Aston在下不求

Strategy CEO Phong Le：Bitcoin Is American-Style Money

Strategy CEO Phong Le (the company’s former name is MicroStrategy, and Michael Saylor is the chairman) published a long post titled “Bitcoin is Freedom” on X on July 4, 2026. He begins with his own background as a Vietnam War refugee: in 1978, his family fled Vietnam, arrived in the United States with the sponsorship of a Catholic church in New York State (Syracuse), and when he was young he relied on public assistance and food stamps, while his father worked three jobs. The post was timed for the Fourth of July in the United States—also the 50th anniversary of the U.S. founding year 1976 in which Phong Le was born. This is a rare, in-depth philosophical discussion from Strategy’s leadership on Bitcoin in recent times.

From Vietnam Refugee to Strategy CEO: Experiencing a Society Where “Service Can Compound”
In the article, Phong Le details the specific circumstances of his family’s move to the U.S. A Catholic church in Syracuse sponsored the family’s relocation; they survived on public assistance, food stamps, and free nutritional lunch programs, while his father worked three jobs at the same time. He describes that America is not easy—“facing poverty and racial discrimination”—but “open,” offering a system in which effort, education, entrepreneurship, and perseverance can grow in a compounding way. His personal trajectory is proof of that system: from refugee to Strategy CEO, with a wife and three children. He defines this as the “American Dream”: not just wealth, but “freedom, family, opportunity, and the ability to decide your own future.”

He also mentions his family’s observations after returning home—Vietnam is flourishing today, but young people’s ultimate dream still remains the United States: American brands, American food, American music, and American-style confidence.

Core thesis: “Bitcoin is the American in the realm of money”
The key passage of the article is: “Bitcoin is the United States of money.” Phong Le compares the U.S. Constitution with Bitcoin’s white paper side by side. He believes the core of the U.S. Constitution is “governing the country with transparent rules rather than governing with individual discretion”; the core of the Bitcoin white paper is “governing money with code, energy, and consensus”—and that both share the same underlying framework of principles. He further defines Bitcoin as “digital capital governed by code, energy, and consensus,” and lists three groups that Bitcoin offers “hope” to: (1) those who work hard and want to protect their property from being eroded by currency inflation; (2) those born in countries lacking reliable rule of law or economic freedom; and (3) anyone seeking asset protection, opportunities, and forms of wealth that are “not dependent on geography, politics, or permission.”

Strategic context: Rare public philosophical remarks from Strategy’s CEO
The strategic significance of Phong Le’s long post lies in the fact that it is written by the CEO of Strategy rather than by chairman Michael Saylor. In the past, Bitcoin-related narratives from Strategy were mainly shaped by Saylor in public appearances and on X; as CEO, Phong Le has focused more on operational and financial aspects. This time, publishing a philosophical discussion on Independence Day in the U.S., directly connecting his personal immigration background to Bitcoin ideology, is Strategy’s first public attempt to define a BTC narrative through first-person experiential account.

Le was born in 1976 (the U.S. 200th anniversary). The post was published a year before the 250th anniversary (2026)—and he explicitly draws the comparison: “Fifty years later, the future of American hegemony is being questioned; I’m not questioning it.” The article’s conclusion defines the relationship between two kinds of freedom: “America grants freedom to my family through a country; Bitcoin grants individuals financial freedom through a network. One opens a door for me; the other opens a door for anyone, anywhere.”

Earlier reporting: microstrategy’s mNAV fell below 1 amid an arbitrage strategy; other reporting: Bitwise believes STRC’s selling pressure is near the cycle bottom rather than a point of Strategy’s collapse. Phong Le’s philosophical piece is the second track of the Strategy narrative—its value foundation beyond operations and finance.

The earliest appearance of this article: Strategy CEO Phong Le: Bitcoin Is American-Style Money

Anthropic Engineer: How Fable 5 Uses Clarifying “Undefined Items” to Work Well

Anthropic Claude Code engineer Thariq Shihipar (X account @trq212) published a long post, “A Field Guide to Fable: Finding Your Unknowns,” on July 3. He observed that Fable 5 repeatedly taught him one lesson: the map is not the territory—there’s a gap between the instructions you give to Claude and the limitations you encounter during real implementation. Thariq argues that success in agentic coding hinges on identifying and clarifying unknowns before and during implementation. He proposes five concrete techniques: blindspot pass, brainstorming, interviews, references, and implementation planning. This is the second set of Fable 5 principles shared publicly by Anthropic engineers, following the Loop Engineering concept Andrew Ng certified on 6/30.

Key insight: the quality bottleneck isn’t the model’s ability—it’s the unknowns in your prompts. Thariq’s core observation is: “the quality of the work is bottlenecked by my ability to clarify its unknowns.” In practice, this means: whatever kind of solution Fable 5 gives you directly reflects how clearly you wrote in your prompt. If your prompt contains assumptions or undefined boundaries you didn’t even notice, Fable 5 will handle them with a reasonable approach that may differ from what you expected—this is the concrete manifestation of map ≠ territory. That’s also why developers often feel that Fable 5 “doesn’t understand requirements”—the real issue is often that the requirements themselves contain unknowns.

The 4-quadrant unknowns framework: Rumsfeld-style classification. Thariq borrows a knowledge classification framework used by politician Donald Rumsfeld and breaks the information gaps in prompts into four quadrants: 1) Known Knowns—explicit instructions that both you and Fable 5 clearly understand; 2) Known Unknowns—gaps you’ve identified but haven’t provided answers for yet; 3) Unknown Knowns—intuitive standards you understand but haven’t articulated (taste, aesthetics, industry conventions); 4) Unknown Unknowns—factors you never even thought would affect the outcome. The most common quality bottlenecks of Fable 5 occur in the 3rd and 4th quadrants—since you don’t even know those assumptions exist, Fable 5 has no way to follow them.

The core methodology of the field guide is to design a workflow that gradually moves unknowns from the 3rd and 4th quadrants into the 1st and 2nd quadrants.

3-stage implementation techniques: pre / during / post. Corresponding to those four quadrants, Thariq proposes implementation techniques across three stages:

Pre-implementation: Blindspot pass lets Fable 5 scan your prompt and mark parts that may be ambiguous or undefined. Brainstorms & prototypes—before the official prompt—encourage Fable 5 to think about the task in multiple ways through a divergent mode. Interviews & references prompt Fable 5 to ask you things like “If X happens, how do you want to handle it?” until the unknowns are filled in, or to anchor on concrete existing implementations.

During implementation: Implementation notes—before each step of hands-on coding—have Fable 5 write down its assumptions for that step, so you can intercept unknowns in real time. This is the key action for gradually moving the 3rd and 4th quadrant unknowns (Unknown Knowns / Unknown Unknowns) into the 1st and 2nd quadrants.

Post-implementation: Quizzes & pitches are the check methods Thariq especially emphasizes. After implementation, have Fable 5 take a short “quiz” on the results, or explain design decisions in a way similar to “pitching to stakeholders.” This often reveals the implicit assumptions it made during implementation.

Case: Thariq uses Claude to edit the Fable launch video. In the field guide, Thariq shares an example from his own implementation: he used Claude to edit the Fable 5 launch video, during which he gradually discovered multiple unknowns—such as video transcription (subtitle transcription), video timing (the sequence of segments), color grading (color correction), and more. These were factors he hadn’t initially considered as influencing the final outcome (Unknown Unknowns). Through Blindspot pass and Interviews, he gradually exposed them, and then moved them into clearly defined instruction scope.

The significance of this case is that the unknowns framework applies not only to code, but to any agentic task—including creative output, ...

