semi

According to the Industrial and Commercial Times, with the expansion of HBM and advanced logic below 2 nanometers, it is estimated that the global silicon wafer supply-demand gap will widen from about 8% by the end of 2026 to 15% in 2027 and 24% in 2028. GlobalWafers, Siltronic, and TSMC are expected to benefit from this. It is noted that mid-2027 will be a critical turning point; if the pace of supply expansion does not keep up, a new wave of shortages and rush buying may occur from the second half of 2027 to the first half of 2028.SEMI statistics show that the global silicon wafer shipment area reached 3.573 billion square inches in the second quarter of 2026, an increase of 9.1% quarter-on-quarter and 7.4% year-on-year. It is estimated that the global monthly production capacity of 12-inch silicon wafers will be about 10.69 million pieces in 2026, an increase of 7% year-on-year, with only about a 3% increase to 11 million pieces in 2027, and an increase of about 6% to 11.7 million to 11.76 million pieces in 2028. The new supply includes about 300,000 pieces from the second phase of GlobalWafers and about 100,000 pieces from Siltronic.It is estimated that the consumption of HBM silicon wafers will increase by 23% and 28% in 2027 and 2028, respectively; new general DRAM production capacity will gradually come online starting in 2027, with an additional monthly capacity of about 350,000 to 450,000 pieces by 2028. In terms of advanced logic, the consumption of logic process silicon wafers is expected to increase by 12% and 11% in 2027 and 2028, respectively. Currently, customer silicon wafer inventory remains at about 100 to 120 days, and it will take time to digest to the normal level of 50 to 70 days.

According to Gate market data, the Philadelphia Semiconductor Index opened down 5.5%.

Singularity Research posted on X that Citrini has been acquired by SemiAnalysis, with the terms of the deal undisclosed. The agency initially estimated the transaction price to be around 30 million USD, later updating that the actual amount could be close to 100 million USD, but emphasized that there is no insider information, and it is merely speculation.Citrini was founded in 2023 and has a Substack newsletter and institutional research products, with over 260,000 subscribers. Its annual revenue is estimated to be around 5.9 million USD. Citrini completed a 5.05 million USD SAFE round of financing in December 2025, and if calculated with approximately 20% dilution, the post-money valuation is about 25 million USD.

According to South Korea's "JoongAng Ilbo," China's semiconductor industry is organized through collaboration among the central government, local governments, state-owned financial institutions, and enterprises, forming a system for fundraising, factory construction, and infrastructure support. The China National Integrated Circuit Industry Investment Fund has a first phase of 138.7 billion RMB, a second phase of 204 billion RMB, and a third phase of 344 billion RMB, totaling 686.7 billion RMB, with investments in companies such as SMIC, Hua Hong Semiconductor, and Yangtze Memory Technologies.Changxin Memory, established for 10 years, has risen to fourth in global DRAM market share, with a net loss of 21.13 billion RMB over the past three years, while R&D and equipment investments during the same period reached 185.2 billion RMB, with Hefei's state-owned capital providing about 80% of the funding for early projects. South Korea's Samsung Electronics and SK Hynix are responsible for investments, employment, technology development, and support for partners. The South Korean government plans to guide 622 trillion KRW in private investment by 2047 and provide 17 trillion KRW in low-interest loans and 1.1 trillion KRW in semiconductor ecosystem funds, but most of the burden will fall on enterprises.Park Kyung-soo, Executive Vice President of the Korea Semiconductor Industry Association: The semiconductor industry will fall behind in the next upturn if investment stops; even during downturns, R&D, equipment investment, and orders for materials and components must be maintained, requiring long-term policy, financial, and tax support. China is catching up with a national-level support system, and South Korea also needs a response system involving the government, political circles, financial institutions, and enterprises to act together.

Tesla (TSLA.O) announced that its Semi Truck is about to enter the European market.

On September 10, Kim Sang-hoon, the head of the packaging division at Samsung Electro-Mechanics, delivered a speech at the KPCA Show International Seminar INSIGHT 2026 held at the Incheon Songdo Conference Center, proposing a multilayer core (MLC) technology direction to address the large-scale and high-multilayer challenges of AI semiconductor packaging substrates.Kim Sang-hoon stated that in the past, the single-layer core (SLC) method, which increased the thickness of a single CCL, led to increased difficulty in processing internal vias and circuits as the core thickened, and the size of the vias became limited. The MLC approach shifts to a combination of CCL and PPG multilayers, or even a mixed structure of two CCLs, which can simultaneously enhance rigidity and wiring freedom, while also allowing for ground or signal wiring to be arranged in the PPG layer. A bonding layer added between the two CCLs can provide more stable support for the substrate, but the process is more complex.Currently, the commonly used bump pitch in mass production is above 90μm, with around 85μm being the turning point for printed fine solder ball processes. The difficulty significantly increases around 80μm, and the 55-45μm range may shift towards metal bumps. Glass cores are another option, offering higher rigidity and lower thermal expansion coefficients, which help reduce warping in large packages, but issues related to fragility and micro-hole plating technology need to be addressed. The substrate must meet signal integrity, power integrity, and mechanical integrity requirements simultaneously, with high-performance packaging substrate layers expected to evolve from about 20 layers at the 90mm level to 120mm and even more than 40 layers.

