Aston在下不求

Is Musk Working for Free at Tesla? Why Will His $158.3 Billion Salary Be Worthless in 2025?

Tesla recently submitted a report to the SEC revealing that CEO Elon Musk's book salary for 2025 could reach a staggering $158.3 billion, yet he won't pocket a dime. Since his compensation structure is entirely stock-based, Musk will receive no cash or vested shares that year. The report indicates that this hefty compensation is mainly composed of the '2025 CEO Performance Bonus' and the '2025 CEO Temporary Bonus' calculated at fair value. With a Delaware court reinstating his 2018 pay case at the end of 2025, the aforementioned temporary bonus has been completely forfeited by law. Total compensation of $158.3 billion, yet Musk receives nothing. According to Tesla's amended annual report (Form 10-K/A), Musk's reported total compensation for 2025 is $158.3 billion. Despite the massive figure, he will not receive any cash salary or non-equity incentive bonuses. Since 2019, Tesla has eliminated the base salary at his request, meaning his pay relies entirely on equity. The colossal numbers in the financial report are estimates of the fair value of stock awards on the grant date according to U.S. accounting standards, rather than actual liquid assets on the books. Thus, there is a significant gap between the reported data and his actual disposable wealth.
Breaking Down the $158.3 Billion Performance Bonus Structure
This $158.3 billion compensation is primarily divided into two main components. The largest portion is the '2025 CEO Performance Bonus' worth approximately $132.3 billion. This plan requires Tesla to achieve a series of challenging operational metrics and market cap targets to gradually unlock shares; as of the end of 2025, no shares have met the vesting conditions. The other part is the '2025 CEO Temporary Bonus' valued at around $26.1 billion, initially designed to ensure the CEO's retention during the legal turmoil of the 2018 pay case. However, the Delaware Supreme Court overturned the initial ruling in December 2025, officially reinstating Musk's 2018 performance bonus plan. Based on this ruling, Tesla's board committee confirmed that the originally proposed '2025 Temporary Bonus' is no longer necessary. This $26.1 billion equity bonus was completely forfeited on April 21, 2026.
High-Risk, High-Reward Link to Company Valuation
Tesla's compensation design demonstrates extreme high-risk, high-reward characteristics. The CEO's wealth growth is entirely tied to the company's market cap and profit performance; if the stringent standards set by the board are not met, the book value of stocks will not convert into actual assets. While this mechanism can effectively motivate management to pursue the company's best interests, it may also encourage more aggressive operational strategies.
Recently, Reuters exposed a confidential IPO filing submitted by SpaceX to the SEC, revealing a compensation plan that sounds like a science fiction plot. Musk's pay is directly linked to Mars colonization and space data centers. If the company's market cap exceeds $7.5 trillion and a permanent human colony on Mars housing at least 1 million people is established, he will receive 200 million shares of super-voting restricted stock. (SpaceX Musk's 200 Million Share Compensation Plan Exposed! Must Transport a Million People to Mars to Cash In)
This article, Is Musk Working for Free at Tesla? Why Will His $158.3 Billion Salary Be Worthless in 2025?, first appeared on <a>...</a>.

Bitcoin Magazine is stepping into TV media, launching BM TV for daily live market discussions.

Bitcoin Magazine has announced the launch of a daily live show "BM TV" this summer, transforming into a comprehensive media platform. The show will be broadcast from its headquarters in Nashville, Tennessee, targeting global financial and professional tech audiences, providing real-time Bitcoin market analysis and macroeconomic commentary. BM TV's airing schedule aligns with the opening hours of the U.S. East Coast stock market. The show plans to go live from 9:30 AM to 11:30 AM EST, perfectly syncing with the U.S. market opening, showcasing its positioning as a professional financial information service. In terms of distribution, a cross-platform streaming strategy will be adopted, airing simultaneously on X, YouTube, Facebook, Rumble, LinkedIn, and the official website BitcoinMagazine.com. This move aims to reach the brand's existing 6 million fans and achieve a global exposure target of over 1 billion. Through a diverse platform layout, the company seeks to ensure that information related to digital assets can be steadily disseminated to different audience groups in an algorithm-driven content environment. The program's content focuses on fintech. BM TV will feature two hours of content daily, hosted live by anchors, complemented by in-depth discussions from analysts. The reporting scope will focus on Bitcoin (BTC), global markets, macroeconomic policies, geopolitics, and cutting-edge technologies like energy systems and artificial intelligence. To enhance the timeliness of information, the show will integrate price trend charts, Bitcoin spot ETF flow data, market forecast indicators, and real-time quotes. Additionally, the platform plans to incorporate remote interviews, inviting experts from finance, energy, and public policy to provide diverse perspective analyses, turning scattered data into contextual professional commentary. Bitcoin Magazine is operated by BTC Inc., a subsidiary of the Satoshi Act Constellation group (NASDAQ: NAKA). CEO Brandon Green states that Bitcoin has moved from the fringe to the mainstream, and the relevant media infrastructure must evolve accordingly. The company's business landscape now encompasses print publications, Bitcoin Conferences, and corporate consulting projects. The establishment of BM TV represents its transition from a singular publishing entity to a large-scale media service provider. Executive producer Spencer Nichols notes that Bitcoin participants have shifted from early retail investors to institutional ones, and with the rise of Bitcoin ETFs, the market's demand for detailed and globally-minded TV analysis has significantly grown. Media head Mark Mason states that BM TV's target audience includes asset allocators, tech developers, and policymakers. The channel aims to maintain the Cypherpunk spirit established since its inception in 2012 while also providing professional analysis that meets the current financial system's demands. BM TV plans to produce over 200 episodes annually. Each episode will generate short videos, news briefs, and editorial articles on BitcoinMagazine.com. This article Bitcoin Magazine is stepping into TV media, launching BM TV for daily live market discussions first appeared on.

ECB and BoE Hold Steady on 4/30: Eurozone and UK Central Banks Take a Wait-and-See Approach to Inflation from the Iran War

The European Central Bank (ECB) and the Bank of England (BoE) both announced on April 30 that they would keep their rates unchanged. The ECB maintained the deposit facility rate at 2%, marking the third consecutive meeting without a change; while the BoE voted 8 to 1 to keep the benchmark rate at 3.75%, with only Chief Economist Huw Pill advocating for a hike to 4%. The decisions reflect a shared structural challenge: inflation driven by the Iran war’s energy crisis has exceeded the central banks' policy targets, yet slowing growth makes rate hikes untenable—prompting discussions about a 'stagflationary dilemma' for the central banks. ECB: Inflation Surges to 3% Yet Rates Unchanged, 'Risks of Upward Inflation and Downward Growth Have Intensified' The Eurozone's April flash CPI data was released on the same day, showing a year-on-year increase to 3%, significantly above the ECB's 2% target, and accelerating from March. Nevertheless, the ECB chose to keep the deposit facility rate at 2%, stating in its announcement that 'risks of upward inflation and downward growth have intensified', acknowledging a policy dilemma caught between two pressures. The ECB made no pre-commitment regarding a rate hike in the June meeting, maintaining a 'data-dependent' stance. This leaves room for subsequent May Eurozone CPI and June unemployment data to dictate the path ahead. The market had initially expected the ECB might lower rates to 1.75% in the second half of the year, but this latest decision and inflation data suggest a recalibration of rate cut expectations. BoE: 8 to 1 Vote Holds Steady at 3.75%, Pill the Sole Hawk The BoE's Monetary Policy Committee (MPC) voted 8 to 1 to hold the benchmark rate at 3.75%. The only vote advocating for a hike to 4% came from Chief Economist Huw Pill. The MPC's minutes noted, 'CPI inflation has risen to 3.3% and may go higher later this year, as the impact of rising energy prices continues to transmit.' In its policy statement, the BoE made a more direct comment on the Iran war's impact: 'The Middle East conflict means that the global energy price outlook is highly uncertain. Monetary policy cannot influence energy prices, but will be set to ensure that the economy adjusts sustainably towards achieving the 2% inflation target.' This wording closely acknowledges that 'the BoE's policy tools cannot directly alleviate energy shock inflation', opting to manage the current environment through aggregate demand rather than interest rate shocks. Eurozone and UK in Sync: Central Bank Policies Limited by Geopolitical Shocks, Future Path Depends on War Outcomes The choices of the ECB and BoE jointly illustrate one thing—when inflation is primarily driven by geopolitical shocks (the Iran war pushing up energy prices) rather than overheating aggregate demand, traditional policy tools of central banks are limited. The Fed also chose to hold steady on 4/29 at 3.5-3.75%, aligning the G7 major central banks in a consistent stance: delaying the rate cut cycle comprehensively until the situation in the Iran war and the Strait of Hormuz becomes clearer, avoiding premature easing that could further spread energy-driven inflation. For the market, this means three expectations need recalibration: (1) Rate cut expectations for H2 2026 are pushed back to early 2027; (2) The relative strength of the euro, pound, and dollar stabilizes due to the central banks' synchronized wait-and-see approach, with exchange rate momentum primarily driven by the war's outcome and safe-haven capital flows; (3) Long-end bond yields are supported by inflation expectations, making rapid declines difficult, which continues to pressure tech stock valuations and corporate financing costs. The next key milestone is the June meeting cycle (ECB 6/4, BoE 6/18, Fed 6/17-18), where all three central banks will face the ultimate judgment of whether the 'timing for a rate cut is ripe'. This article 'ECB and BoE Hold Steady on 4/30: Eurozone and UK Central Banks Take a Wait-and-See Approach to Inflation from the Iran War' first appeared in .

