chip

On-chain data analyst Murphy pointed out that BTC purchased in 2025 has been in a state of unrealized loss until now. Currently, there are about 4.77 million coins from 2025, a decrease of 41.5% from the peak in December last year. The decline slope shows two phases: a rapid decline before February and a slowdown after February, but it still maintains a certain slope. This group is currently the largest supply side in the market.Comparing the data from 2024, 2023, and 2022, it is found that the old coins with unrealized gains have basically passed the steep decline phase; the longer the time, the smaller the slope, and the curve has almost flattened after February. Even if the price drops further, the changes in the number of these coins are not significant, and those that should have changed hands have already done so. In terms of historical data, during the bear bottom in 2022, the high-position coins from 2021 decreased by 51%, and during the bear bottom in 2018, the high-position coins from 2017 decreased by 62%.Murphy believes that if we simply refer to history, the maximum decline of the 2025 coins in this bear bottom may be between 50%-60% (currently at 41%), and this does not yet account for the BTC purchased by institutions such as ETFs and MicroStrategy in 2025, most of which are in a locked state.

According to South Korean media ZDNet Korea, Samsung Electronics is adjusting the original purpose of its NRD-K Line 2 at the Giheung campus from a research and development facility to a wafer foundry line, with the main target product being the 2nm base chips required for NVIDIA's HBM. This production line is expected to be operational in the second half of 2028, adopting a "Send-Fab" model, which involves receiving wafers from other mass production lines and performing specific processes.NRD-K is the most advanced composite research and development campus built by Samsung Electronics to seize future semiconductor technology, with a total investment of approximately 20 trillion won, planning for three production lines, of which the first line is scheduled to be completed by the end of 2024. This adjustment aims to proactively respond to the expansion of HBM demand driven by investments in AI infrastructure. Samsung plans to first apply the 2nm process to custom HBM and HBM5 to maintain performance advantages.Industry insiders point out that the scale of NRD-K Line 2 is relatively small, making it suitable for operation in the Send-Fab format. However, there are still about two years until the production line is operational, and the final purpose may be adjusted again based on market conditions. Samsung had previously considered using this production line for DRAM production, but after modifications to the relevant plans this month, DRAM capacity will be more concentrated in the Pyeongtaek campus. Equipment procurement orders have not yet been officially issued.

On-chain analyst Murphy stated that currently, all chips bought in BTC in 2025 are basically in a state of loss. Therefore, apart from wallet migrations, the reduction in the scale of 2025 chips likely means that holders are cutting losses and selling. Data shows that as of now, approximately 4.77 million BTC chips bought in 2025 remain, a decrease of 41.5% from the peak in December last year.From the downward trend, this group has experienced two phases: a rapid decrease in chips before February this year, and a significant slowdown in the rate of decline after February, although it still maintains a certain slope. Murphy believes that the 2025 chips may be the largest potential supply side in the current market scale. In contrast, the BTC chips formed in 2024, 2023, and 2022 have basically completed the release of high-level locked positions due to still having unrealized gains, and the slope of the curve is gradually flattening, indicating that the selling pressure from long-term holders is weakening.Historical data shows that during the bottom phases of the past two bear markets, high-level chips have shown a significant decline: at the bottom of the bear market in 2022, the chips bought at high levels in 2021 decreased by about 51%; at the bottom of the bear market in 2018, the chips bought at high levels in 2017 decreased by about 62%. If we refer to historical cycles, Murphy believes that in this bear market bottom phase, the reduction of high-level chips in 2025 may be in the range of 50%-60%, and the current decline of 41.5% indicates that there is still some room for release. However, this judgment has not yet considered the BTC bought by institutional investors such as spot ETFs and MicroStrategy, as this portion of chips is mostly in a long-term locked state, which may reduce the actual market supply pressure.

Zhao Haijun, Co-CEO of SMIC, stated that the demand for AI-supported chips has increased by 40%.

The CEO of Lenovo stated that Lenovo will collaborate with NVIDIA (NVDA.O) and plans to launch AI PCs equipped with RTX chips later this year.

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

According to ChosunBiz, Samsung Electronics' System LSI Division has introduced Anthropic's AI programming tool Claude Code for semiconductor design and verification work, with some task cycles shortened to less than one-tenth of the original time. In the verification of data connection structures for customer-customized SoCs, work that was originally expected to take over a month was completed in two days; an engineer with two years of experience completed USB device model development and Android driver adaptation in one day using Claude Code, whereas similar work typically required a month.Samsung opened Claude Code to software developers in May this year and later expanded it to the semiconductor professional development field. The System LSI Division has about 6,000 employees, while its main competitor Qualcomm has about 52,000 employees. Samsung is trying to bridge the manpower gap with AI tools. Meanwhile, Samsung's DS Division is promoting an "AI transformation," extending generative AI from research and development to the entire business process, including production, marketing, and support. The report also pointed out that Claude has exhibited behavior in some verification tasks where it disguised erroneous information as ordinary messages or modified design code without authorization. Samsung emphasized the need for strict control and manual review of the scope and results of AI tasks.

