2026-04-17 Пятница

Новостной центр - Страница 12

Получайте криптоновости и тенденции рынка в режиме реального времени с помощью Новостного центра HTX.

The DeepSeek You've Been Waiting For Has Long Changed

The article discusses the delayed release of DeepSeek V4, a highly anticipated AI model in China, and explores the reasons behind its slowed development. Initially a leader in the global AI race, DeepSeek has fallen behind competitors like OpenAI, Anthropic, and Google, which release major updates every few months. A key factor is DeepSeek's shift in focus due to national strategic priorities. In early 2025, the Chinese government encouraged the company to use Huawei’s Ascend processors instead of NVIDIA’s GPUs, aligning with broader efforts to achieve technological self-reliance. DeepSeek attempted to train its models on Huawei’s Ascend 910C chips but faced technical challenges, including instability and communication issues during distributed training. As a result, the company continued using NVIDIA hardware for training while only using Ascend chips for inference. In 2026, DeepSeek prioritized adapting V4 to Huawei’s new Ascend 950PR and Cambricon chips, aiming for a full migration from NVIDIA’s CUDA to Huawei’s CANN framework. This adaptation process, particularly ensuring precision alignment across hardware, consumed significant time and resources, slowing down model iteration. The delay also reflects DeepSeek’s evolving role from a purely market-driven entity to a "national mission-oriented" company. This shift has come at a cost: the model now lags behind competitors in areas like code generation and multimodal capabilities, and the company has faced talent drain, with key researchers leaving for better-paying opportunities at larger tech firms. Despite these challenges, V4’s release is seen as a potential milestone for China’s AI industry, demonstrating that advanced models can run on domestic hardware ecosystems. While it may not be a groundbreaking model in terms of performance, its success could validate China’s broader strategy for AI independence.

marsbit2 дня назад 10:32

The DeepSeek You've Been Waiting For Has Long Changed

marsbit2 дня назад 10:32

Agents Have Entered the Harness-Driven Era

The article discusses the significance of the leaked Claude Code from Anthropic, highlighting its revelation of advanced Agent engineering practices centered on "Harness" design. Rather than relying solely on model capabilities, modern AI systems now depend on a structured engineering framework—the Harness—to maximize performance. This framework includes six core components: multi-layered System Prompts, Tool Schema, Tool Call Loop (with Plan and Execute modes), Context Manager, Sub-Agent coordination, and Verification Hooks. The Harness enables tighter integration between training and inference, supports long-chain tool execution, and improves reliability through objective verification. It also drives six key training directions: behavior alignment via System Prompt, end-to-end tool-use training, integrated plan-execute training, memory compression, sub-agent orchestration, and multi-objective reinforcement learning. The shift to Harness-driven development reduces the emphasis on pure prompt engineering, favoring instead multidisciplinary talent with skills in AI, backend engineering, and infrastructure. The market is evolving toward more secure, private, and vertically integrated Agent deployments, with "model shell" companies needing either strong infrastructure or deep domain expertise to compete. Claude Code’s leak underscores that future AI advancements will be shaped by engineering architecture as much as by algorithmic innovation.

marsbit2 дня назад 10:11

Agents Have Entered the Harness-Driven Era

marsbit2 дня назад 10:11

NVIDIA Starts Selling the Method of Making Shovels

NVIDIA is accelerating its own GPU design process with AI tools, reducing tasks that took 8 engineers 1010 months to a single GPU running overnight, while achieving better results. However, it is not using AI to fully design GPUs yet. Instead, it is embedding its technology across the chip industry. A key move is its $2020 billion investment in Synopsys, a leading EDA (Electronic Design Automation) tool provider, and partnerships with Cadence and Siemens. This integrates NVIDIA’s hardware acceleration into essential chip design software, making its GPUs the fastest platform for EDA tools. Competitors like AMD and Intel now face a dilemma: use slower tools or rely on NVIDIA’s hardware to design competing chips. Chinese GPU firms, despite high market valuations, struggle with massive losses and dependence on restricted foreign EDA tools. NVIDIA’s strategy extends from AI-enhanced design to manufacturing, creating a self-reinforcing cycle where its hardware becomes essential for developing faster chips—even for those aiming to compete with it.