Meta Teams Up with Stripe to Roll Out USDC Payments for Creators, Supporting Solana and Polygon Networks

Meta has announced a collaboration with the well-known payment platform Stripe to launch a stablecoin payment feature for content creators. This new service supports the USDC stablecoin issued by Circle, providing creators with an income option beyond traditional fiat currency, with the initial trial focused on specific creators in Colombia and the Philippines. Meta's First USDC Payments Launch in Colombia and the Philippines According to Meta's official support page, the option to receive USDC stablecoin as income is currently open to select creators in Colombia and the Philippines. This payment feature is primarily built on the Solana and Polygon blockchain networks. Creators must ensure that their cryptocurrency wallets support one of these networks and can receive USDC to successfully collect payments. Support for Multiple Mainstream Wallets, with Official Reminders on Risks and Security Responsibilities To provide a convenient payment experience, Meta suggests creators use various digital wallets like MetaMask, Phantom, and Binance. However, the officials have also issued a disclaimer, stressing that creators must take responsibility for the security of their account information and wallet keys. Meta warns that only wallets supporting USDC on the Solana or Polygon networks should be used; funds sent to unsupported addresses or networks cannot be recovered. Additionally, if technical difficulties or unforeseen circumstances arise, Meta reserves the right to switch to alternative payment methods at the creator's discretion. Meta Returns to Stablecoin Payments, No Support for USDC to Fiat Conversion Looking back, Meta, which boasts billions of users across Facebook, Instagram, and WhatsApp, previously pushed a digital asset payment project called Libra (later renamed Diem) but had to shut it down due to intense global regulatory scrutiny. Now, through partnerships with third-party companies, Meta is finally making its way back into the stablecoin payment arena. However, if creators need to convert the received USDC into local fiat currency, Meta does not provide direct conversion services. Creators must first transfer USDC from their digital wallets to a cryptocurrency exchange that supports withdrawals in local currency; then they can exchange USDC for local currency within the exchange; finally, they can withdraw the funds to their personal bank accounts or preferred payment methods. This article Meta Teams Up with Stripe to Roll Out USDC Payments for Creators, Supporting Solana and Polygon Networks first appeared on .

Alphabet's earnings blow past expectations, GOOG jumps 6% to a new high

Google's parent company Alphabet released its Q1 earnings report, with revenue and profit performance surpassing market expectations. The key highlight of this report was the significant growth in its cloud computing segment, indicating that the company's massive investments in artificial intelligence (AI) infrastructure are starting to yield tangible revenue conversion benefits. Alphabet continues to ramp up capital expenditures, and buoyed by the stellar earnings report, Alphabet's stock (GOOG) rose 6% in after-hours trading to reach $370, setting a historic high. Alphabet's revenue exceeds expectations, strong growth in cloud business According to Bloomberg data, Alphabet's Q1 revenue, excluding partner revenue sharing, reached $94.7 billion, exceeding analysts' expectations of $91.6 billion. Earnings per share (EPS) stood at $5.11, significantly surpassing Wall Street's estimate of $2.62, reflecting strong operational efficiency and pricing power. At the same time, the company announced a 5% increase in dividends, with quarterly cash dividends reaching $0.22. These figures objectively demonstrate that the company maintains a robust financial structure and profit momentum amidst the current macroeconomic environment. Alphabet's cloud computing division reported sales of $20 billion in Q1, surpassing the expected $18.4 billion. This growth was primarily driven by strong demand for AI software and infrastructure. Notably, its backlog (the value of contracts signed but not yet recognized as revenue) doubled from the previous quarter, exceeding $460 billion. To expand its market, the company will provide Tensor Processing Units (TPUs) to select customers for use in its own data centers, showcasing strong enterprise demand for AI computational infrastructure. Alphabet continues to ramp up capital expenditures To maintain its technological edge in the AI space, Alphabet expects this year's capital expenditures (the funds allocated for the acquisition or upgrading of physical assets) to be revised upward from the initially estimated $175 billion to $185 billion, reaching double last year's scale. CFO Anat Ashkenazi also stated that the company's capital expenditures in 2027 are expected to increase significantly compared to 2026. In terms of industry competition, the company is in a complex competitive relationship with OpenAI and Anthropic. Despite committing to invest up to $40 billion in Anthropic, internal competition from AI code development products (like Claude Code) highlights the high-pressure environment of technology R&D in the AI industry. Growth in Gemini paid users, search engine queries hit new highs With the deep integration of AI technology, the number of paid monthly active users for Gemini Enterprise grew by 40% compared to the previous quarter. The search engine's query volume also reached a historic high. The company has begun providing AI-generated answers in many search results, fundamentally transforming traditional keyword and search advertising business models. From a macroeconomic and digital marketing perspective, this trend will have profound implications for businesses heavily reliant on Google traffic for profit. GOOG rose 6% to $370 in after-hours trading, marking a historic high Additionally, Alphabet's reevaluation of private investment values in companies like SpaceX has positively contributed to its overall profit. Driven by the impressive earnings report, Alphabet's stock (GOOG) surged nearly 7% in after-hours trading to $370, setting a new all-time high. Its performance this month has outpaced other seven tech giants, rising by 26%. This article Alphabet's earnings blow past expectations, GOOG jumps 6% to a new high first appeared on <a>...</a>.

