAI’s losses were reported at as much as $38.5 billion.
AI and large models: OpenAI losses, Anthropic’s public comments and a Copilot security issue
The AI and large model section opened with Anthropic CEO Dario Amodei. In a recent interview, Amodei spoke candidly for the first time about why he left OpenAI, saying he “could not trust a person whose pattern of behavior was disturbing and dishonest.” TechFlow described the comment as a rare public expression of internal conflict within Silicon Valley’s AI circles. The roundup also noted that community discussion viewed the remarks as validation of rumors about governance disorder inside OpenAI, with Reddit and a video listed as the sources.
OpenAI’s financial figures formed another central item. According to the roundup, OpenAI generated $13 billion in revenue in 2025, but its loss reached $38.5 billion. TechFlow wrote that revenue had surged, yet uncontrolled costs produced a huge deficit, leaving the company’s commercialization path under heavy pressure. The column’s own sharp comment said OpenAI had lost the equivalent of “three and a half Twitter acquisition prices” in one year, while Sam Altman’s idea that “AGI needs to burn money” was facing a harsh pressure test.
Microsoft Copilot was also included because of the “SearchLeak” vulnerability. Security researchers warned that the vulnerability could be used for data theft and could expose user emails and two-factor authentication codes. Microsoft had not yet issued a formal response, according to the item. DeepSeek appeared in the same section: the United States listed more than 100 companies as security risks, but DeepSeek had not been included. At the same time, the chain-of-thought behavior of DeepSeek’s web “fast mode” changed noticeably, leading to discussion on DeepSeek Chat and Zhihu. Midjourney, meanwhile, launched a Medical vertical application and entered the medical image generation field. TechFlow noted that compliance and accuracy remained controversial, while Hacker News had 523 comments questioning the ethical boundaries of medical AI generation.
Crypto and Web3: Solana dominates tokenized stock volume
The most important signal in the crypto and Web3 section was tokenized stock trading. Citing Twitter/Buffalu and Twitter/Grayscale, TechFlow reported that tokenized stocks set a new record for the largest single-day trading volume, reaching $142 million. Of that volume, 97% took place on Solana. SpaceX’s $SPCX recorded $3.1 billion in cumulative trading volume across the nine days before and after its IPO, and Hyperliquid was described as becoming a new venue for pre-IPO price discovery.
Exchange activity appeared alongside the on-chain stock data. OKX announced that the RE token would be listed for spot trading, and that its pre-market futures would be converted into a standard perpetual contract. In South Korea, Upbit simultaneously designated AERGO as a trading warning item. TechFlow grouped these items under the same Web3 flow, showing tokenized asset trading, centralized exchange listings and exchange-level warnings arriving on the same day.
Chips, hardware and technology companies: AI-related costs reach products and security layers
The chips and hardware section included AMD, SK Hynix and China’s display industry. AMD was reported to have quietly removed memory encryption functionality from consumer-grade Ryzen CPUs in a new AGESA firmware release. TechFlow said the security feature had been silently deleted, leaving users unaware and exposed to vulnerability risks, while AMD engineers remained silent. SK Hynix had begun supplying samples of its next-generation AI memory, HBM4E with 12-layer stacking. The roundup also said the AI chip arms race was moving into the memory layer, while tight TSMC capacity was driving a surge in demand for Samsung foundry services. Industry discussion, as summarized by TechFlow, held that DRAM supply-chain tightness could continue until 2027. In China, the country’s first 8.6-generation AMOLED production line began mass production in Chengdu with a total investment of 63 billion yuan. Cailian Press described it as one of the world’s first production lines of that generation and said it marked China’s panel industry entering the high-generation OLED era.
Technology company news continued with Apple and Xiaomi. Apple CEO Tim Cook said RAM costs were “unsustainable” and that iPhone prices would rise. TechFlow wrote that AI features were increasing memory demand, and that Apple had publicly stated it would pass costs on to consumers, a rare early price-rise warning in recent years. Xiaomi was linked to a criminal case involving AI-generated rumors: Xi’an police solved a case involving the use of AI to generate false information and arrested four people. TechFlow described it as China’s first publicly announced criminal case involving AI rumor fabrication, signaling that regulation had begun to target misuse of generative AI. The related Zhihu discussion had 580,000 views, with users debating the legal boundaries of AI-generated rumors. Ars Technica also reported a large-scale data leak exposing thousands of sensitive network credentials, with the affected scope involving multiple critical infrastructure networks.
U.S. equities, geopolitics and new product signals
In the U.S. equities and finance section, NASA selected former Google CEO Eric Schmidt’s rocket company for a Mars mission, creating competition with SpaceX. TechCrunch reported that Schmidt’s aerospace company had won a NASA order, marking the first direct challenge to SpaceX’s monopoly position in that area. At the same time, Sequoia partner Roelof Botha joined SpaceX’s board. Another major geopolitical item centered on Trump’s defense of a U.S.-Iran agreement. He said he did “not want to see an economic disaster” and did not want to be “Hoover, who caused the 1929 crash.” The roundup said the United States and Iran had reached a memorandum of understanding: Iran agreed to dilute its stockpile of enriched uranium in exchange for sanctions relief, and the first batch of Iranian oil had broken through the U.S. blockade and moved to market. Republican Senator Bill Cassidy criticized the arrangement as “the worst diplomatic mistake in decades.” TechFlow also recorded market concern that the situation in the Strait of Hormuz could change the global energy landscape.
Geopolitical risk also included Ukraine and Germany. Ukraine launched a second airstrike on a Moscow refinery within one week, while Germany sent warships to the Red Sea. TechFlow said the crude oil supply chain again came under pressure. The German defense minister said two warships had been dispatched in response to military operations related to the Strait of Hormuz. The column’s own comment said Trump did not want to be Hoover, but the market had already started pricing him; the White House narrative had moved from “Make America Great Again” to “do not let the economy crash again,” a downgrade speed compared in the roundup to DeepSeek’s chain of thought.
The new products and trends section began with Lore, an open-source version control system designed for scalability. On Hacker News, Lore received 1,150 votes and 611 comments and was described as a potential challenger to Git. In China’s digital asset rights discussion, a mother sued to inherit 87 game accounts belonging to her deceased son and received court support, while the platform’s standard-form terms were ruled invalid. TechFlow called it China’s first publicly reported successful inheritance case involving virtual property. The related Zhihu topic reached 8.64 million views, with users discussing rights around digital inheritance. The full 1.0 version of “The Scroll of Taiwu” was also released after eight years of development. The domestic independent game’s official release drew 1.28 million views on Zhihu, where users discussed its completion level.
TechFlow’s “hidden thread of the day” connected several of these items: Apple raising prices because of AI, OpenAI burning cash and posting losses, the Copilot vulnerability, and AMD removing a security feature all showed the cost of the AI boom moving from capital markets to consumers and security baselines. At the same time, record on-chain stock trading and Trump’s compromise with Iran to avoid an economic crash were presented as two parallel signals about who ultimately pays the bill when AI infrastructure costs run high and geopolitical risk rises. The original post ended by inviting readers to join the official TechFlow community, including its Telegram subscription group, the Twitter account TechFlowPost and the English Twitter account BlockFlow_News.