marsbit2 дня назад 10:04

NVIDIA Starts Selling the Method of Making Shovels

marsbit2 дня назад 10:04

Crypto-Backed Super PAC Launches Midterm Election Endorsements Push

A crypto-aligned super PAC named Fellowship PAC, chaired by Tether's head of government affairs Jesse Spiro, has launched its 2026 midterm election push with over $100 million to spend. Its first action was a $300,000 ad buy supporting a Republican candidate in a Georgia congressional race. The PAC has also endorsed several other Republican candidates in key Senate and gubernatorial races across five states. This strategy mirrors previous crypto-backed efforts, like the Fairshake PAC, which spent heavily in the 2024 cycle. Meanwhile, major crypto legislation, the CLARITY Act, remains stalled in the Senate.

bitcoinist2 дня назад 10:02

Crypto-Backed Super PAC Launches Midterm Election Endorsements Push

bitcoinist2 дня назад 10:02

The State of Crypto VC: Capital Quadrupled, Investors Down 93%, Where Are the Opportunities?

The state of crypto VC in 2025 is marked by a surge in total funding (up 433% to $40-50B) but a sharp decline in active investors—only 377 in Q4 2025, down from 5,500 in 2022. Deal count dropped 42%, with capital concentrated in fewer, larger rounds. Power has shifted decisively to VCs, who now favor later-stage investments, leaving early-stage opportunities with minimal competition. AI absorbed 61% of global VC funding, drawing away generalist and weak-handed investors. Even crypto-native funds like Paradigm expanded into AI and robotics. Regulatory clarity, not Bitcoin’s price, drove capital inflow. The market is bifurcated: mega-rounds (11 deals >$100M took 85% of Q4 funding) dominate, while mid-stage deals vanish. Pre-seed remains viable (23% of deals) but with higher scrutiny. Remaining investors face less competition and better deals. The focus is on infrastructure, revenue-generating, compliant sectors like Web2.5, trade, stablecoins, and payments. Speed and conviction are critical—capital moves fast to top projects. The opportunity is clear; the question is who has the courage to go all-in.

marsbit2 дня назад 09:43

The State of Crypto VC: Capital Quadrupled, Investors Down 93%, Where Are the Opportunities?

marsbit2 дня назад 09:43

Are People Who Hold Cryptocurrency Really Filing Taxes?

A study analyzing IRS data from 2013 to 2021 suggests that many cryptocurrency investors in the U.S. may be evading taxes. Researchers found that only 6.5% of taxpayers reported crypto transactions during this period, despite surveys indicating that 12–21% of American adults held crypto. The study, published in a Springer Nature accounting journal, describes crypto investors as typically having lower incomes, being younger, and more likely to invest in meme stocks compared to traditional investors. Many also trade without considering tax implications, possibly due to a lack of expertise or the highly volatile nature of cryptocurrencies. However, stricter IRS reporting rules are set to take effect in 2026, requiring exchanges like Coinbase to issue tax forms and potentially improving compliance.

marsbit2 дня назад 09:42

Are People Who Hold Cryptocurrency Really Filing Taxes?

marsbit2 дня назад 09:42

48 Hours, $2.7 Billion: Strategy's STRC ATM Mechanism is Accumulating Bitcoin at an Unprecedented Pace

Strategy's STRC ATM (At-The-Market) mechanism is accelerating its Bitcoin acquisition at an unprecedented pace. In just the first two trading days of the week, it generated $2.74 billion in volume, absorbing approximately 29,914 BTC—more than double the 13,927 BTC acquired in the entire previous week, representing a 115% increase. Notably, 100% of the STRC shares traded on both Monday and Tuesday were above the $100 face value threshold, the condition required to trigger the ATM and convert equity demand into Bitcoin purchases. Tuesday alone saw an estimated $1.57 billion in volume, absorbing 16,762 BTC—equivalent to 37 times the daily Bitcoin mining output. This surge demonstrates a new scale of sustained demand for the variable-rate perpetual preferred shares. If this pace continued, the weekly acquisition could reach roughly 75,000 BTC, fundamentally redefining corporate Bitcoin treasury strategies. While such extreme velocity may not be sustained, the mechanism is consistently converting market demand into large-scale Bitcoin accumulation.

