shift

Justin Drake posted on the X platform that Ethereum L1 will no longer use the SNARK-friendly hash function Poseidon, which has dominated since 2019, and will instead turn to traditional hash functions such as SHA2 or BLAKE2s.This adjustment is based on advancements in the field of SNARK design, shifting the focus from "SNARK-friendly hashing" to "hash-friendly SNARK." By aligning Boolean operations in traditional hashing with binary field native, the proof performance of traditional hashing in SNARK has reached 1 million per second, with an overhead of about 100 times.Research from Binius and Flock has driven relevant progress. The Ethereum Foundation's post-quantum team is advancing this route, with a roadmap that includes deploying production-level leanVM in 2027, and deploying in the consensus layer, execution layer, and data processing layer in 2028.Justin Drake also stated that the enhanced capabilities of AI in cryptanalysis have recently led to setbacks for the lattice-based scheme HAWK and the homomorphic scheme SQIsign, with hash-based schemes being used for post-quantum signatures on the blockchain. The trend of open-source automated research is also accelerating, with SNARK.fast achieving 1.8 million BLAKE3 proofs per second.

CoreWeave warns investors that transitioning from NVIDIA chips to other chips will be difficult and may require a significant investment of time and money.

The price of Bitcoin fell below $64,000 to a weekly low, while gold rose to $4,435 per ounce, a nine-week high.Retail investors flowed $50 million into gold ETFs in a single day.

SK Hynix stated in a recent article that the competition in AI is shifting from the performance of individual chips to the design and operation of the entire infrastructure architecture. The competitiveness of AI data centers no longer depends on individual components, but on whether the five key elements of computing, memory, storage, networking, and power cooling can be seamlessly integrated.The article points out that the continuous expansion of AI model scales has led to a surge in demand for computing power and data movement. Training requires repeatedly reading massive datasets, while inference relies on quickly retrieving user request information, both of which place higher demands on the system architecture of data centers. At the memory level, HBM, server DRAM, and others have formed a hierarchical system, each undertaking different bandwidth and capacity tasks. At the networking level, as large-scale training and inference rely on multi-server parallel processing, networking has become a key factor determining the scalability of data centers. System design is shifting from single-server to whole rack and cluster-level expansion.According to Omdia's forecast, the AI data center chip market will grow from $123 billion in 2024 to $207 billion in 2025, reaching $286 billion by 2030. SK Hynix also mentioned that Microsoft's Fairwater data center in Wisconsin is about the length of five football fields, indicating that infrastructure is being deployed on a larger scale. SK Hynix emphasizes that memory is becoming a key layer connecting computing and data.

Grayscale Research Director Zach Pandl stated that even if the Clarity Act does not pass, the main blockchain operations, the demand for Bitcoin as a store of value, and the growth of stablecoin payments will not be immediately affected, and regulators will fill the regulatory gaps through rule-making.Zach Pandl pointed out that the lack of comprehensive market structure legislation may suppress new investment activities in the U.S., prompting participants in the crypto industry and startups to turn to overseas jurisdictions with clear regulatory frameworks. The U.S. government will continue to support the development of the crypto ecosystem. Strategy co-founder and Executive Chairman Michael Saylor previously stated that regardless of whether the Clarity Act passes, Bitcoin will continue to evolve, but the U.S. needs clarity in digital asset regulation. Senator Bernie Moreno stated that Senate Democrats and Republicans have concluded relevant negotiations and will proceed to a vote.

Serenity posted that it remains bullish on the storage sector represented by Micron (MU) and Samsung, believing that the market often rotates between different supply bottlenecks. This week, stocks in the photonics sector, such as AXTI and LITE, have once again become the focus; compared to the decline in July, the main change is the stock price, followed by scattered narratives and business updates.He pointed out that the market was already aware during the July decline that COHR and LITE's laser products would be sold out for the next two years, and it had also learned about the demand imbalance from AAOI's last quarter earnings call. Aside from the changes in stock prices after liquidation, the fundamentals of the photonics sector have not deteriorated; the bottlenecks in optical transceivers and indium phosphide substrates remain unchanged and may even worsen.In terms of storage, Serenity stated that retail investors are showing signs of capitulation, but the same group had previously signed 16 SCAs with MU and had extremely bullish forecasts. He believes that the current operating profit to market value ratio of the storage business is "extremely unreasonable," and that next year's demand imbalance may be even more severe; the only thing that has changed for the same company is the valuation and narrative, while the market rotates between different sectors.