Sui DEX Aftermath Hacked: 1.14 Million USDC Withdrawn

On April 29, Sui Network's mainstream DEX Aftermath Finance suffered a breach in its perpetuals product, with attackers pulling out approximately 1.14 million USD in USDC through 11 transactions over a span of about 36 minutes. The incident was detected and reported in real-time by blockchain security firm Blockaid, which announced it on the X platform. Currently, Aftermath has paused the affected products. Attack Details: 36 Minutes, 11 Transactions, 1.14 Million USDC The wallet address of the identified attacker, as mentioned in Blockaid's announcement, is 0x1a65086c85114c1a3f8dc74140115c6e18438d48d33a21fd112311561112d41e. The execution speed of the attack was extremely quick— from the first abnormal transaction to the completion of the eleventh transaction, the whole process took only about 36 minutes. The withdrawn amount of 1.14 million USD in USDC came entirely from the Aftermath perpetuals product's protocol treasury. Aftermath Finance is one of the largest decentralized exchanges on the Sui Network. Before the attack, its perpetuals product accounted for about 1/8 of the overall trading volume on the Sui chain and approximately 12% of gas consumption. In other words, this incident affected not only the Aftermath protocol but also created a brief vacuum in on-chain activity across the entire Sui ecosystem. Vulnerability Mechanism: Builder Code Fees Can Be Set to Negative, Inflating Synthetic Collateral According to subsequent explanations from Blockaid and Aftermath, the vulnerability was not in the core smart contract of Perps itself but in the accounting logic of the “builder code fees”—the protocol mistakenly allowed this fee rate to be set to a negative value. The attacker exploited this to artificially inflate synthetic collateral and then withdrew excess funds from the protocol treasury based on this inflated collateral. This type of vulnerability is a common issue in DeFi protocols known as “lack of boundary checks on state variables”: developers assumed that the fee variable would always be positive when designing it and did not implement safeguards against “negative value scenarios.” Once attackers find a way to push the variable into the negative range, unexpected mathematical operations can occur—resulting in the false inflation of synthetic collateral in this case. Aftermath's Response: Perpetuals Paused, Swap/Staking/MEV Unaffected The Aftermath team immediately paused the perpetuals product upon confirming the incident, emphasizing that this vulnerability “is limited to perpetual products; other services including swap, staking, and MEV infrastructure were unaffected.” The team stated that their current focus is on “fund recovery”—but as of the time of this writing, Aftermath has not disclosed whether they have coordinated with centralized exchanges to freeze the funds that were hacked, nor have they announced a specific mechanism and timeline for user compensation. Aftermath advises users who still hold open positions to check their wallet activity and closely follow the team’s subsequent official updates. For the Sui ecosystem, subsequent observation points include: whether other Sui perpetual DEXs (such as Bluefin, FlowX) will absorb liquidity during Aftermath’s pause, whether the hacked funds can be successfully frozen or recovered, and whether Aftermath will release a comprehensive security audit report before resuming operations. 4/29 22:15 (Taiwan Time) Update: All Users Will Be Fully Compensated About 3 hours after the incident was announced, Aftermath Finance released an update on the X platform, stating that with the support of Sui's parent company Mysten Labs and the Sui Foundation, “all users will be made whole, ZERO losses for anyone.” Aftermath also confirmed that the protocol would be back online soon and thanked Blockaid for their quick reporting. In the update, Aftermath clarified a key point: “This vulnerability is not a security issue at the Move contract language level.” This clarification addresses the discussion of Sui ecosystem security: Sui uses the Move programming language aimed at “resource safety,” often cited as a relatively safer contract environment compared to Solidity. Aftermath emphasized that this vulnerability is an application-level accounting logic design flaw (boundary handling of builder code fees), not a weakness at the Move language level, to prevent the market from extrapolating this incident as a conclusion of “Move/Sui being overall unsafe.” Details of the compensation mechanism from Mysten Labs and the Sui Foundation (source of funds, compensation timeline, whether using foundation reserves) have not yet been disclosed, but this swift and high-level response provides important confidence restoration for the Sui ecosystem's DeFi security image—compared to scenarios where users wait long-term for compensation after certain on-chain DeFi hack incidents, this processing speed is considered an active example in the industry. This article Sui DEX Aftermath Hacked: 1.14 Million USDC first appeared on <a>...</a>.

YC Unveils 15 New Startup Directions for Summer 2026: AI Entrepreneurship is More Than Just Stuffing a Chatbot into Products

Y Combinator (YC) recently released its Summer 2026 Requests for Startups (RFS), outlining the directions they are particularly keen for entrepreneurs to dive into this season. From YC's proposals, it's clear that entrepreneurial opportunities are no longer just about "adding a chatbot to existing products" or creating another AI tool for writing emails, organizing meetings, or generating presentations. YC is now more focused on how AI can transform complex systems like internal company knowledge, professional service delivery, semiconductor supply chains, hardware manufacturing, agriculture, healthcare, space electronics, and anti-drone defenses. In other words, AI entrepreneurship is shifting from "enhancing individual work efficiency" to "restructuring organizational and industrial processes." If a company is AI-native from day one, what it sells might not just be a piece of software but a service reconfigured by AI, an enterprise operating system, or even a new supply chain capability. Is YC shifting from a leading indicator to a lagging one? Meng Xing, a partner at Five Sources Capital, recently released a Silicon Valley inspection report pointing out a crucial turning point in the current AI startup scene: YC, once considered the bellwether for startups, may be transitioning from a "leading indicator" to a "lagging indicator" as the pace of AI iterations accelerates. (Is AI more about boosting output or cutting costs? A hundredfold efficiency hasn't translated into a hundredfold revenue, but no one in Silicon Valley dares to hit the brakes.) In March this year, while sitting in the audience at YC's W26 batch Demo Day, Meng Xing put down his pen after the fifth company's pitch. The reason wasn't a lack of effort from these companies, but rather that the topics were too similar. Among over a hundred companies this batch, around 80% are working on vertical agents, such as helping lawyers organize documents, assisting customer service with ticket distribution, and helping HR screen resumes. A year ago, these topics might have made investors think, "that's a neat idea." But by 2026, with Claude Code evolving from a developer tool to an interface nearly anyone can use, and Opus 4.6 further lowering the bar for vibe coding, many vertical agents lacking established business barriers are no longer as scarce as they once were. An average engineer could potentially replicate similar products over a weekend. This also challenges YC's batch system, which has been foundational to its success. The rhythm of YC's process—from application, selection, incubation, to Demo Day—is designed for a relatively stable world where product and market changes occur slowly. However, given the current rapid pace of AI iterations, five months can see several rounds of paradigm shifts. As model capabilities, development tools, and user habits are rapidly rewritten, by the time a startup topic enters the batch process, it may have shifted from cutting-edge to consensus, or even into a red ocean by the time of public pitching. Interestingly, YC's latest Requests for Startups (RFS) appears to be responding to this change. In the Summer 2026 RFS, YC explicitly states that AI has stopped being just a feature and has begun to serve as the foundation for software, services, hardware, and even the physical world. AI is no longer just a feature but the underlying assumption for companies and industries. YC emphasizes in the Summer 2026 RFS that AI has stopped being merely a function within products and has begun to lay the groundwork for software, services, silicon chips, and the physical world. This observation highlights the main theme of this list: AI entrepreneurship cannot just stop at "integrating models into products" but must rethink whether the original work, companies, and industries should be rewritten. Therefore, the topics in this RFS rarely reflect simple consumer app-style entrepreneurial imagination. More accurately, YC is not looking for the next AI application that goes viral based on UI and traffic but is focusing on areas that have historically been challenging to completely transform with software. For example, professional services previously heavily relied on human labor and process experience; corporate knowledge is scattered across Slack, emails, tickets, and meetings; semiconductor supply chains still largely depend on manual coordination; hardware manufacturing in the U.S. iterates much slower than in Shenzhen; agriculture still relies on widespread pesticide spraying; and drone defenses face structural issues where the attacker's costs are far lower than the defender's. These are not problems that can be solved by simply adding a chatbot. They require AI to be integrated into processes, data, hardware, supply chains, and decision-making systems, becoming part of the entire operational framework. From selling software to selling results: AI-native service companies. YC partner Gustaf Alströmer proposed the direction of "AI-Native Service Companies" in the RFS, which could be key to understanding this new wave of AI entrepreneurship shift. Over the past few years, most AI startups have been focused on copilot solutions, helping people complete tasks faster. They sell software, and users still need to operate tools, judge results, and complete deliveries. But YC is now interested in the next step: companies no longer just sell tools but directly sell services. This means customers don't need to buy a suite of AI software to train employees; instead, they can hand over tasks like insurance brokerage, accounting, tax, auditing, compliance, and healthcare administration to AI-native companies. Software here becomes an internal productivity tool rather than the main product sold externally. This shift is significant because the service market is far larger than the software market, and many professional services are already outsourced by companies. If AI-native companies can deliver work at lower costs, faster speeds, and more consistent quality, they are not just attacking SaaS but the entire cost structure of the service industry. Corporate Brain: The Missing Link in Enterprise AI Automation is Not Models but Internal Knowledge. Another key point is the "Company Brain." YC partner Tom Blomfield believes that enterprise AI automation is lacking not in models but in internal knowledge...