American Jewish man accused in Israel of acting as an Iranian spy: only received $1,400 in encrypted compensation

Israel’s prosecution filed espionage charges on July 3 against Eli Lavon, a 21-year-old man with dual U.S.-Israeli nationality. Reporting by The Times of Israel says that while Lavon attended the Mir yeshiva in Jerusalem, in November 2025, he was “absorbed” by an Iranian intelligence unit after receiving messages on Telegram that were disguised as job postings. He then carried out photographing sensitive locations in Jerusalem and sent the intelligence back to handlers in Iran. The entire payment totaled about $1,400, paid in cryptocurrency. Israel arrested him on June 9, 2026, and he was formally indicted on July 3.

This is the first Iranian intelligence case in which a U.S. citizen has been charged by Israel since 2023.

Allegations: 2 counts of contact + 14 counts of information transfer
The prosecution’s specific allegations against Lavon consist of two layers: (1) “contacting foreign intelligence personnel,” 2 counts; and (2) “transmitting information deemed useful to a hostile state,” 14 counts.

The task was to photograph specific sensitive locations within Jerusalem, including key government buildings, military-related facilities, and transportation hubs. All intelligence was sent via Telegram to an overseas contact believed to be a handler for an Iranian intelligence unit. The entire absorption and execution process took about six months (from November 2025 to his arrest in June 2026), with total compensation of about $1,400.

Recruitment method: job ads disguised on Telegram
The absorption method described by Israel’s prosecution matches other recent Iranian intelligence cases: Iranian intelligence units widely broadcast messages on Telegram disguised as job openings, targeting religious communities inside Israel, students, or people with temporary residency. The job ads typically take the form of “photography and data collection work,” but the real tasks gradually escalate into intelligence gathering.

In recent years, Israel has reported a significant increase in Iranian-related espionage activity, and multiple cases have used a similar operational model of Telegram recruitment plus cryptocurrency payment.

Cryptocurrency payments as a low-cost espionage tool: what $1,400 means
The compensation amount in the Lavon case ($1,400) itself carries policy implications. It shows that the Iranian intelligence unit was able to complete cross-border payments at extremely low cost (lower than the spending on typical American intelligence informants) using cryptocurrency, thereby avoiding traditional bank countermeasures and anti–money laundering review. Israeli national security officials have said in recent years that cryptocurrency has become one of the main payment channels for Iranian intelligence activities. Through decentralized networks and anonymous stablecoins, Iran can conduct large-scale recruitment and operations at very low cost—capable of absorbing failures of a single case.

This is the first instance of a U.S. citizen being indicted by Israel using this approach, and it is a concrete example of cryptocurrency payments in cross-border intelligence activity.

This article, “American Jewish man accused in Israel of acting as an Iranian spy: only received $1,400 in encrypted compensation,” appeared first on .

Blackstone sells $3.5 billion worth of AI data center assets, exits plans for the world’s largest campus development—what does this mean?

Blackstone’s real estate investment trust QTS Realty Trust recently sold its equity in three data centers in Northern Virginia for $3.5 billion, and then immediately pulled out of the planned development of what would become the world’s largest data center campus. As a major data center developer with related assets totaling more than $150 billion, the company’s back-to-back retreat signals have led the market to worry that the AI infrastructure boom may be cooling. (Where will the next wave of AI infrastructure go? Citrini report highlights “SiC, GaN and power facilities” as new investment directions.) Blackstone exits the AI data center market one after another: sells assets, abandons development

According to a report by Bloomberg, Blackstone is moving out of the U.S. data center market. The company recently first sold its equity interests in three data centers in Northern Virginia to Digital Realty Trust for $3.5 billion. The deal structure includes $1.2 billion in cash plus $2.3 billion in Digital Realty stock.

The three facilities are located in Manassas—two 96 MW data centers (Blackstone holds 80% equity)—and in Sterling—one 96 MW data center (Blackstone holds 50% equity). All were acquired by Blackstone less than three years ago.

However, just two days later, QTS Realty Trust, Blackstone’s unit, announced it would exit the “Prince William Digital Gateway” development plan. The proposed campus covers about 2,100 acres (roughly twice the size of New York’s Central Park). It is expected to include up to 37 data center buildings, with a total floor area of 22 million square feet. Originally, it would have become the world’s largest data center campus, potentially driving more than $100 billion in regional investment.

The final straw that broke the camel’s back: legal disputes and partner exits

Reports indicate the trigger for QTS’s exit was a string of legal setbacks and the loss of partners. The site is adjacent to a historical battlefield from the American Civil War, and it is also land that had been protected due to development restrictions since the beginning of the planning stage, when local residents mounted strong opposition.

In 2023, Prince William County held a hearing that lasted 27 hours. Hundreds of supporters and opponents attended to present their views. Ultimately, the county government approved changing the land from agricultural and semi-rural use to data center use by a narrow margin.

But community organizations immediately filed a lawsuit. Under Virginia state and local regulations at the time, the two newspaper notices of the hearing were required to be spaced by at least six days; however, the actual notice did not meet that threshold. In March 2026, a Virginia court upheld the earlier ruling, determining that the zoning approval was invalid due to procedural defects.

Compass Datacenters, a unit under Brookfield, was the first to exit in May, pulling out of its responsibility for the more than 800 acres of the development area—leaving QTS as the only remaining developer still appealing. After losing the partner that had been sharing the cost of upgrading public infrastructure, QTS leadership ultimately concluded that continuing the legal fight no longer made economic sense.

Seventy percent of Americans oppose; data centers become “Not-in-my-backyard” facilities

Blackstone’s retreat is not an isolated case and also reflects the political resistance faced by AI data center projects in the U.S. A Gallup poll conducted in March this year showed that 70% of Americans oppose building AI data centers in their residential areas, with 48% saying they are “strongly opposed.” Only about a quarter of respondents said they support the projects, and “strongly support” accounted for just 7%.

The opposition focuses on three main areas: resource consumption (water and power account for 18% each), environmental pollution (including noise, at 16%), and concerns about quality of life and economic burden (about 20%), including worries about rising electricity bills, higher living costs, and public funds subsidizing developers.

At the same time, Goldman Sachs data show that in Virginia, commercial electricity demand (with the 2010–2016 average set as 100) has surged to about 160 since 2020—far above other parts of the U.S. (around 105). On the policy front, Virginia has recently added a tax on data center energy consumption to its budget, and other states are also considering development bans. The issue of data centers is quickly escalating into a flashpoint of local political struggle ahead of the U.S. midterm elections.

(Jim Cramer: It’s not too late for AI data center stocks—the list covers four categories from chips to power.)

The potential chain reaction of Blackstone’s move on AI infrastructure investment

The collapse of the Prince William Digital Gateway is not only a failed local development project; it may also set up a replicable model of resistance for data center construction across the U.S. The strategy used by community organizations—attacking procedural defects through legal channels and pressuring locally elected officials—has already been proven to force even the most well-funded developers to back down.

From selling the Virginia data center assets to abandoning the world’s largest development plan, Blackstone’s consecutive actions show that even the most aggressive data center players—indeed, one of the world’s largest asset managers—are beginning to reassess the development risks of AI infrastructure and the uncertainty of timelines.

Today, power supply bottlenecks, rising resident opposition, and increasingly strict regulations indicate that the timeline for data centers—from planning to going online—will continue to stretch. This will directly compress investment returns for developers who rely mainly on leveraged financing, and it may also force tech giants to rethink the geography and pacing of their capital expenditures.

This article, “Blackstone sells $3.5 billion worth of AI data center assets and exits the world’s largest campus development plan—what does this mean?” first appeared on .