Berkshire Hathaway today released its Q2 2026 financial report, with the most market attention focused on the cash reserves dropping to $36.551 billion in the second quarter, down from about $39.74 billion in the first quarter. This marks the end of Berkshire's 14 consecutive quarters of net selling, turning into a significant net buying for the first time since the fourth quarter of 2022.In the financial report, Berkshire's net stock purchases in the second quarter were nearly $20 billion, including about $10 billion in a private placement from Google's parent company Alphabet to support investments such as its AI data centers. Approximately $6.8 billion was spent to acquire homebuilder Taylor Morrison, which was a complete acquisition and not entirely a public market stock transaction. About $4.5 billion was used to repurchase its own shares. After deducting the above major items, there remains about $3 billion of "unexplained" net purchases of public market equities, with specific stocks to be disclosed in the 13F filing around August 14. Currently, Alphabet officially enters Berkshire's top five holdings, alongside American Express, Apple, Bank of America, and Coca-Cola, with the top five holdings accounting for about 66% of the stock portfolio.Buffett previously stated that the long net selling period was mainly due to high market valuations, making it difficult to find sufficiently attractive opportunities. This shift is seen as a clear signal of more aggressive capital allocation under Abel's leadership as CEO, with Berkshire moving from "patient waiting" to "starting to act."

According to monitoring by Catcher Predict, in the predicted market Polymarket for the event "LoL: Shifters vs GIANTX - LEC Regular Season," the "Yes" option for the sub-market "Both teams kill a dragon" has experienced a dramatic fluctuation in win rate, soaring from 50% an hour ago to the current 90% (a fluctuation of 40%). Please note the impact of relevant breaking news.

According to Fox News, the U.S. Treasury Department has removed certain sanctions related to Iran from its website. A Treasury official stated, "This does not represent any change in U.S. policy toward the Iranian government, the Islamic Revolutionary Guard Corps Quds Force, any designated terrorist organizations or their supporters or agents. Today's removal is part of normal administrative procedures and was conducted after reviewing specific circumstances related to Fly Baghdad."Earlier, the U.S. Treasury Department's Office of Foreign Assets Control (OFAC) published an "Update on the Revocation of Terrorism Designation," lifting certain sanctions related to Iran, including removing Fly Baghdad from the sanctions list. Fly Baghdad was previously listed on the anti-terrorism sanctions list due to its association with the Iranian Islamic Revolutionary Guard Corps Quds Force.

Circle CEO Jeremy Allaire stated during the earnings call that the market is shifting from speculative cryptocurrencies to supporting an open global digital asset market for tokenized stocks, tokenized commodities, and other tokenized assets.

According to South Korea's Digital Daily, Apple has recently negotiated with Changxin Memory Technologies (CXMT) on the supply prices of mobile DRAM such as LPDDR5X to reduce costs for the next generation of iPhones and smart devices, but was rejected. Changxin Memory insists on maintaining a price level comparable to or even higher than that of Samsung Electronics and SK Hynix, breaking Apple's usual bargaining advantage as an end manufacturer.The report analyzes that Chinese terminal companies such as Huawei and Xiaomi have locked in a large amount of Changxin Memory's production capacity through long-term high-price contracts, providing a solid price defense line in the domestic market. Meanwhile, Samsung Electronics and SK Hynix are facing reduced supply of general DRAM due to the continuous expansion of HBM capacity, combined with Changxin Memory holding the price floor, significantly alleviating the shipping pressure and price decline risks for Korean manufacturers in the general DRAM sector, allowing them to focus resources on high-value AI storage products such as HBM4, LPCAMM2, and eSSD. Industry insiders expect that in the second half of the year, the operating profit margins of the Korean semiconductor camp and global market pricing power will further strengthen.

Fortune magazine recently published a commentary article pointing out that with the rapid rise of China's open-source AI models, the competitive landscape of the global AI field is changing. The real contest has shifted from U.S.-China rivalry to the battle between open-source and closed-source models. The article cites data from the independent evaluation platform Artificial Analysis, stating that the intelligence index of DeepSeek V4 Flash is only 1 point lower than that of GPT-5.6 Luna. Even though OpenAI has reduced prices by 80%, the former's single-task cost is still 60% lower.The author notes that while U.S. export control policies aim to limit China's AI development, they have objectively stimulated innovation at the architectural level among Chinese companies, forcing them to shift towards open-source and low-cost routes. Chinese models attract global developers to participate in improvements by opening weights, while transferring inference costs to overseas cloud service providers, achieving a low-cost global layout. Regarding the viewpoint that open-source models "cannot be profitable," the article argues that their profit model is similar to that of open-source software: users pay for hosting services, and most companies are not inclined to self-deploy.The article also points out that recently, several figures in the U.S. tech community, including former White House AI official David Sacks, have begun to support open-source, and NVIDIA CEO Jensen Huang has also called for the industry to embrace open-source. The author urges both China and the U.S. to cooperate in the field of AI security, viewing open-source AI as a "global public good," rather than falling into a zero-sum competition that costs trillions of dollars.