Global Accounts Based on the Bitcoin Network: Grid Integrates 65 Fiat Currencies and Cryptos for AI-Powered Payments

Current CEO of Lightspark and former PayPal President David Marcus introduced a new payment infrastructure "Grid Global Account" at the 2026 Bitcoin Conference. This product integrates fiat currencies, stablecoins, and Bitcoin within a single account, pioneering a mechanism that allows AI to assist in managing and executing payments, providing a fresh solution for cross-border transfers and corporate fund allocation. Grid Global Account: Supports Visa Card Usage and International Withdrawals. Previously, cross-border remittance fees were high and time-consuming. Since Lightspark has become a major player in Visa, users can directly use stablecoins or digital assets within their Grid account at 175 million locations worldwide that accept Visa, both in physical stores and online. The system supports local payment networks in over 65 countries. For instance, a freelancer in Mexico receiving USD payments from a US company can exchange the funds for Mexican Pesos anytime within 24 hours and instantly transfer to a local bank account. Additionally, the USD funds in the account can be converted into stablecoins on different blockchain networks like Solana or Optimism. Users only need one wallet address to store, receive, and trade USD stablecoins and Bitcoin. AI Agents for Shopping and Payments. In response to the interactive model of the AI era, the Grid account collaborates with the "Bread" wallet system to introduce an agent authorization feature that allows AI agents to utilize funds within a secure scope. Users can set clear spending limits and authorization ranges for the AI. The system works on the principle of "AI proposes plans, the system reviews rules, and the account is responsible for disbursing funds," ensuring that ultimate control over funds remains with the user. For example, users can instruct their AI assistant in messaging apps (like WhatsApp) to purchase coffee beans, and the AI will automatically generate a virtual card number for payment; or instruct the AI to transfer money to friends abroad, where the AI will obtain the recipient's local account details and complete the cross-border transfer. If the transaction amount exceeds the user-defined limit, the system will pause and display a notification, and it will only proceed after the user manually clicks to approve the transaction. Built on the Bitcoin Network's Underlying Financial Infrastructure. The Grid Global Account is not only open to the general public but also specifically designed as an underlying financial infrastructure for global enterprises and multinational platforms. For companies engaged in numerous cross-border transactions, the high fees and costs previously charged by traditional financial systems can turn into revenue for the platform after implementing the Grid account system, including derivative income from stablecoins, foreign exchange fees, or Visa card transaction fees. Furthermore, Lightspark has chosen to build this extensive system on the Bitcoin network, valuing its open and absolutely neutral characteristics, which ensures that the payment services constructed by enterprises will not be subjected to control by any single centralized entity or face the risk of arbitrary rule changes. This article Global Accounts Based on the Bitcoin Network: Grid Integrates 65 Fiat Currencies and Cryptos for AI-Powered Payments first appeared on <a>...</a>.

NVIDIA drops the Nemotron 3 Nano Omni open-source multimodal model

According to NVIDIA's official blog post on April 28 (by Kari Briski), NVIDIA has launched the Nemotron 3 Nano Omni — an open-source multimodal model that integrates visual, speech, and language capabilities into a single framework, aiming to provide a lower latency and cost-effective "perception layer" for AI agent systems. Core specs: 30B-A3B MoE, 256K context, 9x throughput, topping 6 leaderboard categories. Key architecture: 30B-A3B hybrid mixture-of-experts (total parameters 30B, activated 3B). Integrates Conv3D and EVS encoding with a 256K context length. Inputs: text, images, audio, video, documents, charts, GUI screens. Outputs: text. Performance signal: 9x throughput compared to other open-source omni models at the same interactivity level; secured the top position in 6 benchmark categories across document intelligence, video understanding, and audio comprehension (NVIDIA's announcement did not list specific scores, directing readers to the developer blog for details). NVIDIA positions the Nemotron 3 Nano Omni as the "eyes and ears" in agent systems, working alongside family models like the Nemotron 3 Super (high-frequency execution) and Nemotron 3 Ultra (complex planning), and is also compatible with third-party cloud models. Three typical agent application scenarios: Computer Use Agent: native 1920×1080 resolution visual reasoning. Document intelligence: reasoning across images, tables, screenshots, and mixed media inputs. Audio/Video understanding: integrating speech, visuals, and recordings into a single reasoning stream. Adopting firms: Foxconn, Palantir on board, H Company CEO expressing support. NVIDIA's announcement clearly distinguishes between "production adoption" and "under evaluation": Production adoption: Aible, Applied Scientific Intelligence (ASI), Eka Care, Foxconn, H Company, Palantir, Pyler. Under evaluation: Amdocs, Dell, Docusign, Infosys, IQVIA, Lila, Oracle, Quantiphi, TCS, Zefr, etc. H Company CEO Gautier Cloix stated in the announcement: "To build useful agents, you can’t wait seconds for a model to interpret a screen. By building on Nemotron 3 Nano Omni, our agents can rapidly interpret full HD screen recordings — something that wasn’t practical before." Translation: "When creating useful agents, you can't wait seconds for a model to interpret a screen. Built on Nemotron 3 Nano Omni, our agents can quickly interpret full HD screen recordings — which wasn’t feasible before." Open-source strategy and deployment: weights / datasets / training methods made fully public. Along with the launch, NVIDIA made public: model weights, training datasets, training techniques/methodologies. Deployment pipeline covers three layers: local workstations: NVIDIA DGX Spark, DGX Station NIM microservices: build.nvidia.com third-party platforms: Hugging Face, OpenRouter, and over 25 NVIDIA Cloud Partners, inference platforms, and cloud service providers offer customized tools using NVIDIA NeMo. The Nemotron 3 family (Nano/Super/Ultra) has accumulated over 50 million downloads on Hugging Face in the past year, and this Omni extends the family’s capabilities into multimodal and agentic domains. This article NVIDIA announces Nemotron 3 Nano Omni open-source multimodal first appeared on .

Japan's 4 ministries joint request: Strict KYC and anti-money laundering for crypto asset real estate transactions

According to the official announcement released by the Japan Financial Services Agency (FSA) on April 28 (full PDF), a joint request was issued by four ministries: the FSA's Comprehensive Policy Bureau Risk Analysis Division, the Ministry of Land, Infrastructure, Transport and Tourism's Real Estate Industry Division, the National Police Agency's Criminal Organization Countermeasures Division, and the Ministry of Finance's International Bureau Investigation Division, to seven industry groups regarding "Real Estate Transactions Using Cryptocurrency (Request)" (for real estate transactions using crypto assets). The four ministries made requests to six real estate associations and JVCEA. The recipients of this request are seven industry groups: Japan Real Estate Transaction Association Federation, All Japan Real Estate Association, Real Estate Association, Real Estate Distribution Management Association, National Housing Industry Association, Real Estate Distribution Promotion Center, and the Japan Cryptocurrency Exchange Association (JVCEA). The background explanation points out that real estate properties handled by real estate agents have high value and can be exchanged for large amounts of cash, posing a risk of being used for money laundering and other crimes. In recent years, there has been an increase in real estate purchases for asset preservation and investment purposes, and both domestic and foreign criminal organizations may use real estate transactions to convert criminal proceeds. The document specifically notes that "the characteristic of cryptocurrency allowing for instant cross-border transfer is deemed to carry a high risk of being used for money laundering as a settlement method in real estate transactions." Specific requirements for real estate operators: Strict adherence to the Payment Services Act and the Act on Prevention of Transfer of Criminal Proceeds. The specific requirements for real estate operators (real estate agents) include two points: 1. The act of exchanging cryptocurrency for fiat currency may constitute a cryptocurrency exchange business. It is advised to be cautious of actions such as "exchanging cryptocurrency for fiat currency or acting as an intermediary for exchanges," which may fall under the cryptocurrency exchange business as per the Payment Services Act. Engaging in such activities without registration poses a risk of violating the Payment Services Act. If any suspicious unregistered activities related to cryptocurrency exchange business are discovered, information should be provided to the police authorities; real estate operators should not utilize unregistered cryptocurrency exchanges when receiving cryptocurrency as payment for sales and subsequently exchanging it for fiat currency, as these actions do not fall under cryptocurrency exchange business. 2. Strict transaction confirmations per the Act on Prevention of Transfer of Criminal Proceeds (Law No. 22 of 2007). When conducting cryptocurrency real estate transactions, strict KYC must be executed, and a "suspicious transaction report" must be submitted to the relevant administrative authority. In case of any event-related concerns, the police authorities must be notified. Requirements for cryptocurrency exchange operators: High-value transactions not matching the customer's profile must raise red flags. Specific scenarios for cryptocurrency exchange operators include: In cases where "customers receive payment for real estate sales in cryptocurrency and conduct high-value transactions inconsistent with the customer's profile," strict transaction confirmations must be conducted per the Act on Prevention of Transfer of Criminal Proceeds, and suspicious transaction reports must be submitted to the relevant administrative authority, along with notifying the police authorities of any event-related concerns. The Foreign Exchange Act reiterates: Transactions involving cryptocurrency exceeding 30 million yen and acquiring real estate from non-residents must be reported. The document emphasizes two reporting obligations under the Foreign Exchange and Foreign Trade Act (Law No. 228 of 1949, referred to as the "Foreign Exchange Act"): Any individual receiving cryptocurrency equivalent to 30 million yen or more from abroad must submit a "Report on Payments or Receipt of Payments". When non-residents acquire real estate in Japan, they must submit a "Report on Acquisition of Real Estate or Rights Related to Real Estate Located in Japan". It is particularly noted that real estate acquired in Japan after April 1, 2026 (Reiwa 8) is subject to reporting regardless of the acquisition purpose. This request is classified as "administrative guidance" (request) level and not new legislation or amendments. However, the joint signatures from the four ministries and simultaneous dispatch to both the real estate and cryptocurrency industries indicate Japan's concrete identification of the AML risks associated with "real estate transactions using cryptocurrency as a settlement method," making it worth monitoring for future enforcement and industry compliance trends. This article on Japan's 4 ministries joint request: Strict KYC and anti-money laundering for crypto asset real estate transactions first appeared on.