JPMorgan Warns: AI Chip and Tech Giants’ Stock Prices Decouple; Semiconductor Rally May Be Hard to Sustain

In a recent report, JPMorgan (JPMorgan) issued a warning that the ongoing trend in which the semiconductor sector has outperformed tech giants since last September may face a reversal. If the monetization capabilities of downstream customers’ AI do not improve significantly, the market could be hit by a tailwind reversal as capital expenditures slow down. AI chip stocks and tech giants’ performance decouple—JPMorgan flags potential risks
JPMorgan analyst Nikolaos Panigirtzoglou noted in the report that since last September, semiconductor stocks covering AI chip and memory makers have almost continuously outperformed tech giants such as Amazon, Microsoft, and Google. The analyst believes that this extreme decoupling phenomenon is not reasonable in the long run and has become a potential weakness in the semiconductor industry.
Two scenarios for narrowing the stock-price gap: AI monetization capability and corporate capital expenditures
To address how the future divergence in stock performance may converge, the report lays out two potential scenarios:
Optimistic scenario:
The AI profit models of tech giants, AI model providers, and end users improve significantly. As revenue and earnings rise, they will be able to catch up and gain a larger share in the overall value-add landscape of the AI industry.
Pessimistic scenario:
The semiconductor sector’s outperformance is built on sacrificing the profits of downstream customers like tech giants. This may lead customers to start reducing their willingness to invest in future capital expenditures, creating a strong headwind for demand for semiconductor products.
Could cloud capital expenditures slow in 2027? Nvidia and Micron face valuation headwinds
At the heart of the debate is whether massive AI hardware spending—such as GPUs from Nvidia (NVDA) and memory from Micron (Micron, MU)—can translate into sustainable revenue for tech giants. Although JPMorgan’s official baseline stance leans toward optimistic development and expects AI to eventually boost cloud revenue, the report also highlights near-term risks.
Currently, most bottom-up industry analysts’ consensus forecasts that tech giants’ capital expenditures will sharply slow starting in 2027. If that prediction comes true, it would mean the market has already priced in strong semiconductor demand. Once signs appear that tech giants’ budgets are loosening, it could trigger a sharp price correction. “Along with market concerns about the cyclical nature of semiconductor demand and the risk of excess capacity, chip suppliers may face pressure from inventory overhang and price-cutting competition.”
Calling out MicroStrategy: Adds major risk to the crypto market
Beyond the AI and semiconductor industries, JPMorgan also estimates that the U.S. scale of money creation will grow steadily from $1.6 trillion in 2025 to $1.8 trillion in 2026. The report specifically singled out MicroStrategy (Strategy, MSTR), warning that its financial maneuvers could add more unknowns and extreme volatility to the crypto market.
This article
JPMorgan Warns: AI Chip and Tech Giants’ Stock Prices Decouple; Semiconductor Rally May Be Hard to Sustain
First appeared on: .

Empery Digital Dumps Bitcoin Strategy: $65 Million Bet on AI Data Center

Cryptopolitan reported on July 2 that Nasdaq-listed company Empery Digital (NASDAQ: EMPD) announced it is abandoning its bitcoin treasury strategy. Instead, it plans to spend $65 million to acquire a 25% stake in an AI data center in the U.S. Midwest—marking the first clear strategy reversal among the group of bitcoin-treasury imitators associated with Michael Saylor. In the past, Empery accumulated more than 4,000 BTC at an average price of over $117,000. After BTC fell below $70,000, the company’s paper loss exceeded 40%. The company’s major shareholder, Tice P. Brown, publicly called on CEO Ryan Lane to step down and to fully liquidate its crypto assets.

The turning point: a 40% book loss + pressure on the CEO to resign
Empery initially operated as an electric vehicle manufacturer before pivoting into a bitcoin treasury company, copying the BTC accumulation model of Michael Saylor, chairman of MicroStrategy. In the past, the company accumulated more than 4,000 BTC at an average price of over $117,000. Earlier this year, BTC briefly dropped below $70,000, and the company’s unrealized losses widened to over 40%. The major shareholder Tice P. Brown, who holds about 10% of the shares, issued an open letter demanding that CEO Ryan Lane resign and that the company fully liquidate its crypto holdings. When BTC further sank, Empery was forced to sell 370 BTC at roughly $66,632 per coin to repay a $105 million margin loan. The company currently still holds 2,914 BTC.

New direction: a 150MW data center in the Midwest, $230 million total price
Empery announced it will invest $65 million to buy a 25% stake in a Midwest AI data center. The site currently has about 150 MW of power capacity and can be expanded to 300 MW. Partner Hunt Properties (a Dallas-based real estate company) holds another 75% through its subsidiary TexStack Infrastructure, and the total property purchase price is about $230 million. Empery has already paid $2.9 million in initial capital. The remaining $62.1 million will be paid on the closing date in Q3 2026, and due diligence must be completed by July 29. In addition, the company simultaneously announced it would remove the bitcoin treasury dashboard, stating that “the metric can no longer fully reflect the company’s total NAV.”

DAT’s plot twist: the first reversal in the Saylor imitator camp
Empery is the first case of an explicit reversal among bitcoin-treasury imitators of Saylor’s strategy. Over the past 12–18 months, these companies have shared a common pattern: purchasing BTC through equity or debt financing, trading their stock at a premium with the market treating them as “bitcoin-exposed” businesses, and expecting NAV to rise by keeping BTC gains going. Empery’s case shows that when BTC falls below the cost range, shareholders may push the company to abandon its strategy—where a 40% book loss plus a major shareholder’s public letter becomes the key catalyst.

Earlier reports covered MicroStrategy’s mNAV arbitrage strategy when mNAV fell below 1, along with another story about Anthropic betting on AI for drug discovery. Empery’s double-transition case—from EV to BTC to AI—reflects a rising frequency of small listed companies switching narratives during the post-SVB/post-BTC peak period.

This article, Empery Digital Dumps Bitcoin Strategy: $65 Million Bet on an AI Data Center, first appeared on .

FBI Director Patel Reports Late: Owns MSTR Shares Worth $100,001–$250,000 for 6 Months

On July 1, the U.S. NOTUS and several media outlets reported that FBI Director Kash Patel purchased MicroStrategy (MSTR) stock worth between $100,001 and $250,000 on November 21, 2025, but did not file the disclosure with federal regulators until May 26, 2026—later than the 45-day deadline required by the STOCK Act. MSTR is the world’s largest publicly traded company holding bitcoin, and it also has long-standing government procurement contracts with the U.S. Department of Justice (DOJ), the FBI’s parent agency. A first-time offender can be fined $200 by law; however, DOJ has not yet issued a penalty to Patel. Timeline: Bought in November; disclosed over 6 months later Patel’s purchase date was November 21, 2025, the amount fell within the $100,001 to $250,000 range, and the disclosure to the federal government was filed on May 26, 2026. The gap of more than 6 months far exceeds the STOCK Act’s requirement that senior officials in the executive branch publicly disclose stock transactions within 45 days. In a letter to the U.S. Office of Government Ethics, Patel said the transaction was “inadvertently omitted” from previously submitted financial disclosure forms. Deputy Assistant Attorney General William Taylor said the late filing was an “unspecified miscommunication,” adding, “I continue to believe that Director Patel complies with the relevant laws regarding conflicts of interest.” FBI officials also said the late filing was “not noticed” and “not intentional.”

MicroStrategy–DOJ Contract Relationship: Direct Conflict Raises Questions The sensitivity of this case lies in the fact that MicroStrategy, over the past decade, has had government procurement contracts worth millions of dollars with the DOJ—while the FBI is an organization under the DOJ’s law-enforcement umbrella. MicroStrategy is known to outsiders as a “bitcoin treasury company,” and is the world’s largest publicly traded holder of bitcoin. Patel personally owns shares of MSTR, while the agency he leads is part of MSTR’s government customer base—exactly the type of conflict of interest the STOCK Act seeks to uncover through timely disclosure. Dylan Hedtler-Gaudette, policy vice president at the nonpartisan watchdog Project on Government Oversight (POGO), bluntly said, “This is illegal—there’s no other way to describe it.” The Office of Government Ethics declined to comment on the case.

Broader Context: Trump Administration Officials x Crypto Asset Disclosure Controversies Patel’s late filing is just one of the latest controversies involving crypto-related financial disclosures among members of the Trump administration. Earlier reporting said that Trump’s own 2026 financial disclosures showed more than $1.4 billion in crypto-related income, with meme coins accounting for much of it. Another report said that on June 30, the U.S. Department of Commerce lifted export controls on Anthropic’s Fable 5, and that communication between the White House and the AI industry has intensified. What ultimately happens in the Patel case—whether DOJ issues a $200 first-offense fine, or triggers further investigations—will be the focus of ongoing observation.

