2026-08-06 Quinta

Notícias de cripto - Página 258

Mantenha-se a par do mercado de cripto. Notícias em tempo real, análises, preços, histórias em alta e análise de especialistas — tudo num só lugar.

AI Billing Black Box Exposed: 1.7 Million Overcharged, Anthropic Refunds But Doesn’t Admit Fault

A startup named Vaudit, founded by former Oracle director Michael Hahn, audits AI bills for companies and claims to have identified approximately $1.7 million in overcharges across 60 businesses, totaling $34 million in reviewed bills. The alleged discrepancies primarily involve charges for Anthropic's Claude Code. Common issues cited include billing for newer, more expensive models when older, cheaper ones were used; charging for failed or errored requests; and "retry storms" where AI agents silently retry failed tasks, accumulating costs unnoticed. Major clients like Panasonic, HP, and Honda were among those audited. While Vaudit reports that around 80% of the disputed charges were refunded by providers like Amazon, Google, Microsoft, Anthropic, and OpenAI after申诉, the AI companies largely deny systemic problems. Anthropic stated overcharges do not appear widespread and it does not bill for uncompleted requests or errors, while OpenAI said it found no evidence of such issues affecting its customers. The situation highlights the inherent opacity and complexity of AI billing, which is based on token usage that is difficult to track and predict, especially with multi-agent, multi-model workflows. This complexity is creating a new market for third-party AI bill auditing services like Vaudit, which charges fees based on recovered amounts. Separately, Anthropic faces a proposed class-action lawsuit alleging its high-tier subscription plans deliver far less usage than advertised. The case underscores growing scrutiny over AI service pricing and transparency as major providers prepare for IPOs.

marsbit06/29 09:34

AI Billing Black Box Exposed: 1.7 Million Overcharged, Anthropic Refunds But Doesn’t Admit Fault

marsbit06/29 09:34

Tencent Buys Baidu Chips

China's internet giants, once defined by building closed, self-sufficient empires, are undergoing a fundamental shift. A key signal is Baidu's plan to spin off its AI chip unit, Kunlun Xin, for a Hong Kong IPO targeting a $50 billion valuation, potentially exceeding its parent company's worth. Concurrently, Alibaba's T-Head is also pursuing independence. Most significantly, reports indicate that rival Tencent has become a major customer for Kunlun Xin's chips. This move, where competitors begin procuring each other's core technologies, marks a decisive break from the past era of internal duplication and isolation. It signals the maturation of China's AI industry into a more open, specialized ecosystem. The underlying driver is the immense and clear cost of AI infrastructure, particularly the exploding demand for inference compute driven by AI agents and applications. Hardware is no longer just an internal cost center but a profitable, strategic business in itself. Globally, a parallel trend is evident as OpenAI, Google, Amazon, and others develop their own AI chips to control costs and optimize performance. The competition has moved beyond model benchmarks to a deeper, foundational war over token cost efficiency, inference cluster performance, and secure, scalable computing power. Baidu and Alibaba aren't dismantling their empires but are instead decoupling non-core, capital-intensive infrastructure to participate in and shape a larger, collaborative industrial base. The era of the all-encompassing super-app is giving way to an age of strategic specialization and open ecosystem building in the AI race.

marsbit06/29 09:18

Tencent Buys Baidu Chips

marsbit06/29 09:18

The Token Itself Is an Asset: Three Types of Tokenized Stocks, Which One Suits You?