Google signs a secret AI contract with the Pentagon, hundreds of employees' petitions fall on deaf ears

According to reports, Google has signed an agreement with the U.S. government, allowing the Department of Defense (now renamed the Department of War by Trump) to use its AI models for classified missions. This move positions Google alongside tech giants like OpenAI and xAI, officially making it a supplier of classified AI tech for the U.S. military. (OpenAI snags Pentagon AI contract! Sam Altman admits to 'hasty decision', emphasizes no involvement in mass surveillance or autonomous weapons) Google's AI models will be deployed in defense missions. The Information reports that this contract allows the Pentagon to use Google's AI technology for 'any lawful government purpose.' These classified networks are typically used for handling highly sensitive military tasks, including mission planning and weapon targeting. A Google spokesperson stated that this new agreement is an amendment to existing contracts. AI weaponization and security regulations: Google has no authority to veto military decisions. In response to concerns about AI weaponization, the contract explicitly states that the AI system 'is not intended to be used for domestic mass surveillance or autonomous weapons (including target selection) without proper human oversight and control.' Additionally, the contract includes a key clause that clarifies that the agreement 'does not grant any rights to control or veto lawful government combat decisions.' Furthermore, the agreement requires Google to comply with U.S. government requests, assisting in adjusting corporate AI safety settings and filtering mechanisms. (ChatGPT uninstall numbers skyrocketing threefold, OpenAI's Pentagon cooperation raises cybersecurity concerns) Over 600 Google employees petition against the agreement. Reports indicate that more than 600 Google employees (including members of its top AI lab, DeepMind) have signed a letter to CEO Sundar Pichai, strongly urging the company to reject any classified AI projects with the military. In the letter, employees emphasize: 'We want to see AI benefit humanity, not be used in inhumane or extremely harmful ways, including lethal autonomous weapons and mass surveillance.' Google has not immediately responded to the situation, clearly indicating that despite strong internal opposition, this collaboration with the U.S. military has ultimately been finalized. (When AI becomes a nuclear-grade weapon: Ben Thompson discusses Anthropic and Pentagon conflict) This article Google signs a secret AI contract with the Pentagon, hundreds of employees' petitions fall on deaf ears first appeared on .

OpenAI ChatGPT misses revenue targets; CFO admits they might not cover computing costs

With the growth of ChatGPT users slowing down and facing pressure from both Anthropic and Google, OpenAI is encountering unprecedented financial stress just as it rushes towards an IPO. (Before burning through hundreds of billions, can OpenAI turn ChatGPT into a cash cow?) OpenAI raises alarms: both user and revenue targets are likely to fall short by 2025. According to the Wall Street Journal, OpenAI has failed to meet its internal target of over 1 billion weekly active users for ChatGPT by 2025, as well as its undisclosed annual revenue target. CFO Sarah Friar has expressed concerns to other executives that if revenue growth doesn’t pick up, OpenAI may struggle to pay the costs of signed data center contracts. The board has also been scrutinizing the company’s computing power procurement agreements, questioning CEO Sam Altman’s strategy of continuing to expand computing capacity amid slowing business. The slowdown in ChatGPT's growth is not a recent issue. Late last year, Google Gemini emerged rapidly, capturing a significant market share from OpenAI; this year, Anthropic's competitiveness in programming development and enterprise markets has continued to rise, leading OpenAI to miss monthly revenue targets for several consecutive months. The loss of subscription users has become a pressing concern for management. The $600 billion computing gamble starts to waver. Altman has long believed that the shortage of computing power is the biggest bottleneck for OpenAI's growth, and last year signed contracts committing future expenditures totaling approximately $600 billion. However, as ChatGPT's growth momentum slows, this 'computing power positioning strategy' has begun to spark internal discussions. Friar and other executives are actively pushing for cost control and financial discipline, diverging from Altman’s expansion plan. In response, Altman and Friar issued a joint statement emphasizing that the two are 'working together every day to secure as much computing power as possible, and our positions are completely aligned.' Currently, the company is cutting back on non-core projects like Sora to control expenses while continuing to bet on the rapid growth of its programming development tool Codex, recently launching the top-performing GPT-5.5 in several benchmark tests. Can the largest financing in history raise enough cash? The IPO timeline is full of variables. In March of this year, OpenAI completed the largest financing round in Silicon Valley history, raising $122 billion led by Amazon, Nvidia, and SoftBank, with the company valued at $852 billion. However, insiders reveal that this funding is expected to run out within the next three years before achieving the revenue targets. Not to mention that part of the funding comes with conditions depending on whether specific collaboration agreements are fulfilled. Regarding the IPO plans by the end of the year, Friar remains cautious, believing OpenAI has yet to meet the strict financial reporting standards required for publicly listed companies; however, Altman leans towards aggressively pushing the IPO timeline. With Elon Musk’s lawsuit against OpenAI and Altman also officially opening this week, the company is currently burning the candle at both ends. Under the quadruple pressure of growth bottlenecks, intensified competition, internal disagreements, and legal lawsuits, whether OpenAI can successfully ring the IPO bell by the end of the year remains uncertain. This article OpenAI ChatGPT misses revenue targets; CFO admits they might not cover computing costs first appeared on <a>...</a>.

E-Cash.org might be the initial version before Satoshi Nakamoto unveiled 'Bitcoin'.

Bitcoin history researchers have found that e-cash.org was registered 29 days earlier than bitcoin.org. Records show that this domain aligns closely with Satoshi Nakamoto's development footprint before the whitepaper was published in 2008. It's speculated that 'e-cash' was likely Satoshi's preliminary naming scheme for 'Bitcoin', representing his early choice of asset and domain name. Ultimately, he opted to use Bitcoin (BTC) as the name for digital cash, discarding the early 'e-cash' version. The registration timeline of e-cash.org perfectly matches the drafting period of the whitepaper. According to WHOIS historical databases, e-cash.org was registered on July 20, 2008, through Dynadot, while the well-known bitcoin.org was only registered on August 18 of the same year. This 29-day gap coincided with a critical phase where Satoshi frequently modified the titles of his papers and technical documents. Bitcoin historian Gwern Branwen pointed out that when Satoshi communicated with cryptographer Wei Dai in August 2008, the private draft he shared was titled 'ecash.pdf', and the download URL pointed to 'ecash-pdf.html'. This indicates that before the term 'Bitcoin' was officially settled, Satoshi was highly inclined to use 'e-cash' as the system's name. Analyzing the technical characteristics of this domain, e-cash.org has consistently concealed its registrant information through privacy protection services since its registration in 2008, which aligns with the anonymous service (Anonymous Speech) that Satoshi used for registering bitcoin.org. As of the latest query record on April 25, 2026, the domain status remains Client Transfer Prohibited, and its validity has been renewed until July 20, 2028. Although this domain has never published content in the past 18 years and has not publicly associated with any commercial entities, the ongoing renewal actions indicate that its holder deliberately retains this historically significant virtual asset. (Hashcash inventor Adam Back discusses crypto-punk, Bitcoin, spam, and anonymous system design) Historical literature indirectly supports the early naming transition of 'Bitcoin'. On October 31, 2008, when Satoshi officially published the whitepaper to the Cryptography Mailing List, he ultimately titled it Bitcoin: A Peer-to-Peer Electronic Cash Paper. However, the concept of 'electronic cash' consistently ran through the document before its formal public release. Wayback Machine's archived records confirm that e-cash.org has remained blank from the start, which aligns with the behavior pattern of developers shelving early test domains after finalizing the brand name. Bitcoin historians believe that the transition from 'e-cash' to 'Bitcoin' is a crucial part of blockchain research. Despite the high correlation in the timeline and behavior patterns, there are currently no digital signatures or encrypted emails that can directly confirm e-cash.org as Satoshi's personal holding. All inferences are based on indirect evidence, including the registration timing, privacy protection preferences, and the correspondence with early draft document naming. Since the domain is still under strict privacy service protection, the true identity of its behind-the-scenes holder remains a mystery. This 'high probability but unverified' state reflects the nature of Bitcoin's early anonymous development and provides researchers with highly valuable temporal clues when tracing the origins of digital currency. This article 'E-Cash.org might be the initial version before Satoshi Nakamoto unveiled Bitcoin' first appeared on.