This article first appeared on .

Claude Fable 5 Complete Tutorial: Get Started with Loop Engineering

Claude Fable 5 is Anthropic’s flagship model released on June 9, 2026. It targets a 1M context, autonomous multi-day operation without supervision, and xhigh-effort tier reasoning. Fable 5 was temporarily paused on June 12 due to export controls by the Ministry of Commerce, then the ban was lifted and it went back online on June 30. It is currently Anthropic’s highest-tier model available via API. This article summarizes the key points from Anthropic’s official “Prompting Claude Fable 5” handbook, breaking down the core behavior changes in Fable 5 compared with Opus 4.8, and how to adjust the corresponding prompts and harnesses. It is a complete first Chinese guide for getting started with Fable 5, suitable for developers, AI agent builders, and enterprise technology decision-makers.

What is Fable 5: 1M context, xhigh effort, multi-day autonomous operation
Fable 5 API model: claude-fable-5. It also has the following specifications:
- Context window: 1M tokens (million-level context). Maximum output per call: 128K tokens.
- Pricing: Input $10/M tokens, Output $50/M tokens (prompt caching can get a 10% discount).
- Local reasoning in the United States: 1.1× price surcharge.
- Available platforms: Claude API, Claude Platform on AWS, Amazon Bedrock, Google Cloud, Microsoft Foundry.

Adaptive thinking:
- Unique “thinking” mode; cannot be disabled.
- thinking.display can be set to summarized (read the summary) or omitted (do not read, default).

Safety classifier:
- Fable 5 includes a built-in refusal classifier. Requests involving attack-oriented cybersecurity, biological and life sciences, or requests to extract model reasoning will be refused (HTTP 200 but stop_reason: refusal).
- Mythos 5 has no such classifier; it is only available to approved Project Glasswing customers.

Data retention: 30 days (not available for ZDR zero data retention).

Anthropic’s official positioning of Fable 5 improvements over Opus 4.8 is in seven areas:
1) Long-horizon autonomy
2) Higher first-shot correctness for complex problems
3) Vision
4) Enterprise workflows
5) Code review and debugging
6) Handling ambiguous requirements
7) Concurrent sub-agents

The shared underlying logic behind these seven improvements is one thing: Fable 5 can continuously and stably produce outputs during the time when the user isn’t watching.

Core paradigm shift: From Prompt Engineering to Loop Engineering
The biggest mindset change with Fable 5 isn’t writing better prompts—it’s changing from prompt steps to designing loops. With Opus 4.6/4.7, the mindset was: “Write a good prompt; Claude returns a good answer.” With Fable 5, the mindset is: “Design an environmental loop that lets Claude self-correct, repeatedly cycling through plan → act → review → improve until the task is complete.”

In the official handbook, Anthropic explicitly provides this loop design instruction:
“Establish a method for checking your own work at an interval of [X] as you build. Run this every [X] interval, verifying your work with subagents against the specification.”

A translation of this passage:
“Establish a self-check mechanism to validate your own work at intervals of [X]. Every [X] interval, run verification and check your work against the specification using subagents.”

The key is to use a set of independent context verification subagents (fresh-context verifier subagents), not let the main agent critique itself—because the main agent already has bias.

This shift also responds to the previously reported “Harness Engineering” concept: the next battleground for AI is not the model, but the layer of architecture outside the model; and Akshay’s viewpoint that “the model is just a node in the loop.” Loop engineering is Anthropic’s official version of this line of thinking.

Effort parameter grading:
5-level control of how many tokens Fable uses
Effort is Fable 5’s most important cost-control knob. It is divided into 5 levels, from highest to lowest, affecting how many tokens Claude is willing to spend on thinking, tool calls, and text responses:
- max: highest capability, no cost ceiling. Used for frontier problems requiring the deepest reasoning. Frequent use will generate large costs, not recommended for routine use.
- xhigh: expanded capabilities, suitable for durations over 30 minutes, token usage...”

Do meme numbers come true? Tesla shares return to $420 and sparks a retail-frenzy

On Tuesday (6/30), Tesla (TSLA) closed at a meme-like $420.60 for the first time in 16 years since listing, sparking market attention on the momentum behind retail “meme trading.” According to data from Vanda Research, whenever Tesla’s stock price moves above $420, the average total net buy by retail investors over the next five days rises from $87 million to $135 million. The phenomenon reflects the subcultural symbolism behind the number, creating a distinctive psychological pull for the retail investor base that is highly engaged with Tesla. Although retail fund inflows slowed on Wednesday as the market waited for the Q2 delivery report, historical data and technical indicators suggest that this price level has already formed real behavioral support in the retail market—further deepening the unique link between Tesla’s share price and its retail community.

420 in Tesla’s past controversy

The number 420 (4/20) is a global cannabis-culture code, originating from U.S. students’ slang in the 1970s (“4:20 in the afternoon to smoke weed”), and later becoming the unofficial “Cannabis Day” every year on April 20. 420 also carries high sensitivity in Tesla’s development history. In 2018, Elon Musk publicly posted a tweet saying he was considering taking Tesla private at $420 per share and claimed “Funding secured.” The tweet sent Tesla’s stock soaring in the short term, but the privatization plan never materialized, leading to an SEC investigation and a class-action lawsuit by shareholders. Musk was then temporarily removed from his role as Tesla’s chair and paid a fine.

At the time, the SEC alleged in the lawsuit that Musk chose 420 because it was a cannabis-culture inside joke that would make his then-girlfriend laugh. Musk denied this in court, emphasizing it was purely the result of rounding a 20% premium in the stock price, with no connection to cannabis.

Afterward, whenever Tesla’s stock price repeatedly hit or neared $420, it would trigger discussion in the market and even from Musk himself. For example, when the stock first reached around $420 by the end of 2019, Musk posted “Whoa… the stock is so high lol,” clearly playing on the meme.

Tesla stock hitting $420 triggers a retail buying rush

According to data provided by Viraj Patel, Global Macro Strategist at Vanda Research, when Tesla’s stock breaks above $420, the average five-day net buy amount by retail investors increases significantly from $87 million to $135 million. Since mid-2024, across 14 recorded instances of the stock crossing the $420 threshold, 9 of them have sparked substantial additional buying from retail investors. This indicates that, in the eyes of retail traders, a specific number already carries psychological support strength that goes beyond conventional technical analysis—turning it into an important retail behavior indicator that cannot be ignored in quantitative trading.

Dave Mazza, CEO of Roundhill Investments, noted that Bitcoin (BTC) challenging $69,000 in the past, as well as GameStop retail’s fixation on whole-number levels, are typical cases of “meme trading.” Musk’s long-running management of his personal public image has made 420—a subcultural symbol originating from cannabis culture—a shared marker within the Tesla community. Steve Sosnick, chief strategist at Interactive Brokers, said this phenomenon has persisted for years within the Tesla community, and rising discussion on the social platform X has further reinforced retail investors’ consensus to enter.

TSLA is still down nearly 3% this year. Despite the meme effect being talk-worthy, the stock’s real trajectory remains supported by the broader market and technical factors. On Wednesday, Tesla’s stock closed at $425.3. However, Vanda data also shows that retail fund inflows actually declined on Wednesday, mainly because the market is holding its breath for the Q2 vehicle delivery report scheduled for Thursday.

Tesla’s stock is down nearly 3% this year, underperforming the S&P 500’s +9%.

This article

“Do meme numbers come true? Tesla shares return to $420 and spark a retail-frenzy”

First appeared on .