The number of registered users on the Zhipu MaaS open platform (i.e., API users) has approached 7 million, an increase of about 2 million compared to early July, including 23,000 enterprise customers. The user count for the developer product ZCode, which competes with Codex, surpassed one million within one month of its launch.Since the beginning of this year, Zhipu's ARR has grown 15 times. It is understood that after the release of new models by Kimi and Alibaba, Zhipu's ARR growth rate has not been affected. Some investors have revealed that the current ARR has reached $2 billion, but Zhipu's official stance has denied this, with people close to the company indicating that the actual figure should be higher. On July 21, it was reported that Zhipu has built a domestic AI computing power infrastructure with a scale of 1GW. It is understood that more than 50,000 domestic computing chips have been activated to alleviate the growing inference demand. On July 31, Zhipu's long-term Coding Plan, which had significantly increased in price, was opened for purchase.

Microsoft plans to publicly release its next-generation Maia 300 AI chip this fall (possibly as early as September). To meet delivery demands for 2027, Microsoft has engaged with TSMC to secure manufacturing capacity for over 300,000 units of the chip.

According to The Information, Microsoft's self-developed artificial intelligence chip project has shown signs of recovery after experiencing a slow start.

Discovered Materials completed a $9 million seed round financing, with investors including Lightspeed India Partners, Peak XV Partners, and several angel investors.

According to data from the Bank of Korea, South Korea's current account surplus in June reached 49.73 billion USD, setting a monthly record for the second consecutive month, further expanding from 38.61 billion USD in May, and compared to 13.97 billion USD in the same period last year. The cumulative surplus for the first half of the year was 191.01 billion USD, the highest in history for the same period, exceeding the Bank of Korea's previous annual target of 250 billion USD by 76%. The goods trade surplus of 47.89 billion USD also set a monthly record, with exports increasing by 84.5% year-on-year to 112.37 billion USD, surpassing the 100 billion USD mark for the first time, among which chip exports soared by 196.9%, and computer peripherals grew by 282.7%.South Korea's current account has maintained a surplus for 38 consecutive months since May 2023. The services account deficit widened to 1.29 billion USD, mainly influenced by increased payments related to intellectual property; the travel account surplus was 440 million USD, the second highest since October 2008. The current account surplus for South Korea in 2025 is projected to be 123.05 billion USD, setting a historical annual record.

Crypto analyst Murphy stated that the concentration of Bitcoin chips has risen to 14.8%, approaching a historically high-risk area. The concentration of chips itself cannot determine the future direction of BTC prices, but historical data shows that when the curve begins to turn, if BTC prices have risen previously, the probability of continuing to rise is greater; conversely, the probability of continuing to fall is greater.Currently, this indicator is still in an upward phase, so it is temporarily impossible to determine the direction of the next round of fluctuations. However, market data shows that as the concentration of chips continues to increase, potential volatility risks are accumulating, and the BTC market may face larger price changes.

According to the latest supply chain survey by TrendForce, despite the United States implementing export controls on high-end GPUs, China's AI industry has not slowed down; instead, it is accelerating the construction of a domestic AI supply chain. The report points out that the supply structure of high-end AI chips in China is shifting from reliance on imported GPUs to a dual-track model of "domestic GPUs plus self-developed ASICs," with domestic solutions expected to capture nearly 90% of the Chinese high-end AI chip market by 2026.This year, the Chinese government is actively promoting the adoption of domestic AI chips, with policies prioritizing support for high-potential local manufacturers to significantly expand their market share in the high-end AI server market. Meanwhile, the domestic ecosystem is gradually maturing in key areas such as advanced processes, advanced packaging, and thermal management. TrendForce expects that, against the backdrop of an accelerated domestic substitution process, domestic chip manufacturers like Huawei Ascend, Cambricon, and Haiguang are likely to continue benefiting, and the related industrial chain will usher in structural growth opportunities.

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.

According to The Wall Street Journal, Apple is testing memory chips from the Chinese DRAM manufacturer Changxin Storage (CXMT), planning to apply them to product lines such as the iPhone and MacBook. The background of this move is the AI demand driving up global memory prices; Apple has previously lobbied the U.S. side, stating that the current pricing from Korean and American suppliers is "unsustainable." The report states that Apple is seeking approval from the White House to conduct business with Chinese chip suppliers. Currently, the U.S. allows the purchase of commercial off-the-shelf chips but restricts customized products.Changxin Storage has recently rapidly expanded its market share, with second-quarter revenue growing more than 8 times year-on-year, and its global DRAM market share rising to 7%, completing its IPO on the Science and Technology Innovation Board last month. Meanwhile, PC manufacturers such as HP and Acer have begun using Changxin Storage memory chips in some devices sold in non-U.S. markets to alleviate supply pressure. If Apple ultimately reaches a procurement agreement, it may face political resistance from Washington.

According to Bloomberg, informed sources revealed that the U.S. government department responsible for investigating violations of chip export controls is currently conducting a systematic review of how Chinese AI companies obtain advanced chips from Nvidia through legal means, focusing on their behavior of renting computing power in third countries. This move stems from recent technological breakthroughs achieved by Chinese AI companies, indicating that they still have the capability to acquire and use top-tier hardware despite U.S. export restrictions on chips to China. Analysts believe that this review may prompt the U.S. to further tighten regulatory rules on computing power leasing in third countries to close existing loopholes in the export control system.

According to Gate market data, Microchip Technology rose 9% in pre-market trading as the company forecasted that both revenue and profit for the second quarter would exceed expectations.

SK Hynix plans to invest 35.2 trillion won in the second phase of chip factory construction in Yongin, South Korea.