MicroStrategy Adds 3,273 BTC: Average Price $77,906, Total Holding 818,334 BTC

According to a press release from Strategy Inc. on April 27 and a Form 8-K filed with the SEC on the same day, the company purchased an additional 3,273 bitcoins (BTC) last week (April 20 to April 26), bringing its total holdings to 818,334 BTC, all funded through $255 million raised via ATM (at-the-market) stock issuance. Between April 20-26, they added 3,273 BTC at an average price of $77,906 per coin. In comparison, the spot price of Bitcoin during that period was in the range of $75,000 to $79,400, with this addition made around the midpoint of that range, indicating that Strategy is maintaining a 'mechanical weekly entry' strategy rather than timing the market. Total holdings of 818,334 BTC, average cost $75,537, total investment $61.81 billion. As of April 26, Strategy holds a total of 818,334 bitcoins, with a total investment amount of $61.81 billion (including fees), and an average purchase cost of $75,537 per coin. With BTC priced at $79,000 at the end of April, this asset's market value is approximately $64.6 billion, yielding unrealized profits of about $2.8 billion (around 4.5%). In terms of holding scale, Strategy is still about 180,000 BTC short of surpassing Satoshi Nakamoto's estimated position as the 'largest individual holder' at 1 million BTC—at the current pace of adding 3,273 BTC per week, it would take an estimated 50-60 weeks, or about 12-14 months, to reach that target. Funding Source: ATM Sale of 1,451,601 MSTR Shares, Netting $255 Million. This week's purchase funds came entirely from ATM issuance of MSTR Class A common stock (NASDAQ: MSTR), selling a total of 1,451,601 shares and netting $255 million. Aside from MSTR, there were no records of issuance for the four special stocks (STRF, STRC, STRK, STRD) this week. For MSTR shareholders, the weekly ATM issuance represents a dual track of 'equity dilution' and 'BTC holding growth'—as long as BTC's price increase offsets the dilution, the BTC per share ratio can still grow. Saylor has previously stated that if BTC's annual increase exceeds 2%, he can cover the special stock dividends, and the current model still operates within this range. Remaining $53.8 Billion Issuance Capacity for Five Securities. Currently available for future issuances, according to the 8-K disclosure: MSTR common stock: $26.47 billion (including the $21 billion increase from March 23), STRC (Variable Rate Stretch preferred stock): $19.46 billion, STRD (10% Stride preferred stock): $4.01 billion, STRK (8% Strike preferred stock): $2.1 billion, STRF (10% Strife preferred stock): $1.62 billion. The total remaining issuance capacity is approximately $53.7 billion, meaning that even if BTC does not rise further, Strategy still has over 2 years of 'ammunition' to continue adding. However, the actual issuance pace depends on market acceptance—if MSTR's stock price falls or the special stock's ARR risk premium increases, available space will be compressed. Implications for the BTC Market: Stable Buying Pressure, No Longer a Marginal Pricing Force. From 2024-2025, Strategy became a marginal buyer of Bitcoin due to a surge of over 30,000 BTC in a single week, but starting in Q1 2026, the buying pace has clearly slowed: this week, they added 3,273 BTC compared to last week's 34,164 BTC (April 13-19), a significant reduction. Possible reasons include: (1) ATM issuance pace influenced by MSTR stock price, leading to a rise in dilution costs when the stock price falls; (2) ETFs and other companies (Metaplanet, BitMine, etc.) beginning to share institutional buying pressure; (3) after BTC moved from the $73K to $79K range, Strategy is in no rush to chase prices. For the market, Strategy is shifting from a 'marginal pricing force' back to a 'stable weekly buyer' role—while still suppressing short-term volatility in BTC, they are no longer the sole driving factor. The next observation point is whether the weekly pace will return to over 10,000 BTC at the beginning of May. This article MicroStrategy Adds 3,273 BTC: Average Price $77,906, Total Holding 818,334 BTC first appeared on <a>...</a>.

Why do I say that the USVC launched by Naval is a bit like a financial version of the Pi Yao coin?

If you've recently scrolled through English financial Twitter or Chinese crypto media, it's hard to miss the ads for USVC. The marketing message for this fund is pretty straightforward: launched by AngelList, overseen by Naval Ravikant, with a minimum investment of $500 and zero performance fees, allowing regular folks to invest in companies like xAI, Anthropic, and OpenAI that were previously exclusive to Silicon Valley insiders. It's a clever narrative. Private companies have indeed been the biggest wealth growth arena over the past decade: in 1980, the median age for companies to go public was 6 years, and by 2024, it's pushed back to 13 years; retail investors typically have to wait until after the IPO to enter, and usually, the most lucrative stage has already passed by then. USVC packages this structural unfairness as a product that can be solved with just $500. However, after reading the 60-page prospectus submitted to the SEC and cross-referencing the sharp questions already emerging in the community, you'll realize that while this product isn't illegal, the gap between the marketing language and legal disclosures is significant enough to warrant a pause for anyone ready to hit the 'invest' button. "by AngelList" doesn't mean what you think it does. The USVC website starts with "a new fund by AngelList." But the investment advisor for USVC isn't AngelList itself; it's a subsidiary called AngelList Asset Management, LLC. This company was originally named Strawberry Tree Management Company LLC and didn't officially change its name until November 20, 2025. This firm was only established in December 2023 and registered with the SEC in March 2024, currently managing approximately $329 million in assets, and has never managed a closed-end fund registered under the 1940 Act. The prospectus emphasizes this point in several paragraphs because it's a legally required disclosure of "significant risk." In other words, you might think you're backed by the AngelList that has been running for over a decade and where Naval invested in Uber and Twitter. In reality, the company handling your money is a two-year-old firm that just recently adopted the AngelList name. Naval does serve as the chairman of the investment committee, but day-to-day investment decisions are made by Portfolio Manager Ankur Nagpal. The "zero performance fee" sounds great, but it's just a switch for something worse. The biggest selling point of USVC is "no carry." Traditional VCs take 20% of the profits, but USVC doesn't, so your returns are your returns. It does sound like a win for consumers. However, carry is not just a cost item; it's an incentive alignment mechanism. GPs only get paid when investors make money, so GPs are motivated to make the right investment decisions, reject bad deals, and control risks. USVC removes carry, which seems cheaper, but what replaces it? There's a section on page 37 of the prospectus that's easily overlooked: in the deal where Carry Technologies was acquired by AL Venture, LLC, Nagpal expects to receive consideration as a common stockholder of Carry, which includes contingent consideration linked to the growth of USVC's asset size. In layman's terms: how much money the Portfolio Manager can pocket directly depends on how large USVC's AUM can grow—not how well it performs, but how big it gets. This incentive direction is the opposite of traditional carry. Traditional carry motivates GPs to make money and avoid screwing up; AUM-linked compensation incentivizes managers to grow the size and not scare off investors. For a fund that's 65% cash, 20% bet on one company like xAI, with valuations determined by the manager themselves, this incentive structure particularly needs attention. The real version of a 1% management fee is 2.5%; without fee waivers, it's 3.61%. User gemchanger was the first to finish the prospectus and pointed out that although USVC's homepage touts "1% management fee, 0% carry," the actual total expense ratio isn't low; the so-called "0% carry" seems more like USVC itself not taking a performance cut, rather than the overall structure being truly carry-free. This criticism is entirely valid. The fee table in the prospectus looks like this: Item Ratio Management Fee 1.00% Shareholder Service Fee 0.25% Underlying Fund Fees (AFFE) 0.95% Other Fees 1.41% Gross Expense Ratio 3.61% Investment Advisor Waivers and Subsidies -1.11% Net Expense Ratio 2.50%. USVC does indeed only charge 1%. But your money will be invested into downstream Investment Vehicles and SPVs, which each charge management fees of 1% to 2.5% and also take 20% to 30% carry. These costs are reflected in your total cost through AFFE (Acquired Fund Fees and Expenses). USVC isn't misleading you; its contract literally states 1%. But you think you've escaped carry, while in reality, it's just moved from USVC's own level to the downstream fund level it invests in. You're still paying carry, just to someone else. More realistically, the prospectus reveals that if you invest $1,000 with an annual return of 5%, the total accumulated fees over 10 years would be $399. That’s nearly 40% of accumulated costs. And the 2.5% net expense ratio is still conditional. It relies on two waiver agreements signed by the investment advisor, which only last until October 2026...