Standard Chartered hails Morpho 2030 $60: 30x surge potential, to outperform BTC/ETH

On July 1, Standard Chartered Bank’s digital assets head, Geoff Kendrick, officially launched coverage of Morpho and set a 2030 target price of $60—implying roughly 30x upside—and expects Morpho’s performance to surpass Bitcoin and Ethereum. Standard Chartered’s year-by-year path: 2026 $3.50, 2027 $11, 2028 $22, 2029 $40, and 2030 $60. Standard Chartered expects the DeFi asset market to grow 37x by 2030, with Morpho’s vaults business seen as the key differentiator in drawing traditional finance capital on-chain.

Geoff Kendrick’s year-by-year path: $3.5 → $11 → $22 → $40 → $60. Kendrick is Standard Chartered’s head of digital asset research, covering institutional-grade crypto asset investment targets. This Morpho coverage marks the bank’s first independent analyst view on DeFi lending protocols. Year-by-year price path: $3.50 at the end of 2026, $11 in 2027, $22 in 2028, $40 in 2029, and $60 at the end of 2030. Compared with the current Morpho price, the 2030 target implies about 30x upside, meaning that in Standard Chartered’s model, Morpho will significantly outperform the 5-year return expectations for BTC and ETH.

Standard Chartered’s thesis: DeFi assets grow 37x over five years; Morpho vaults are differentiated. The bank’s long-term view is based on DeFi asset market growth of 37x by 2030. Under this assumption, Morpho vaults are considered a key differentiator: vaults allow institutions to deploy large-scale traditional financial assets on-chain, earn predictable returns, and manage risk. Morpho is currently the second-largest DeFi lending protocol, behind Aave; together, the two account for 57% of deposits and 63% of active loans across lending protocols.

Anton Cheng joins the Morpho narrative: accelerated institutional-grade DeFi infrastructure. This round of analyst coverage from Standard Chartered, along with another report from yesterday about DeFi developer Anton Cheng joining the Morpho APAC team, forms a corroborating narrative chain: on one side, institutional analysts back the story with long-term target prices; on the other, a seasoned DeFi developer chooses Morpho as the next stop. Standard Chartered also highlighted Aave’s target price of $3,500 (another major DeFi lending giant).

In September 2024, Anton Cheng’s Morpho Hyperstructure deep-dive was published separately; in November 2025, a separate publication compared Morpho’s architecture with AAVE and Compound.

This article, “Standard Chartered calls Morpho 2030 $60: 30x surge potential, to outpace BTC/ETH,” first appeared on .

Challenging Nvidia’s position? AI chip startup Etched gains TSMC support, wins a $1 billion order

AI chip startup Etched has recently announced that with assistance from TSMC, it has successfully completed its first batch of chip manufacturing and signed contracts worth more than $1 billion, while also disclosing that the company has raised $800 million in funding. Etched’s flagship offering is an end-to-end system called “frontier inference clusters,” designed to tackle the high-cost and performance bottlenecks in AI inference. For industry leader Nvidia, it could be seen as an upstart. We’re coming out of stealth. We’ve built our first racks after a successful A0 tapeout, $1B+ in customer contracts, and $800m raised. Early customer tests show us achieving SOTA throughput, latency, and power efficiency on inference workloads. Our first racks ship this summer. pic.twitter.com/FLccrkLTza — Etched (@Etched) June 30, 2026 has set up a factory in Taiwan and is pushing hard to ramp up production! Etched clinches a $1 billion order
Etched’s post says it is currently testing its first batch of end-to-end systems with customers, while actively expanding production capacity and having already established a factory in Taiwan. Etched emphasized that early testing results have achieved industry-leading (SOTA) throughput, latency, and power efficiency for inference workloads. More complete performance data is expected to be released this summer. Compared with existing competitors, the system can execute AI inference models at faster speeds, lower costs, and with better energy efficiency. The company has already signed customer orders worth more than $1 billion.

Built on TSMC’s 4nm process, Etched bets on transformer-specific chips
Etched’s flagship chip is called Sohu, using TSMC’s 4nm process. Unlike general-purpose GPUs, Sohu burns the transformer architecture directly into the chip. By specializing the design, it delivers higher inference performance and energy efficiency. Etched previously claimed that servers equipped with eight Sohu chips can outperform Nvidia’s H100 by about 20 times.

Dropped out of Harvard to start a new company! Latest valuation reaches $5 billion
Etched was founded in 2022 by Harvard dropouts Gavin Uberti, Chris Zhu, and Robert Wachen. The company has raised a total of $800 million. The largest funding round was a $500 million raise completed in December 2025, led by Stripes, with a post-money valuation of $5 billion. According to a report by Bloomberg, investors include VentureTech Alliance, a venture capital firm closely tied to TSMC, as well as Peter Thiel and quantitative trading and market institutions including Jane Street, Hudson River Trading, Jump Trading, and Two Sigma, along with several other heavyweight figures in the AI field such as Geoffrey Hinton, Fei-Fei Li, Andrej Karpathy, and investor Stanley Druckenmiller.

AI chip market competition heats up; specialized chips become a battleground for tech giants
Compared with Etched’s early situation in 2023 when it struggled to raise funds, today’s AI chip investment environment is clearly much hotter. Competitors like Cerebras completed an IPO this year, and Groq also completed another major funding round. Large cloud providers such as Amazon, Google, and Microsoft continue to design and build their own AI chips. OpenAI has also unveiled its first custom chip, manufactured by Broadcom.

Etched says that if testing goes smoothly and mass production proceeds on schedule, the end-to-end system is expected to gain a competitive advantage in throughput, latency, and power efficiency—helping it increase market share in competition with major cloud providers and AI service companies.

This article Challenging Nvidia’s position? AI chip startup Etched gains TSMC support, wins a $1 billion order First appeared on .

SEC Seeks Input on Novel ETF Regulation: Crypto, Tokenization, and Event Contracts Included

On June 30, 2026, the U.S. Securities and Exchange Commission (SEC) issued Press Release 2026-60, opening a 60-day public comment period to re-examine the regulatory framework for so-called “novel ETFs.” The scope of the request covers crypto-asset ETFs, event contract ETFs, tokenized securities ETFs, and single-stock strategy ETFs. The request document, Release No. 33-11426, poses 27 questions but does not propose specific rule changes. The 60-day deadline begins after the notice is published in the Federal Register. Atkins declared: ETF innovation needs a consistent regulatory framework. In the press release, SEC Chair Paul S. Atkins said, “ETF innovation relies on a consistent, transparent, and effective regulatory framework.” He emphasized that the purpose of this inquiry is to collect public views on how the U.S. ETF market can continue to grow and innovate while effectively serving investors. One of the core questions is whether ETFs investing in non-securities assets (such as crypto assets, tokenized assets, and event contracts) can still fall under the definition of “investment companies” in the Investment Company Act of 1940. This definitional issue directly affects the registration pathways and regulatory obligations available to industry participants. On May 21, it was reported that the SEC would put off prediction market ETFs and that Chair Atkins requested public input; this public comment period is the formal implementation of that policy direction.

Covered scope: crypto assets, event contracts, tokenized securities, single stocks. This inquiry clearly lists four categories of novel ETFs. The first category is crypto-asset ETFs, covering currently approved Bitcoin and Ethereum ETPs and other crypto-asset ETFs that may be listed. The second category is event contract ETFs, directly touching the ETF-ization paths for prediction markets such as Polymarket and Kalshi; on June 24, it was also reported that Cboe Predicts’ binary options officially launched. The third category is tokenized securities ETFs, aligning with the tokenized securities exemption policy Atkins advanced in June; on the same day, it was also reported that tokenization leader Securitize would list on the NYSE on 7/2. The fourth category is single-stock strategy ETFs, covering structured products such as leveraged, inverse, and target-date strategies.

