CoinVoice

Author: Zhou, ChainCatcher
After the “pioneer” of perpetual futures, BitMEX, shut down, BitMart also collapsed.
On July 15, BitMart only just released its 2026 first-half report. The report uses an eight-year anniversary theme, talks throughout about long-termism and building for the next cycle, and even lists six major development directions for the second half.
Just 11 days later, BitMart announced an orderly shutdown of its trading platform operations.
However, this doesn’t mean the company will disappear immediately. An official announcement states that the platform’s trading services will continue until August 26, and the account withdrawal window will remain open until January 2027. Users can still retrieve their assets within the deadline.

CoinVoice learned: According to MLM monitoring, since the HYPE team tokens started unlocking in December 2025, 4.93 million HYPE tokens have been allocated to team members. Based on the current price, the value is approximately $270 million, accounting for 0.493% of the total supply.
Meanwhile, 1.19 million HYPE tokens have been sold on the open market for $32.5 million; additionally, 3.14 million HYPE tokens have been transferred to an OTC trading platform, with the transfer value estimated at about $132 million.
In total, about 4.33 million HYPE tokens have been sold, generating approximately $165 million. On average, about 540,000 HYPE tokens are sold per month, with a value of roughly $20.6 million.

Written by: Bu Shuqing
Cash flow data show that global investors are undergoing a profound reshuffling of asset allocations.
According to Chase the Trends trading desk, Bank of America’s latest edition of its (Fund Flow Report) (The Flow Show) reveals that, amid rising expectations for Fed rate hikes and persistently higher yields on the long end, “smart money” is accelerating toward emerging markets, technology stocks, and commodities, while remaining cautious about U.S. domestic equities and U.K. stocks.
As of the week ending July 22, global equity funds saw net inflows of $30.4 billion for the week, bond funds net inflows of $14.9 billion, gold funds net inflows of $2.0 billion, while money market funds recorded net outflows of $33.9 billion.

CoinVoice learned that Storj Labs voluntarily filed for Chapter 11 restructuring on July 26 with the U.S. Northern District of West Virginia Federal Bankruptcy Court to address historical legacy debt issues while keeping the company operating. Storj said that its services will not be affected during the restructuring period, and customer obligations will continue to be fulfilled. Its parent company, Inveniam, supports the restructuring, viewing it as the right path to achieve sustainable development.
Storj software engineering director Kaloyan Raev said that the fundamentals are strong, and what constrains growth is early legacy liabilities. After the restructuring, a “clean foundation” will be established, and the company plans to have management, the token community, and investors jointly participate in the post-restructuring allocation of company ownership. Storj previously focused its business on the core distributed storage sector and is currently winding down its previously acquired and non-core businesses. This restructuring is another major governance adjustment for the company in the decentralized storage space following its 2024 acquisition of Storj by Inveniam.

CoinVoice learned: China’s leading DRAM maker, Changxin Technology (688825.SH), was listed on the STAR Market today for its first trading day. It opened at 49.5 yuan, which is 471.59% higher than the issue price of 8.66 yuan. Its market cap reached 3.3 trillion yuan, ranking first among A-shares. This IPO raised about 57.9 billion yuan (before the full exercise of the over-allotment option), making it the largest IPO on the STAR Market to date. Based on the opening price, investors who received one lot could earn about 20,000 yuan.
On the same day, Nomura Securities initiated coverage of Changxin Technology, assigning a target price of 116 yuan. The target was based on an expected earnings per share of 5.8 yuan in 2028 and a 20x target price-to-earnings ratio, implying an upside of about 13.4x versus the issue price and an implied market cap of about 7.76 trillion yuan. Nomura believes this valuation is about twice that of Micron, mainly driven by expectations for growth in Changxin’s share of the memory market and the valuation premium in the China market.

Written by: Sandy Kaul, Head of Digital Assets and Innovation at Franklin Templeton
Compiled by: Jiahua, ChainCatcher
AI’s evolution is driving the investment narrative
AI has always been evolving. Around the early 2010s, early capabilities such as machine learning, natural language processing, and predictive analytics helped ignite the “big data” era—enabling people to process structured and unstructured data at speeds and scales previously unimaginable. Back then, AI was more like a tool that assists human work.
In the early 2020s, generative AI emerged, and its uses and roles took a major leap forward. AI evolved from a helper into a “co-creator”—able to generate content, respond to various questions, and even take over part of people’s work. The potential of generative AI has not yet been fully unleashed; products are getting stronger and are increasingly penetrating more corners of everyday life.

CoinVoice learned: according to Artemis data, on July 21 Robinhood Chain recorded 323,969 daily active users, exceeding Base’s 274,520. This marked the second time it surpassed Base since the mainnet went live on July 1. On the same day, TVL hit a new high of $588.9 million, up 14% day-over-day; at the end of June, this metric was still under $5 million. Morpho accounted for 44% of TVL at $260.6 million, while Ethena accounted for 26%.
On-chain stablecoin supply reached $438.8 million, up 30% week-over-week. Spot DEX 24-hour trading volume was $624 million, also exceeding Base’s $603 million. However, the main driver of activity is meme coin trading rather than the tokenized stocks that the chain focuses on; the market cap of tokenized stocks is only $19.3 million. Base still leads by a wide margin on most metrics: DeFi TVL is about $4.64 billion—15 times that of Robinhood Chain—while stablecoin supply is $4.84 billion, 11 times as much. Perpetual contract trading volume is $255.5 million, far above Robinhood Chain’s $13.5 million. Robinhood Markets’ stock price rose more than 8% on launch day, and on July 21 it was reported at $107.55.