Market news says that Samsung will integrate Mistral AI into its semiconductor business.

On September 8, South Korean semiconductor concept stocks showed mixed performance, with related asset markets remaining active. According to data from the Gate platform, SK Hynix (000660) is currently priced at $1,332.38, with a 24-hour increase of 0.56%; Samsung Electronics (005930) is currently priced at $200.26, with a 24-hour decrease of 0.18%.According to CoinGlass data, both SK Hynix and Samsung have high trading activity in the contract market. Among them, trading of related stock derivatives on Gate is active: the SKHX contract has a position amount of nearly $158 million, ranking second across the network, with a 24-hour trading volume of nearly $177 million; the Samsung contract has a position amount of $9.51 million, with a 24-hour trading volume of $15.8 million, both ranking second across the network.

According to South Korea's "Cultural Daily," Samsung Electronics' wafer fab in Taylor, Texas, is expected to enter trial operation around the end of this month, as orders for 2-nanometer chips from Tesla, Broadcom, Arm, and others continue to pour in, with pre-production capacity nearing full load. The plant has accumulated an investment of $37 billion since 2022, with plans to begin trial operation early next month and mass production next year, with an initial monthly capacity of about 50,000 wafers, close to TSMC's approximately 80,000 wafers.The report states that the Taylor plant will produce Tesla's next-generation artificial intelligence chip AI5, Broadcom's new generation communication chips, and Arm's smart device AI chips, among others. Samsung's 2-nanometer yield has increased from about 60% earlier this year to around 80% recently. The company sent personnel for process, equipment, and yield to the plant in June, and began installing equipment in July. At a recent earnings meeting, Samsung projected that this year's 2-nanometer order volume would more than double compared to last year, and stated that NVIDIA's inference chip Grok3 began mass production last month; Samsung also recently raised the foundry price for 4-nanometer chips by up to 15%.The second Taylor plant is scheduled to start construction by the end of this year and begin mass production in 2030. Samsung has requested major partners to complete the relevant production equipment safety certifications in advance, and some partners are considering sending dozens of personnel to the local area. At the same time, Intel issued about $20 billion in new shares last month and expanded investments in 2-nanometer level 18A and 1.4-nanometer level 14A, while Japan's Rapidus completed its 2-nanometer factory in Hokkaido last month and entered the wafer trial production stage.

Taiwan Semiconductor Manufacturing Company, Micron, and other companies are increasing capital expenditures to expand production, driving the combined orders of the top five semiconductor engineering firms—HanTang, Axiom, FanXuan, Yankee, and ShengHui—to exceed 880 billion yuan, a record high. Taiwan Semiconductor Manufacturing Company recently stated at a semiconductor exhibition that it is building up to 20 wafer fabs, with the overall capacity expansion scale increasing multiple times compared to the past, but it still cannot meet customer demand. The U.S. tariff policy has driven the demand for manufacturing plants in the United States. Axiom has the largest order amount of 440.73 billion yuan, and Chairman Yao ZuXiang pointed out that the cumulative amount of turnkey projects undertaken in Singapore over the past four years has reached 600 billion yuan, with expectations for new projects to follow. HanTang's order amount is approximately 193.937 billion yuan, setting a new record, benefiting from continued plant construction by major clients like Taiwan Semiconductor Manufacturing Company and Micron.FanXuan's order amount reached a new high of 135.1 billion yuan, and Chairman Gao XinMing revealed that order visibility extends at least to 2028, with related projects for clients planned for 2029 and 2030. FanXuan has deployed materials, manpower, and local construction teams to support clients in Taiwan, Arizona in the United States, Japan, and Germany in synchronizing production expansion needs, and is investing in the development of technologies such as CoPoS. ShengHui's order amount exceeds 60 billion yuan, with Taiwan accounting for 68% and semiconductor orders accounting for 63%. In the first half of the year, the after-tax net profit was 2.944 billion yuan, with earnings per share of 23.73 yuan, setting a new high for the same period. Yankee's after-tax net profit in the first half of the year was 2.317 billion yuan, with earnings per share of 17.46 yuan, and the order amount is approximately 51.77 billion yuan, with order visibility reaching the end of 2027.

The supply chain reports that Amazon, a leading North American cloud service provider under AWS, has recently expanded the delivery of AI server cabinets, with strong shipping momentum from partners such as TSMC, World Advanced, Wistron, and Hon Hai. Amazon has recently revised its capital expenditure for this year, increasing the original estimate from $200 billion to $220 billion, primarily for AI infrastructure, including data centers, servers, and network equipment.Amazon CEO Andy Jassy is optimistic that the demand for AI computing power will continue to exceed supply until 2027, and he has even seen strong demand for 2028. AWS's AI business and self-developed chip scale have grown to $25 billion, maintaining triple-digit percentage growth. Annapurna Labs' customized ASIC chips are mostly produced by TSMC using advanced 3-nanometer processes.World Advanced is an important partner for AWS's self-developed AI chips and is optimistic that its quarterly revenue and profits will reach new highs this season, with continued growth expected in the fourth quarter. The demand for 3-nanometer AI accelerators is better than previously assessed, and designs can be finalized before the end of the year. AWS's ASIC AI cabinets are assembled by Wistron, which also supplies switch trays, while Hon Hai is involved in CPU trays. Hon Hai's rotating CEO, Liu Young-way, stated that this year's ASIC project has made new progress, setting a target for ASIC server market share at over 40%.