Policy signal: Framework reshuffle amid the ETF market $4T→$12T. The macro backdrop for the SEC’s solicitation is the expansion of the ETF market. U.S. ETF assets grew from about $4 trillion in 2019 to about $12 trillion in 2025. Until now, much of the growth relied on the SEC’s streamlined approval process, effectively bypassing the formal exemption application procedure. TD Cowen analyst Jaret Seiberg believes the design of this public solicitation is intended to “create the record needed for future policy changes,” reflecting Chair Atkins’ policy priorities embracing crypto and tokenization. For the industry, this inquiry is the most important guidance for understanding the SEC’s future formal regulatory direction for novel ETFs and the potential timeline for expanding product categories. On June 30, it was also reported that New York Life first tokenized assets, showing that institutional capital is moving quickly into the product categories covered by this inquiry.

This article, “SEC Seeks Input on Novel ETF Regulation: Crypto, Tokenization, and Event Contracts Included,” first appeared on .

Anton Cheng joins Morpho: letting go of purism, returning to institutionalized cooperation

DeFi developer and author Anton Cheng (antonttc) published an article titled “On This Era” on his personal blog on June 25, documenting his reflections as he wrapped up 2.5 years of freelancing and prepared to head to Paris to join the Morpho APAC team. In the past, Anton has written in-depth pieces on Morpho, including what’s different between Morpho, AAVE, and Compound, as well as the paradigm for the next generation of DeFi foundations (Hyperstructure): the lending protocol Morpho. Starting from his perspective, the article examines a generational shift in the mindset of DeFi developers.

Self-awareness of the “ascetic personality”: triggers from Mandate drama and fork events
Anton describes two triggers that hit this March. On one side, a Mandate drama unfolded within the Ethereum Foundation—large-scale restructuring, layoffs of about 20%, and the reorganization into five major “clusters.” At the same time, his own Monarch platform was forked, triggering intense anger. Anton wrote that as a developer in the DeFi community committed to open source, transparency, and decentralization, he may not necessarily receive equivalent support from the Foundation.

What prompted his self-reflection was a conversation with AI. The AI reminded him: “A martyr is someone who voluntarily chooses suffering, bloodshed, or death to prove the purity of their beliefs.” From this, Anton realized he has a form of psychological “purity obsession” typical of an ascetic personality—overly pursuing moral purity that instead becomes a source of self-harm. In the article, he redefines his view on making money and compromise: “Making money isn’t a sin—only greed is.” He then starts evaluating whether his Monarch platform sacrificed functionality and user experience in the name of absolute neutrality.

From independent developer to Morpho APAC: choosing institutionalized collaboration
Anton’s turning point was catalyzed by a recommendation from DeFi developer Merlin. Anton wrote that after a “major life reassessment,” he decided to join Morpho’s APAC team and give up his identity as an independent developer. Morpho had just completed a $175 million fundraising on June 9, co-led by Paradigm, a16z crypto, and Ribbit Capital, valuing the company at $2 billion. The deal elevates Morpho from a DeFi lending protocol to a provider of infrastructure for an “Open Credit Network.” For a researcher long focused on the architecture of lending protocols, Morpho APAC is a workplace with resources and partners.

Anton also honestly wrote about his worries about joining an institution: the plot of “the dragon slayer becoming a dragon” is still something DeFi developers have long been wary of. He clearly states in the article that this risk isn’t just hypothetical—it’s a real scenario. Still, he chose to accept the role because he believes institutionalized collaboration can help DeFi go further.

“We will ultimately be forgotten by history”: the generational sense of the times among DeFi developers
In the article, Anton asks people around him a question he often asks: “How do you think people 50 years from now will describe the era we’re in today?” He lists several possible historical narratives—unchecked capitalist expansion, an AI bubble, the pre-AGI era, and the decline of Western civilization. At the same time, he admits, “It’s impossible to distinguish whether my predictions are things I want to see, or things I believe will happen,” and describes his mental state as “believing very strongly in what I believe, and caring very little about whether I’m right.”

Anton cites the psychology concept “End of History Illusion”—the tendency for people to underestimate how much their future may change. He uses the example of how much his life has changed from age 20 to now, far beyond expectations, to argue that his future self should have an even bolder imagination. The article ends with a line from the animated film “Earth. On the Movement of the Earth”: “We will ultimately be forgotten by history,” but it immediately follows with another line that encourages him to keep moving forward.

In a separate report on May 25, Vitalik discussed the Ethereum Foundation’s transformation and argued that the “EF should not be Ethereum’s leader, but a mission-driven node,” which mirrors the generational backdrop Anton reflects on.

This article, “Anton Cheng joins Morpho: letting go of purism, returning to institutionalized cooperation,” first appeared on .

Claude moves into Azure Foundry GA: first run of GB300 clusters

On June 29, Anthropic will officially make the Claude model family generally available (GA) on Microsoft Foundry on Azure. With the Opus 4.8 and Haiku 4.5 dual models, Azure customers can use them directly in Azure clients, with support for Azure Active Directory authentication, unified billing, and full integration with the Anthropic Messages API. This deployment is among the first frontier AI services running on NVIDIA GB300 Blackwell Ultra GPU clusters in Azure. With the launch of both Opus 4.8 and Haiku 4.5, pricing is the same as buying directly from Azure: Opus 4.8 input pricing is USD 15 per million tokens, and output pricing is USD 30 per million tokens; Haiku 4.5 input pricing is USD 0.25 per million tokens, and output pricing is USD 1.25 per million tokens. Azure customers can call the service through the Foundry unified interface without needing to apply for an Anthropic account or handle cross-platform billing. Azure AD authentication allows enterprise customers to use their existing identity management mechanisms to control access permissions for using Claude models.

First NVIDIA GB300 Blackwell Ultra deployment: Azure Accelerator upgrade
What’s special about this deployment is the underlying infrastructure: it runs entirely on NVIDIA’s next-generation GB300 Blackwell Ultra GPU clusters and is the first frontier AI service on Azure to enable this architecture. GB300 is NVIDIA’s latest AI accelerator announced in June; compared with the previous generation GB200, it delivers significant improvements in inference throughput and energy efficiency. This deployment also extends the collaboration between Microsoft and Anthropic from the May Series H investment down into the underlying compute layer.

Regional availability: eight data centers across North America, Europe, and Asia-Pacific
The initial available regions cover three major regions—North America (East US, West US, Canada Central), Europe (West Europe, North Europe), and Asia-Pacific (Australia East, Japan East)—for a total of eight Azure data centers.

Claude’s GA on Microsoft Foundry, and Anthropic’s previous collaborations with AWS Bedrock and Google Cloud Vertex AI, are proceeding in parallel; on April 7, another report said that Anthropic partnered with Broadcom and Google to obtain 3.5 gigawatts of TPU compute power, reflecting a landscape where these three hyperscalers jointly support Claude training and inference.

This article “Claude moves into Azure Foundry GA: first run of GB300 clusters” first appeared on .

Taiwan VASP Association’s Three Missions: Sub-legislation Promotion, Enforcement Against Overseas Platforms, and Cross-Border Joint Anti-Scam Prevention

On June 30, the day the Legislative Yuan passed in a third reading the “Virtual Asset Services Act,” the Republic of China Virtual Asset Service Commercial Co-op Association (VASP Association) immediately issued a statement in response. It said that the passage of the special act signifies that Taiwan’s virtual asset industry has officially entered a new era of governance under the “special act.” The association announced three core missions: promoting related sub-legislation to ensure a smooth transition for the market; aligning with international standards and strengthening the management of overseas platforms; and linking up the criminal-investigation, police, and judicial authorities with financial institutions to enhance cross-border anti-scam and anti-fraud mechanisms. It will also guide industry participants through three committees—(1) listing and delisting review, (2) discipline, and (3) the committee on preventing financial fraud and legal compliance.

Sub-legislation Promotion: Seven practical regulations will be implemented in stages
The VASP Association stated it will actively assist the competent authority, the Financial Supervisory Commission (FSC), in drafting the sub-legislation related to the special act. The association listed seven practical regulations: establishment rules, personnel management rules, management rules, guidelines for handling internal control systems, regulations for managing abnormal transactions, outsourcing operation rules, and rules for preparing financial reports. These sub-legislations are key steps for the practical implementation of the bill, and will directly affect compliance pathways and capital/organizational/internal-control requirements for businesses during the transition period.