"Tokenized Stocks: Three Types, Which One Fits You? For investors outside the US, buying stocks like SpaceX or Nvidia is difficult, requiring brokers, cross-border transfers, and often accredited investor status. Blockchain offers an alternative through tokenized stocks, a term encompassing three distinct products with vastly different ownership, voting, and profit rights. 1. **Full Real Ownership**: Companies like Superstate register native equity directly on-chain (e.g., Solana). Holders are on the official shareholder registry, with full voting rights, dividends, and legal ownership. This offers maximum rights but potentially less DeFi flexibility. 2. **SPV-Backed Tokens (Surrendered Ownership for DeFi Composability)**: Issuers like Backed (xStocks) and Ondo use offshore Special Purpose Vehicles (SPVs) to hold underlying shares 1:1 and issue tracking tokens. Investors get price exposure and dividends (reinvested as more tokens) but hold a claim on the SPV, not direct stock ownership. This enables use as collateral in DeFi protocols (Kamino, Morpho) and 24/7 minting/redemption, but carries SPV counterparty risk (highlighted by the PreStocks collapse). 3. **Perpetual Futures (Pure Price Speculation)**: Platforms like TradeXYZ (on Hyperliquid) and Ostium offer perpetual contracts. These are synthetic derivatives with no underlying stock ownership, using funding rates to track spot prices. They require only a price oracle, allowing extremely fast listing (e.g., SpaceX pre-IPO) and high leverage, attracting speculators. Their trading volume far exceeds tokenized spot products. The core value of tokens is that they don't need to replicate full stock ownership. Most retail investors never vote. Tokenization creates layered financial tools: full equity for institutions, composable tokens for DeFi users, and perpetuals for leveraged traders."

marsbit06/29 09:18

The Token Itself Is an Asset: Three Types of Tokenized Stocks, Which One Suits You?

marsbit06/29 09:18

AI as the Boss: Nearly Bankrupts 10 Companies...

A recent study from Princeton University tested 14 AI models, including large language models (LLMs) and a rule-based algorithm, in a simulation where they acted as CEOs of a virtual SaaS startup over 500 days. The goal was to grow an initial $1 million capital. The results were stark: only four "CEOs" ended with a profit. The top performer was Claude Fable 5, multiplying the capital 47-fold to $47.15 million. Claude Opus 4.8 and GPT-5.5 followed. Notably, the fourth profitable entity was a simple, pre-programmed rule-based algorithm, which outperformed many advanced LLMs with $15.76 million in profit. Five other models, including several major LLMs, went bankrupt before the simulation ended. Key takeaways from the research highlight that successful AI CEOs demonstrated a tendency for exploration and adaptation over caution. They excelled in discovering hidden information, predicting future cash flow, adapting quickly to changes (like competitor moves), and engaging in strategic "if-then" planning. The study also found that equipping LLMs with programming-agent frameworks, optimized for coding tasks, actually harmed their performance in this CEO role, suggesting a need for domain-specific adaptations. The article concludes by contrasting AI's current operational proficiency within defined frameworks with the type of visionary, intuitive decision-making—exemplified by figures like Steve Jobs—that truly drives transformative business strategy. This critical "matrix-drawing" capability, it argues, remains uniquely human.

marsbit06/29 09:08

AI as the Boss: Nearly Bankrupts 10 Companies...

marsbit06/29 09:08

Tokens as Assets: Which Type of Tokenized Stock Is Right for You?

**Tokenized Stocks: Three Models, Which Suits You?** For investors outside the US, accessing stocks like SpaceX or NVIDIA is often difficult, requiring compliant brokers and cross-border transfers. Blockchain offers an alternative through tokenized stocks, but this term encompasses three distinct models with vastly different ownership, voting rights, and economic benefits. The first model offers full, direct ownership. Platforms like Superstate register shares directly on-chain (e.g., Solana), with holders listed on the official shareholder registry, granting full voting rights, dividends, and legal status. The second model sacrifices direct ownership for DeFi composability. Issuers like Backed and Ondo use offshore Special Purpose Vehicles (SPVs) to issue tokens 1:1 backed by real shares. Holders gain price exposure and automated dividend accruals (via token balance increases), and tokens can be used as collateral in DeFi protocols. However, this introduces SPV counterparty risk, as seen in the PreStocks collapse. The third model abandons ownership entirely for pure price speculation. Perpetual futures platforms like TradeXYZ (on Hyperliquid) and Ostium create synthetic markets using price oracles and funding rates to track stock prices. They require no underlying shares, enabling rapid listing (e.g., SpaceX pre-IPO) and high leverage, which explains their trading volumes being 4-5x higher than tokenized spot markets. The core insight is that tokens derive value without needing to replicate full stock ownership. Most retail investors rarely exercise voting rights. These three models cater to different needs: direct ownership for institutions, DeFi-composable tokens for on-chain users, and perpetuals for leveraged, speculative traders. Tokenization is not a mere stock substitute but a new class of layered financial instruments.