CoinVoice learned: According to Token Terminal data, the tokenized stocks market reached a record $2.3 billion in mid-July, nearly doubling from when it first broke $1 billion in March. On July 21, On the same day, Ondo Finance’s tokenized stock circulating supply (514.5 million shares) and number of holders (93,880), Backed Finance’s tokenized market value ($579.4 million), Robinhood Chain’s tokenized stocks (126,720 shares) and number of holders (36,170) all hit record highs.
Arcus, a tokenized stocks exchange launched by the dYdX team, recorded record highs the same day: $11.9 million in daily perpetual contract trading volume and $6.8 million in open interest. By on-chain distribution, Ethereum accounts for 34% of the tokenized stocks market share, BNB Chain 30%, and Solana 23%. Ondo Finance leads with $955 million in on-chain stock scale. It has recently partnered with Japan’s SBI Group to tokenize Japanese stocks and has launched 24/7 minting/redemption and voting power features. Within eight months of launch, xStocks’ cumulative trading volume surpassed $25 billion. Previously, Ondo executives projected that the tokenized stocks market value would reach $2.5 to $3.0 billion by year-end.

CoinVoice learned: In response to allegations by Michael Kratsios, Director of the Office of Science and Technology Policy at the White House, that Moonshot AI (Kimi) referenced Anthropic Fable when developing the K3 model and built a large-scale distillation platform, a member of the Moonshot AI team, Randy Xia, said that Fable was released on July 1 and K3 went live on July 15. Following this timeline, it means the model research, distillation, training, and release were all completed within 15 days. He joked, “We trained a brand-new frontier model in just 15 days—that’s basically creating a Guinness World Record.”

CoinVoice learned: According to Blockaid monitoring, at 5:30 a.m. Beijing time, the Arbitrum ecosystem protocol AFX was attacked. The attack targeted a cross-chain bridge operated by this protocol; as of now, about 24.15 million USDC has been stolen. Blockaid said it is working with the Arbitrum team to respond to the incident, to communicate with the affected protocol, and to help freeze the stolen funds.
In addition, Steven Goldfeder, co-founder of Offchain Labs, responded on X that the Arbitrum team has noticed reports of the cross-chain bridge attack on Arbitrum and is investigating. Goldfeder confirmed that the relevant transactions originated from a third-party protocol. No attack or exploitation of the Arbitrum native bridge occurred. The team will coordinate with the third-party team and provide more details.

Written by Matt Hougan, Bitwise CIO
Compiled by Saoirse, Foresight News
The crypto market is finally showing signs of a bottom. Since July 1, Bitcoin has risen 9%, while the Nasdaq 100 has fallen 6% over the same period. Crypto ETF flows have turned from negative to positive, and market sentiment has continued to improve. While it’s still too early to say the market has fully stabilized, these encouraging signals have already prompted many to start asking about the path ahead for the next leg of the rally.
Last Friday, a financial adviser asked me: “If the market has already bottomed out, what asset will lead the next wave of the crypto bull market?”

STON.fi——The leading AMM protocol on The Open Network (TON)——announced today that its application has officially launched a cross-chain swap feature, providing users with a unified, self-custody interface to directly transfer stablecoins between TON, TRON, Ethereum, Base, BNB Chain, Polygon, Avalanche, Arbitrum, and Robinhood Chain.
The launch connects TON with major liquidity and application ecosystems. Users can freely route funds between the stablecoin market, TON native assets, DeFi protocols, and Telegram native applications without relying on centralized exchanges, cross-chain bridges, or wrapped assets.

Author: Wintermute
Compilation: Deep Tide TechFlow
Deep Tide Guide: Wintermute releases an industry manifesto: the battle over encrypted infrastructure is over. The next battlefield is not DeFi, but machine economics. When AI agents, warehouse robots, and automated experimentation systems become economic actors, the underlying assumption of traditional finance—"the other side is human"—will completely fail. The difference in the crypto track—"the other side is code"—turns from a flaw into a core advantage. Three directions worth paying attention to: agent economy layer, physical AI, and machine-driven discovery.
The old problem is dead, and a new one is happening

CoinVoice learned that, according to a report by “AI Technology Review,” Chinese large-model leader Zhipu has spent several hundred million yuan to complete the acquisition of Zhongke Jiahe, an AI heterogeneous computing software infrastructure company. The move is intended to fundamentally address Zhipu’s shortcomings in underlying engineering and compiler capabilities for large models, in order to cope with structural shortages in computing resources brought on by a surge in user numbers and the challenges of high-concurrency inference.
The technology of Zhongke Jiahe traces back to the Compilation Laboratory of the Institute of Computing Technology of the Chinese Academy of Sciences, and was founded by Dr. Cui Huimin. Its core team has previously been deeply involved in the development of compilers for a number of homegrown chips, including Loongson, Sugon, Cambricon, and Huawei Ascend. Zhongke Jiahe’s core advantage lies in its virtual instruction set technology, which can unify different brands and models of chip ecosystems through an intermediate-layer software layer—assembling scattered domestic chips into a unified, ultra-large-scale cluster—thereby significantly improving overall utilization of computing power. According to official claims, its inference engine SigInfer can reduce large-model inference latency by up to 74 times.