On June 30, additional in-depth reporting covered the transition timeline after the VASP Act’s third reading. For businesses that have already completed anti-money laundering registration, they must apply for licenses within 12 months after the act takes effect and obtain certificates within 21 months; if necessary, the deadline may be extended by an additional 3 months.

The association said it will guide businesses through the transition via three committees: the listing/delisting review committee, the discipline committee, and the committee on preventing financial fraud and legal compliance. The establishment of the three committees aligns with the common management requirements for virtual asset service providers set out in the provisions (including internal control and audit systems, information and communications technology system security management systems, and business continuity policies), and also reflects the statutory role of trade associations in self-regulation and member management.

Aligning with International Standards: Establishing an enforcement basis for unlicensed overseas platforms
In its statement, the association said it will introduce mechanisms in line with international standards, including anti-money laundering measures, cybersecurity controls, investor protection, and maintenance of market order. Referencing international regulatory trends, it will establish clear management and enforcement grounds for overseas virtual asset platforms that provide services to, or solicit business from, citizens of Taiwan without authorization—so as to avoid regulatory arbitrage and risk spillover. This section directly echoes the 15 additional resolutions passed by the Legislative Yuan calling for strengthened management of overseas transaction platforms established without approval. It is expected that through sub-legislation, administrative guidance, and cross-border cooperation mechanisms, enforcement against unlicensed overseas platforms will be strengthened.

On June 30, it also released a full side-by-side comparison of the three frameworks: Taiwan vs. the EU MiCA vs. the U.S. GENIUS Act. Under the strict restrictions applied in all three jurisdictions to overseas issuers, MiCA requires third-country issuers to meet conditions for EU passport admission; GENIUS Act requires foreign stablecoin issuers’ home-country regulatory framework to be recognized as “equivalent” by the U.S. Department of the Treasury and to complete OCC registration; whereas Taiwan requires overseas VASPs to set up a branch within Taiwan and obtain FSC approval.

Cross-Border Joint Prevention: On-chain analytics and intelligence sharing with police, prosecutors, and investigative authorities to fight scams
The association announced that it will connect investigative, police and prosecution authorities, competent authorities, financial institutions, and compliant VASP businesses to establish a more timely and tighter intelligence-sharing mechanism, suspicious transaction reporting, and a joint anti-scam prevention system. Specific measures include on-chain transaction trail analysis, identification of high-risk addresses, monitoring of abnormal transactions, and mechanisms for assisting in fund-flow verification. These efforts will help prevent and block scam funds and improve overall effectiveness in combating financial crimes.

This mission echoes the provisions stating that when a virtual asset service provider discovers suspected illegal activity or abnormal transactions, it should immediately suspend relevant transactions and cooperate with the government in strengthening anti-scam efforts. It also echoes the policy direction in additional resolutions from the Legislative Yuan calling for stronger oversight of overseas platforms and closing loopholes used for money laundering and scams.

On June 30, it also released a complete 2026 guide to the Virtual Asset Services Act, including a table comparing article-by-article sections, analysis of 15 additional resolutions, and an 8-question FAQ.

This article—“Taiwan VASP Association’s Three Missions: Sub-legislation Promotion, Enforcement Against Overseas Platforms, and Cross-Border Joint Anti-Scam Prevention”—appeared earliest on.

The yen falls through 162 to a 40-year low: where is the Japanese government’s intervention line?

The yen-to-US dollar exchange rate has broken through the 162 level, hitting the lowest level since 1986. Although the Bank of Japan raised its policy rate to 1% on June 16, the interest-rate differential between Japan and the U.S. remains significant because market expectations are that the U.S. Federal Reserve will keep interest rates high. This situation has continued to trigger capital outflows and downward pressure on the yen, which in turn has caused a sharp increase in Japan’s import energy costs and placed a burden on domestic inflation and consumers. The market is currently closely watching whether the exchange rate will test the 164 to 165 range further, and when the authorities will step in again to curb the decline.

Even after the BoJ’s rate hike, the Japan-U.S. interest spread remains wide, and the yen keeps weakening

Even though the Bank of Japan (BOJ) raised its policy rate to 1% on June 16—its highest level since 1995—market sentiment toward the yen remains pessimistic. The main reason is that the U.S. Federal Reserve’s recent stance has been more hawkish, making it difficult for the interest-rate differential structure between the two countries to reverse in the short term. In addition, Japan’s structural issues—such as population aging and large public debt—also limit room for a substantial further rate hike. Under these circumstances, investors tend to borrow low-cost yen through carry trades and shift investments into overseas assets with higher yields. This kind of capital outflow continues to exert downward pressure on Japan’s exchange rate.

The yen-to-US dollar exchange rate breaks through the 162 level, reaching the lowest level since 1986.

Multiple interventions have failed; Satsuki Katayama says she is prepared to take action

In response to the yen’s continued weakness, the Japanese government has previously injected a record 11.73 trillion yen (about $72.4 billion) into foreign exchange intervention between April 28 and May 27. However, in the foreign exchange market where daily global trading volume reaches $9.5 trillion, intervention by a single country often cannot sustainably change the overall trend. Japan’s Finance Minister Satsuki Katayama reiterated again that the authorities are ready to take decisive action and have reached a policy consensus with U.S. Treasury Secretary Scott Bessent. As Bloomberg analysts noted, the market is currently closely monitoring whether the exchange rate will further slide into the 164 to 165 range, and when the authorities will move in again to prevent further declines.

Yen depreciation creates pressure on energy imports and inflation

Yen depreciation has a double-edged impact on Japan’s domestic economy, with the most negative effect coming from rising import costs. Because Japan relies heavily on imported energy—especially oil and natural gas from the Middle East—recent regional conflict involving the U.S. and Israel against Iran has further increased uncertainty in energy supplies. Energy import prices denominated in U.S. dollars have surged sharply, directly intensifying domestic inflation pressure. From food to electricity, prices of essential goods rising across the board not only erode consumers’ real purchasing power but also pose potential political risks to support for Prime Minister Sanae Takaychi’s cabinet.

Exchange-rate gains for exporters surge, supporting Japan’s stock market hitting new highs

Despite inflation challenges at home, the weak yen provides a real financial boost to Japan’s export-oriented companies. When the yen weakens, the dollar or euro revenue earned overseas by companies, after being converted back into domestic currency, increases significantly in book value. Taking Toyota Motor Corp. as an example: it is estimated that for every 1-yen depreciation of the yen, operating profit could rise by about 50 billion yen, potentially generating up to $5.8 billion in foreign-exchange gains for Japan’s automakers.

This article: The yen breaks 162 to a 40-year low—where is the Japanese government’s intervention line? first appeared on …

Ukraine’s first case: seized $8.3M in cryptocurrency transferred to ARMA for state management

At the end of June, the United States time, Ukraine transferred confiscated crypto assets for the first time to the National Agency for Finding, Tracing and Management of Assets (ARMA). According to a report by Decrypt, the transferred USDT worth $8.3 million came from a member of a multinational hacking group who was indicted and accused of carrying out ransomware attacks totaling more than $100 million against companies in Europe and the United States. In the same case, $11.1 million in assets were also confiscated, including real estate, cars, and $1 million in cash. This transfer marks ARMA’s first time handling digital assets, and represents Ukraine’s concrete implementation of “state-managed crypto assets.”

ARMA reform backdrop: EU funding injection + requirements for transparency

ARMA (full name: National Agency for Finding, Tracing and Management of Assets) is responsible for overseeing assets confiscated in criminal cases. This first-time handling of digital assets is a tangible result after the agency’s large-scale reforms in 2025. The reforms were prompted by long-standing criticism that ARMA was opaque and inefficient in asset management. After the reforms, it unlocked hundreds of millions of euros in EU support and required ARMA to publish transparent procedures for managing confiscated assets. For the EU, Ukraine’s progress in building this kind of judicial infrastructure is an important indicator for subsequent aid and the process of joining the EU.