MediaTek lands a major deal with Google for the 8th generation TPU! ASIC hype boosts three concept stocks.

The semiconductor design giant MediaTek (2454) has recently been reported to successfully penetrate Google's 8th generation TPU training chip supply chain. Not only are they involved in the core I/O Die design, but they are also implementing advanced processes and high-end packaging technologies, marking their formal entry into the high-end AI ASIC battlefield. With the shipping scale and demand heating up, Taiwanese supply chain players like King Yuan Electronics (2449), Sigmastar (6510), and Hon Hai (7769) are expected to ride the wave of new growth momentum. (MediaTek's stock price breaks through the 2,000 NTD mark, hitting the limit up, with the earnings call countdown igniting AI ASIC themes.) MediaTek's entry into Google's 8th generation TPU training chip supply chain has been highlighted by the Commercial Times, which states that Google's latest TPU architecture, the 8th generation, is divided into the TPU 8t for training and the TPU 8i for inference, with the training chip requiring support for large AI model computations, significantly raising the technical bar. Supply chain players reveal that MediaTek has targeted the TPU 8t, taking charge of I/O Die design and integration services, demonstrating their capability to support high-intensity AI training demands. In terms of processes and packaging, this project utilizes TSMC's N3P (3nm enhanced process) combined with the CoWoS-S advanced packaging solution, further boosting chip performance and integration density. Analysts indicate that being able to engage in such high-spec projects signifies that MediaTek's technical expertise has risen to the forefront of AI chip design. (Google Ironwood TPU: 10x performance + four partners to combat Nvidia) The ASIC business is becoming a new engine, with TPU shipping scale driving revenue growth. As major cloud firms continue to invest in AI infrastructure, the demand for customized ASIC chips is rapidly increasing. MediaTek has been actively positioning itself in the ASIC field in recent years, leveraging high-speed interfaces and system integration advantages to secure its place. The company expects its ASIC business revenue to exceed $1 billion this year, becoming one of the key growth drivers. The market further indicates that the production volume of Google TPU is continuously being revised upwards, with projections that by 2027, overall shipments could reach tens of millions of units. Whether orders are shared among MediaTek or other design firms, this will contribute to long-term stable revenues for the supply chain. Testing and equipment demand is heating up, and three ASIC concept stocks will benefit simultaneously. Due to the use of advanced processes and high-complexity packaging for AI training chips, the demands for testing precision and strength have significantly increased, driving rapid growth in related supply chain demand. Industry insiders point out that King Yuan Electronics and Sigmastar, which focus on high-performance computing testing, will directly benefit from the rising demand for AI chip testing. At the same time, testing equipment manufacturers like Hon Hai are also expected to benefit from the trend of expanded equipment investments. As chip prices rise and testing processes upgrade, system-level testing (SLT) is gradually becoming the standard procedure for high-end chips, further enhancing the importance of testing in the overall supply chain. Competition is shifting towards collaborative specialization: AI chips enter a new era of high-spec competition. In terms of competitive dynamics, the market has been watching whether Marvell Technology would penetrate the TPU supply chain, but there has yet to be any clear moves. Industry analyses suggest that as the difficulty of AI chip design continues to increase, future competition will shift from simply grabbing orders to collaboration, even developing new business models like 'Consign die'. In the long run, AI chips will evolve towards chiplet integration and heterogeneous integration, increasingly relying on advanced packaging technologies, thereby boosting TSMC's capacity utilization and the overall technical threshold of the semiconductor industry. In this trend, MediaTek's successful entry into the core of Google's training chips not only represents its own transformation achievements but also symbolizes the ongoing strengthening of Taiwan's supply chain in the global AI competition. This article MediaTek lands a major deal with Google for the 8th generation TPU! ASIC hype boosts three concept stocks. first appeared in.

Analysis suggests the DeFi TVL loss of $13 billion is due to a chain of liquidations, not actual losses.

Recently, DeFi has faced severe tests, with KelpDAO suffering a targeted attack on its infrastructure, leading to a loss of about $292 million in assets, subsequently triggering a significant drop in the market's total locked value (TVL) of around $13 billion. Despite the seemingly disastrous data, analysts point out that the primary cause of the capital reduction stems from the chain liquidations of leveraged positions and asset rotations. DeFi is not dead; it still demonstrates resilience under extreme pressure. The hacker's attack path has shifted from smart contracts to infrastructure. This KelpDAO attack differs from past incidents involving smart contract vulnerabilities, as it primarily targeted the LayerZero verification infrastructure. According to preliminary investigations, this incident is suspected to be related to the North Korean hacker group Lazarus Group. KelpDAO employed a single-validator architecture, and despite multiple prior recommendations for defensive configurations, its centralized structure led to its issued liquidity-staked token rsETH losing adequate collateral. This phenomenon triggered market concerns about bad debt spreading to large lending protocols like Aave's WETH pool. When the attack surface expands from the code level to the underlying infrastructure, the on-chain system's risk premium increases, forcing investors to reassess the potential costs of storing assets in decentralized protocols. Leverage amplifies price volatility. Data shows that DeFi's total locked value plummeted by $13 billion within 48 hours; however, this figure includes a large amount of duplicated virtual value. Taking Aave, the leader in the lending market, as an example, prior to the attack, users widely adopted looping strategies—depositing liquidity to re-stake tokens, borrowing ETH, and then exchanging it for more tokens to amplify rewards. This leverage effect boosts TVL during stable market conditions but can trigger a severe liquidation wave when collateral concerns arise. Although it appears that hundreds of millions have been lost on paper, the actual net capital loss is only a small fraction of that. Additionally, the current natural yield rates in the DeFi space are generally low, with Aave's USDC deposit annualized yield at about 2.61%, even lower than traditional financial institutions like Interactive Brokers at 3.14%. This has led developers to leverage to fill the reward gap, further exacerbating volatility during such events. Historical cases validate that DeFi has the capacity for recovery. Over the past few years, DeFi has faced multiple larger incidents, including the Terra collapse, and hacks of Wormhole and Ronin bridges, with losses in the billion-dollar range. Even after such massive capital withdrawals, core protocols have been able to gradually restore trust through treasury asset compensation, loan adjustments, and community governance. DefiLlama founder 0xNGMI pointed out that although protocols like Aave experienced massive capital outflows in a short period, this resembles a redefinition of risk assets rather than a total collapse. Historical experience shows that when market sentiment stabilizes, capital usually flows back due to cost considerations and potential rewards. Currently, while Aave faces liquidity pressure, it has sufficient adjustment mechanisms to protect the integrity of the protocol. During the KelpDAO incident, the market observed a significant capital transfer phenomenon. The SparkLend protocol maintained sufficient ETH withdrawal liquidity by preemptively removing risk assets such as rsETH, with its total locked value growing from $1.8 billion to $2.9 billion over the weekend, reflecting that capital has not fully retreated from the on-chain ecosystem but is instead seeking safer havens. This article suggests the DeFi TVL loss of $13 billion is due to a chain of liquidations, not actual losses, first appearing at <a>...</a>.