According to Jinshi reports, the CEO of Amazon Web Services (AWS) stated that the artificial intelligence business is transitioning from the training phase to the inference phase.

Ronald Keung, the head of Asian Internet Research at Goldman Sachs, stated that as the global usage of Chinese AI models surges, developers may begin to charge commercial licensing fees to cloud platforms for hosting their open-weight models. Currently, most Chinese models are released under permissive licenses such as the MIT License, allowing any user to use, modify, and host them for free, including for commercial purposes.Keung pointed out that Chinese AI models like Kimi K3 from the Dark Side of the Moon and GLM-5.2 from Zhiyu AI have performance levels that are close to those of leading models in the United States. The adoption rate among small and medium-sized enterprises, both domestically and globally, is very fast, and even large enterprises are beginning to consider using Chinese models. Goldman Sachs believes that model developers can convert the growing use of open-source into actual revenue by requiring third-party service providers to purchase commercial licenses to run model inference on their own infrastructure.

According to PRNewswire, Nasdaq-listed company Canton Strategic Holdings announced that it has completed the sale of its biotechnology research and development department, Gravitas Life Sciences, to Gravitas Collective Corp. The transaction was unanimously approved by the company's board of directors and was officially completed on July 17, 2026.This sale marks the company's further transformation into an operational enterprise focused on driving the digitization of traditional financial markets around the Canton Network, promoting asset tokenization and the digitization process of financial markets, as well as financial infrastructure business.

Bloomberg New Energy Finance (BNEF) latest forecast shows that by 2035, electricity consumption by data centers in the United States will account for about 20% of the nation's total electricity consumption, a significant increase from the current level of about 5.9%. The agency has raised its forecast for data center electricity demand in 2035 to 106 GW, which is 36% higher than the 78 GW predicted in April this year. Currently, the operating capacity of data centers in the U.S. is about 40 GW, accounting for approximately 3.5%-4% of the national electricity demand, while under BNEF's baseline scenario, this proportion is expected to reach 8.6% by 2035. The high-growth model from the Electric Power Research Institute (EPRI) indicates that if the combined effects of cryptocurrency mining and AI computing power are taken into account, the upper limit of this proportion also points to 20%.In response to the explosive growth in AI computing power demand, Bitcoin mining companies are actively transforming. Companies like Core Scientific and Riot Platforms have partnered with tech giants such as AWS and Google to convert their existing mining sites into AI data centers. Currently, Bitcoin mining companies have secured about 6 GW of electricity capacity, which is expected to expand to 12 GW by 2027, with some analysts estimating that about 20% of mining companies' computing power capacity will shift towards AI workloads by then. Data from the Electric Reliability Council of Texas (ERCOT) shows that data centers now account for about 90% of local large load applications, with many sites originally used for cryptocurrency mining being repurposed as AI computing facilities. This trend is also directly reflected in the capital markets, as Core Scientific has seen a significant rebound in its stock price after emerging from bankruptcy and partnering with AI cloud service provider CoreWeave.

According to CoinDesk, cryptocurrency wallet company Exodus Movement (EXOD) announced that it will cut about 25% of its global workforce to reduce operating costs and shift its business towards stablecoin payments and card payment infrastructure.Exodus stated in regulatory filings that this restructuring is part of the company's strategy to build a full-stack payment platform. The company has been continuously integrating the acquired electronic money institution Monavate and cryptocurrency payment company Baanx to expand its payment capabilities and global business layout.Exodus expects that this round of layoffs will result in pre-tax restructuring costs of $2.5 million to $3.5 million, mainly for employee severance and related costs. Affected employees will receive severance compensation, benefits continuation, and transitional support.The company anticipates that after the restructuring is completed, it can reduce cash operating expenses by $10 million to $13 million annually, with related savings fully realized by 2027.In the market, EXOD rose about 2.2% in pre-market trading on Monday, but its cumulative decline over the past year is still close to 85%. This adjustment reflects that cryptocurrency companies are accelerating their transformation towards stablecoin payments, financial infrastructure, and other business directions.

Tom Lee stated that as AI "bottleneck" related stocks adjust, AI downstream assets are strengthening, including SMH, DRAM, MAGS, IGV, as well as AI downstream related ETHA and Ethereum ETH.He pointed out that over the past month, Ethereum has outperformed DRAM by about 5,500 basis points, or 55%.

Coinbase co-founder and CEO Brian Armstrong posted on the X platform, publicly acknowledging that the creator coin previously promoted by the Base chain has not succeeded and that a strategic adjustment was completed earlier this year.Armstrong stated that Base's current core focus areas are trading, payments, and AI agents, which are interconnected—payments require foreign exchange trading, and AI agents will also involve a large number of trading and payment scenarios. Currently, most resources have been concentrated on the trading track.