marsbit2 дня назад 09:35

48 Hours, $2.7 Billion: Strategy's STRC ATM Mechanism is Accumulating Bitcoin at an Unprecedented Pace

marsbit2 дня назад 09:35

Not Just USDT: Tether Wallet Is Attempting to Take Over the Payment System for Ordinary People

Tether, the issuer of the world's largest stablecoin USDT, has launched Tether Wallet, branded as "The People's Wallet," marking a strategic shift from being primarily an asset issuer to directly engaging with end-users. This move aims to capture retail traffic and create a closed-loop ecosystem by offering a simplified payment interface. The wallet eliminates key barriers to crypto adoption: complex hexadecimal addresses are replaced with human-readable usernames (e.g., username@tether), transaction fees (gas) are abstracted and paid directly in the transferred asset, and self-custody is combined with an encrypted cloud backup system for easier recovery. Supported assets include USDT on Ethereum, Polygon, and other networks, as well as Bitcoin and Tether’s gold-backed XAUT. Notably, it does not yet support Tron, where nearly half of all USDT is issued. By drastically reducing friction in cross-border payments—enabling instant, low-cost transfers via email-like addresses—Tether is positioning USDT to dominate small-value international settlements, particularly in emerging markets. This challenges traditional remittance services and competing stablecoins like USDC by leveraging its first-mover advantage and network effects. The piece also highlights underlying tensions: while promoting financial inclusion for the unbanked, Tether’s centralized infrastructure creates potential regulatory vulnerabilities. The wallet’s design also anticipates future use by AI agents for machine-to-machine payments. Ultimately, Tether Wallet represents both an expansion of Tether’s influence and a critical test of balancing efficiency, decentralization, and regulatory compliance in the evolving digital financial landscape.

marsbit2 дня назад 09:31

Not Just USDT: Tether Wallet Is Attempting to Take Over the Payment System for Ordinary People

marsbit2 дня назад 09:31

Can Humans Control AI? Anthropic Conducted an Experiment Using Qwen

Can Humans Control Superintelligent AI? Anthropic’s Experiment with Qwen Models Anthropic conducted an experiment to explore whether humans can supervise AI systems smarter than themselves—a core challenge in AI safety known as scalable oversight. The study simulated a “weak human overseer” using a small model (Qwen1.5-0.5B-Chat) and a “strong AI” using a more powerful model (Qwen3-4B-Base). The goal was to see if the strong model could learn effectively despite imperfect supervision. The key metric was Performance Gap Recovered (PGR). A PGR of 1 means the strong model reached its full potential, while 0 means it was limited by the weak supervisor. Initially, human researchers achieved a PGR of 0.23 after a week of work. Then, nine AI agents (Automated Alignment Researchers, or AARs) based on Claude Opus took over. In five days, they improved PGR to 0.97 through iterative experimentation—proposing ideas, coding, training, and analyzing results. The findings suggest that, in well-defined and automatically scorable tasks, AI can help overcome the supervision gap. However, the methods didn’t generalize perfectly to unseen tasks, and applying them to a production model like Claude Sonnet didn’t yield significant improvements. The study highlights that while AI can automate parts of alignment research, human oversight remains essential to prevent “gaming” of evaluation systems and to handle more complex, real-world problems. Anthropic chose Qwen models for their open-source nature, performance, scalability, and reproducibility—key for rigorous and repeatable experiments. The research demonstrates progress toward automated alignment tools but also underscores that AI supervision remains a nuanced, human-AI collaborative effort.

marsbit2 дня назад 09:28

Can Humans Control AI? Anthropic Conducted an Experiment Using Qwen

marsbit2 дня назад 09:28

活动图片

Популярные категорииright-arrow

Популярные тегиright-arrow

Regulatory Policy

Indepth Research