Trump's Second $TRUMP Whale Party: 297 Attendees, Coin Price Down 96% from Peak

According to the Washington Post, U.S. President Trump held his second $TRUMP meme coin whale dinner at Mar-a-Lago in Florida on Saturday, April 25. The event brought together winners of the "Annual Holdings Competition" organized by his team, with the tagline "the world's most exclusive crypto and business conference." CoinDesk and CNBC added financial details: the current price of $TRUMP is $2.59, which is a staggering 96% drop from the peak post-Trump's inauguration. 297 $TRUMP whales attended, with the top 29 enjoying a champagne toast with Trump. The winners, who ranked highest by the amount of $TRUMP they held during a specified window, were granted entry to Mar-a-Lago. The top 29 were further upgraded to a "VIP champagne toast" session, receiving Trump-branded perfume, posters, watches, and other merchandise. According to blockchain analysis firm Nansen, these 297 whales hold a combined total of around $29 million—just one-fifth of the $148 million from the first event in 2025. The entry threshold has also been significantly lowered. According to Fortune, the lowest-ranked qualifying holder this time had around $8,460, down over 80% from the approximately $55,000 minimum threshold of the first event. $TRUMP hit a high of $70 shortly after Trump's inauguration in January 2025, and currently trades at $2.59, with a 14% drop over the last 24 hours. Tether CEO, Cathie Wood, and Mike Tyson Speak on Stage The speaker lineup this time spans finance and entertainment: Tether CEO Paolo Ardoino, Ark Invest founder Cathie Wood, bestselling author Tony Robbins, former heavyweight champion Mike Tyson, Anchorage Digital CEO Nathan McCauley, and ProCap Financial co-founder Anthony Pompliano. CoinDesk reported that the discussion topics included financial inclusion, the role of the dollar, and the intersection of AI and crypto. Trump Defends Crypto Legislation in Keynote Speech Trump used his keynote speech to advocate for two crypto bills that are set to enter the House for a second reading, arguing that "legalizing the crypto industry within the U.S. can bring funds back home." This speech to his own token whales contrasts with recent investigative actions against the event from the Democratic side—Senators Elizabeth Warren, Adam Schiff, and Richard Blumenthal have jointly requested the White House to disclose the list of attendees and information on conflicts of interest. $TRUMP Down 96% from Peak, Liquidity Concentrated Among Few Whales The current price of $TRUMP is $2.59, with a market cap of about $518 million (based on circulation), which represents a 96% drop compared to the peak of $70 during the weekly launch in January 2025. Notably, whale concentration is very high—only 297 addresses control a total of $29 million, nearly 5.6% of the total market cap—reflecting that the circulation structure of $TRUMP is still highly concentrated. Other crypto ventures by the Trump family continue to face challenges. Earlier this week, Justin Sun filed a lawsuit against Trump family’s World Liberty Financial, alleging they froze his $1 billion worth of WLFI tokens and pressured him to invest in their USD1 stablecoin, marking another unresolved front in Trump’s crypto landscape. This article, Trump’s Second $TRUMP Whale Party: 297 Attendees, Coin Price Down 96% from Peak, first appeared on .

Stablecoins aren't just for cross-border payments; they're getting localized! a16z's latest report: Asia accounts for two-thirds of trading volume

The crypto investment giant a16z has released a report titled "9 charts on what stablecoins are becoming," featuring nine key charts that depict the structural transformations happening with stablecoins. The core conclusion of this report isn't about new tokens or narratives but highlights the evolution of stablecoins from 'trading tools' and 'savings vehicles' to 'core financial infrastructure,' increasingly localized, which starkly contrasts the market's original perception of them as cross-border payment solutions. The US GENIUS Act has boosted stablecoin trading volume to a staggering $4.5 trillion per quarter. For years, regulatory uncertainty has acted as a ceiling for institutional participation in the stablecoin market. The turning point came with the US GENIUS Act establishing the first federal framework for stablecoin issuance. According to a16z's data, adjusted stablecoin trading volumes had been rising for several quarters before the bill passed, but the growth accelerated significantly afterward, reaching about $4.5 trillion by Q1 2026. Europe's MiCA framework presents a different picture. After its full implementation at the end of 2024, several major exchanges have delisted USDT for compliance, resulting in a temporary surge in non-USD stablecoin activity exceeding $40 billion. After the volatility settled, monthly trading volumes stabilized at a new baseline of $15 billion to $25 billion, far above the levels seen before MiCA was implemented. In other words, regulation hasn't stifled non-USD stablecoins; instead, it has created a market for them that was virtually nonexistent in the past. Business activity is growing the fastest: C2B up 128%. The structural significance of this transformation lies in the use cases themselves. For the entirety of 2025, C2C (consumer-to-consumer) transfers will still dominate with 789.5 million transactions, but the fastest growth is in C2B (consumer-to-business), skyrocketing from 124.9 million transactions in 2024 to 284.6 million, a year-on-year increase of 128%. The infrastructure data supporting stablecoin payment cards corroborates this trend. The stablecoin card program supported by Rain (including Etherfi Cash, Kast, Wallbit, etc.) saw monthly collateral deposits grow from nearly zero in November 2024 to over $300 million per month by early 2026. Although this still reflects collateral balances rather than direct payments, the trajectory is quite clear: the commercial use of stablecoins is rapidly expanding. The velocity of money circulation has doubled: from 2.6 times to 6 times. The turnover frequency of each unit of stablecoin supply is accelerating. The stablecoin velocity metric calculated by a16z, which measures adjusted monthly transfer amounts relative to circulating supply, has nearly doubled since early 2024, rising from 2.6 times to 6 times. The increase in velocity indicates that the market's demand for stablecoin transactions is growing faster than the issuance rate, meaning the existing supply is being 'used more frequently.' This is a hallmark of a real payment network where currency is 'used' rather than just 'held.' The pure payment scale of stablecoins reaches between $350 billion to $550 billion. After stripping away the 'financial nature' of transactions, institutional treasury transfers, and exchange operational mechanisms, the estimated pure payment scale of stablecoins among different entities in 2025 is between $350 billion to $550 billion. Structurally, B2B (business-to-business) remains the largest segment, but the growth of C2C direct transfers and merchant payment activities is equally noteworthy. Geographic distribution is highly concentrated: Asia accounts for nearly two-thirds. Geographic data shows that stablecoin payment activity is not evenly distributed. The Asian market contributes nearly two-thirds of the payment volume, mainly from Singapore, Hong Kong, and Japan. North America accounts for about a quarter, Europe around 13%, while Latin America and Africa combined account for less than $1 billion. This distribution has direct implications for fintech players in Taiwan and Southeast Asia. The real growth momentum for stablecoin payments is concentrated in the Asian time zone, suggesting that the target customer base, trading partners, and regulatory arbitrage opportunities for related businesses differ from those in the US fintech market. The narrative around cross-border transactions has been overturned: domestic transactions have surged to three-quarters. One of the most counterintuitive charts in the report challenges the mainstream narrative that 'stablecoins are merely cross-border remittance tools.' Actual data shows that the share of cross-border activity in overall payment volume has been declining rather than increasing. The proportion of domestic transactions in payment volume grew from about half in early 2024 to nearly three-quarters by early 2026. Brazil serves as a vivid example. The locally pegged stablecoin BRLA, based on the Brazilian real, saw its monthly transfer volume grow from nearly zero in early 2023 to about $400 million per month by early 2026, largely driven by its integration with the local real-time payment network, PIX. a16z interprets this as stablecoins finding a new positioning: not just remittance or forex tools but as 'local payment mediums running on global infrastructure.' This shift in positioning has significant implications for the competitive landscape of banks, payment institutions, and stablecoin issuers. This article Stablecoins aren't just for cross-border payments; they're getting localized! a16z's latest report: Asia accounts for two-thirds of trading volume first appeared on <a>...</a>.