Foresight News06/29 09:07

Tokens as Assets: Which Type of Tokenized Stock Is Right for You?

Foresight News06/29 09:07

Trading Moments: Bitcoin's 200-week moving average has turned into a resistance level, can the July rise still be realized?

**Market Recap: Key Global Developments and Outlook** Global markets are navigating shifting dynamics. Geopolitical tensions eased as the US and Iran agreed to halt further military actions, planning talks for June 30. This pushed oil prices down, with WTI crude dropping below $70. Meanwhile, gold saw a "death cross" (50-day moving average crossing below the 200-day), pressured by a strong dollar and rising real yields. A methodological revision to the US PCE inflation index, set for September, is expected to artificially lower reported core inflation, drawing criticism for lack of transparency. In US equities, major indices extended losses, with the S&P 500 and Nasdaq recording their longest losing streaks since last year. Hedge funds aggressively sold tech stocks, particularly in semiconductors, leading to a sharp rotation into defensive sectors like healthcare and utilities. SpaceX is set for rapid inclusion in the Nasdaq 100, potentially triggering significant passive fund inflows, while its valuation faces scrutiny. Bitcoin is on track for its worst monthly performance since 2022, down over 18% in June. It has failed to reclaim its 200-week moving average (now acting as resistance near $62.6k), raising the risk of a drop toward $55k. Historically, July is a strong month for BTC, with an average gain of 7.6%. Analysts suggest the current sell-off could present a buying opportunity, with a key test being whether Bitcoin can stabilize above $61k to confirm a reversal. In Asia, South Korean stocks initially fell on semiconductor selling but recovered after the government announced a massive investment plan for chips and AI. Japanese retail sales showed strength, supporting consumer recovery. Chinese markets saw a rebalancing, with healthcare stocks surging on policy catalysts and consumption shares rebounding. Hong Kong tech stocks also rallied. **Key upcoming events:** * June 30: US-Iran technical talks; MiCA transition deadline in Spain; NetEase's dual-primary listing in Hong Kong. * July 1: ECB's Sintra Forum featuring key central bankers. * July 6-7: SpaceX's inclusion in Nasdaq 100 triggering passive fund flows. * September 30: New US PCE methodology takes effect.

marsbit06/29 08:42

Trading Moments: Bitcoin's 200-week moving average has turned into a resistance level, can the July rise still be realized?

marsbit06/29 08:42

Lao Huang: Prompt is Dead, the Entire AI Community is Frenziedly Chasing Loops

The article "Prompt is Dead: The AI Industry is Obsessively Chasing Loops" discusses a major shift in AI development, where "Loop Engineering" is replacing traditional prompt engineering. Industry leaders like NVIDIA's Jensen Huang, Andrew Ng, and engineers from Anthropic and OpenAI argue that manually crafting prompts is becoming obsolete. Instead, the new focus is on designing autonomous, self-improving AI systems (loops) that can operate 24/7. A loop system typically involves five key phases: Discovery (finding tasks), Handoff (assigning to agents), Validation (critical independent review), Persistence (saving progress), and Scheduling (automated operation). The core idea is to move humans from being the operational "engine" to being the system "architects" who design the loop, define goals, and set up verification mechanisms. A major challenge and necessity is implementing robust, independent validation to prevent AI from uncritically approving its own work. The trend is seen as part of a move towards "inference-time compute," where allocating computational budget effectively becomes a key engineering skill. While loops can produce higher-quality outputs, they are more expensive and time-consuming than simple prompting. The article warns of risks like "verification debt," "comprehension corrosion," and "cognitive surrender," where engineers might stop understanding the code their systems generate. Ultimately, the article concludes that in an era of automated loops, human judgment and oversight remain the most critical and scarce resources.

marsbit06/29 08:37

Lao Huang: Prompt is Dead, the Entire AI Community is Frenziedly Chasing Loops

marsbit06/29 08:37

活动图片