Ukraine’s crypto economy scale: $206.3 billion in annual trading volume; officials hold $2.8 billion in BTC

Ukraine’s presence in Europe’s crypto market is far from small. According to Chainalysis, between mid-2024 and mid-2025, Ukraine’s crypto trading volume reached $206.3 billion, ranking fourth in Europe. The total value of bitcoin holdings publicly reported by Ukrainian officials is about $2.8 billion—reflecting a very high rate of crypto adoption in the country, along with a relatively mature public disclosure mechanism. For the state, establishing a credible “state-managed confiscated assets” mechanism is a necessary foundation—handling wartime funds, pursuing recovery of cross-border criminal assets, and enforcing sanctions (especially related assets connected to Russia) all require this kind of infrastructure.

A signal for the crypto industry: a new precedent for state-level crypto management

ARMA’s process for handling digital assets for the first time, and the subsequent asset disposal approach, will serve as a concrete benchmark for what the agency will disclose externally after its reforms.

This article
Ukraine’s first case: $8.3M in seized cryptocurrency transferred to ARMA for state management
First appeared on .

U.S. and Iran Pause Hostilities: Hormuz Shipping Resumes, Both Return to Technical Talks

On June 28-29 (U.S. time), CNBC reported that both the U.S. and Iran paused hostilities and that shipping through the Strait of Hormuz has resumed. A U.S. official told CNBC, “Technical talks will continue in all areas of the MoU; for now, both sides will pause (hostile actions), and ships can transit freely.” After reports on June 27 of U.S. airstrikes against Iran, the weekend saw another round of escalation—Iran’s IRGC launched ballistic missiles and drones at U.S. forces stationed in Kuwait and Bahrain—but later in the weekend both sides agreed to halt military action and return to the 60-day ceasefire MoU framework signed on June 17.

Escalation over the weekend: Iran attacks U.S. military installations in Kuwait and Bahrain. In the early hours of Sunday, Iran’s IRGC launched ballistic missiles and drones at U.S. bases in Kuwait and Bahrain. A residential home in the Muharraq area of Bahrain was hit, and Kuwait’s international airport briefly suspended operations. This was the largest military action since the 60-day ceasefire framework was signed on June 17.

President Trump once posted threats of “annihilation” against Iran, but later said the ceasefire framework “is still valid technically,” and that negotiations with Iran were “going very smoothly.” Iran’s Ministry of Foreign Affairs does not accept the claim that the ceasefire is “still valid.” It publicly stated that there are substantive differences between the two sides over the current status of the ceasefire, while agreeing to pause military actions in order to maintain diplomatic channels.

“Technical talks continue”: The key next step under the 60-day MoU framework. U.S. officials specifically emphasized that “technical talks will continue on all areas of the MoU,” meaning that the 60-day ceasefire framework signed on June 17 has not broken down—it has only entered a tense monitoring period.

The MoU covers: tactical ceasefire, freedom of navigation through the Strait of Hormuz, a de-escalation mechanism in Lebanon, and monitoring of Iran’s nuclear facilities.

After earlier reporting that the U.S. and Iran reached a 60-day roadmap on 6/17, this escalation is the biggest test of the framework within 12 days of its establishment. The fact that both sides are willing to return to the technical talks table suggests that a diplomatic solution remains the main track in the near term.

Spillover effects on crypto and U.S. stocks: Risk-off sentiment fades briefly, but the structure remains fragile. The impact of this U.S.-Iran de-escalation on markets shows an asymmetric pattern: stocks rebound, while crypto does not respond significantly. In a report yesterday, IBIT showed an average investor loss of 40%, and BTC ETF monthly outflows totaled $4 billion. The geopolitical de-escalation did not significantly reverse the outflow trend for crypto assets. CoinDesk the same day noted that BTC dipped to $59,700 in the short term and did not rise along with U.S. equities.

For the crypto market, geopolitical risk is only one of multiple pressures in the first half of 2026 (liquidations driven by AI, interest-rate expectations, and the four-year cycle). A single event cooling off is not enough to change the overall trend.

Key points to watch next: When the 60-day MoU moves into substantive implementation, and whether Iran’s specific requirements for ceasefire conditions can align with the U.S.’s positions.

This article, “U.S. and Iran Pause Hostilities: Hormuz Shipping Resumes, Both Return to Technical Talks,” was first published on .

TSMC and Winbond team up for a “strong-strong” collaboration! WoW advanced packaging technology to build a local memory supply chain

Against the backdrop of a severe global memory shortage, Taiwan’s semiconductor foundry leader TSMC (2330) is accelerating efforts to nurture Taiwan’s local DRAM supply chain. Reports in the market indicate that TSMC has incorporated Taiwan’s homegrown memory heavyweight Winbond (2344) into its AI chip supply chain. The two sides will collaborate on next-generation “wafer-on-wafer” (WoW) advanced 3D wafer packaging technology. This move not only helps TSMC secure a stable source of memory for its AI chips, but also signals that Taiwan’s memory industry has officially entered the global high-end AI core supply chain.
TSMC Winbond WoW vertical stacking, breaking the “memory wall” bottleneck of AI chips
According to a report by United News Network, Winbond, leveraging its mature 12-inch wafer mass production capability and high yields, will provide key DRAM and other memory wafers to work with TSMC through “wafer-on-wafer” (Wafer-on-Wafer, WoW) integration, jointly stepping into the core supply chain for AI servers and high-performance computing (HPC).

Although both Winbond and TSMC currently have not commented on specific customers or market rumors, industry consensus is that this collaboration will help Winbond undergo a major transformation—enabling Taiwan’s memory industry to seize a new era of business opportunities in global AI high-end chip integration.

Reduce reliance on the three major U.S.-Korea firms: TSMC builds Taiwan’s local memory wafer supply chain
In the past, for the WoW technology, the memory wafers required by TSMC mainly came from major memory makers such as Samsung Electronics (Samsung), SK Hynix (SK Hynix), and Micron (Micron). However, as the global AI boom surged, international memory original equipment manufacturers’ production capacity was nearly fully utilized, leading to severe supply constraints.

To diversify risk, global AI supply chains have started seeking more diversified and more flexible sources of memory. Industry analysis suggests that this cooperation is not merely a straightforward procurement contract—it is a strategic move by TSMC to build a “Taiwan local AI chip ecosystem.” By nurturing local suppliers, TSMC can plan its AI chip supply capacity and strengthen supply-chain resilience.

What is WoW technology? The key to doubling AI chip performance
The semiconductor industry points out that WoW technology is regarded as an important core for next-generation AI chip integration. The technology uses a “hybrid bonding” process to directly vertically bond logic chips and memory wafers, forming tens of thousands to millions of micro-scale copper interconnects within the stack.

Compared with traditional packaging methods, WoW can significantly shorten the data transmission distance. This can deliver higher bandwidth, lower latency, and also better energy efficiency—making it an indispensable key technology direction for future edge AI devices and high-performance computing.

Winbond’s niche technology gains favor: Taiwan’s global strategic position in semiconductors upgraded again
TSMC’s screening standards for WoW collaboration partners are extremely strict. In addition to needing mature 12-inch wafer mass production capability, TSMC places heavy emphasis on high yields, specialized processes, and experience in wafer integration. Winbond has long focused on niche-type DRAM and embedded memory (NOR Flash) markets. With deep expertise accumulated in specialized memory processes and quality management, it has therefore become an important partner for TSMC’s AI memory supply chain plans.

Looking ahead, as WoW technology ramps up and is introduced into more AI platforms, it is expected to open up a new wave of long-term growth opportunities for Winbond, while making Taiwan’s semiconductor supply chain’s strategic position in the global AI landscape even more solid.

This article “TSMC and Winbond team up for a strong-strong collaboration! WoW advanced packaging technology to build a local memory supply chain” first appeared on .