According to Ai Yi's monitoring, the top spot on the Binance contract real smart money 30-day leaderboard has changed hands, with trader Semisemi earning over 10.78 million USD in the past 30 days. Among them, the BTC long position has realized profits of over 8.06 million USD, and the ETH long position has realized profits of over 3.91 million USD, with the main earnings contributed by BTC and ETH trading.Currently, Semisemi still holds BTC and ETH long positions worth 95.59 million USD, with a total unrealized loss of approximately 1.885 million USD.

Today, South Korean semiconductor concept stocks are under pressure and have retreated, but trading in related asset markets remains active. According to data from the Gate platform, SK Hynix (000660) is currently priced at $1,175.12, with a 24-hour decline of 2.11%; Samsung Electronics (005930) is currently priced at $183.02, with a 24-hour decline of 1.94%.According to CoinGlass data, trading in the SKHYNIX and SAMSUNG contract markets is active, with the open interest of the two underlying contracts increasing by 4.03% and 7.30%, respectively. The trading volume of the contracts increased by 451.63% and 1069.65%, respectively. In Gate's related stock derivatives, the SKHX contract open interest reached $180 million, ranking second in the entire network, with a 24-hour trading volume of $180 million; the SAMSUNG contract open interest reached $8.14 million, ranking second in the entire network, with a 24-hour trading volume of $5.94 million.

SemiAnalysis released the latest episode of its podcast, which pointed out that SK Hynix and Micron have been facing challenges in achieving the highest HBM4 speeds. In particular, SK Hynix had to redesign its base chip, resulting in delays in HBM4 deliveries to the NVIDIA Rubin platform.Samsung currently has the best HBM4 technology. Broadcom has traditionally relied heavily on Samsung for its HBM supply. Samsung's performance in the HBM3E generation was poor, which posed a disadvantage for Broadcom's ASICs. However, now Samsung is leading its competitors with HBM4, and this same supply chain relationship has become an advantage for Jalapeño.

The semiconductor and AI independent research organization SemiAnalysis released a deep security report on Neocloud (new cloud), revealing various cross-tenant security vulnerabilities discovered during the ClusterMAX 3 testing period. In a four-month test covering 25 vendors and 32 clusters, the team achieved multiple instances of cross-tenant remote code execution (RCE) solely by exploiting publicly known vulnerabilities and basic configuration checks. Affected entities included banks, telecommunications companies, universities, research institutions, AI laboratories, and even a national intelligence agency.Typical issues included: shared Kubernetes control plane leading to tenant metadata visibility, container escape, exposure of BMC/IPMI management networks, incorrect configuration of InfiniBand security keys (P_Key, SA_Key, M_Key), unfortified default trust mode of BlueField DPU, Grafana monitoring dashboards using god-level API keys, and lack of VXLAN isolation in front-end networks. The report specifically pointed out a cascading vulnerability case: a misconfiguration of shared vCluster combined with software versions being two years out of date ultimately completed the POC verification of cross-tenant RCE within an afternoon.Notably, the report questioned the mainstream narrative that "AI has fundamentally changed the pace of cybersecurity": statistics on CVEs for NVIDIA GPU drivers, CUDA, PyTorch, Kubernetes, Docker, and the Linux kernel showed that there was no significant increase in vulnerabilities after the popularization of AI coding models, with most data supporting the "no change hypothesis." The report also detailed the incident where an OpenAI-trained agent attacked Hugging Face, where the AI agent achieved cluster-level privilege escalation through a message board established via Artifactory, which went undetected from May to July. While building POC verification for existing vulnerabilities, the team found that Claude Fable and GPT-5.6 Sol frequently rejected security-related requests, ultimately relying on open-source models such as DeepSeek V4, Kimi K3, and GLM-5.2 to complete the task.SemiAnalysis stated that the core issue in the Neocloud (new cloud) industry is not the new risks brought by AI, but rather the long-term absence of basic patch management, tenant isolation, and security design. They recommended that vendors establish automated security announcement monitoring systems and rectify single points of failure that could expose all users' architectural patterns.

P Equity Research and SemiAnalysis researcher Nick Doyle and others discussed high bandwidth flash (HBF) in X Space. Nick stated that it is still too early to determine how much the cost premium of HBF relative to HBM can shrink; existing data mostly comes from vendor claims, such as Sandisk stating that the cost per bit is about one-eighth that of HBM. Yields, testing, and other factors will improve with scale, but structural costs such as TSV, stacking, and pSLC mode will always exist, and durability is a key unknown; if wear exceeds expectations, costs will rise.The application scenarios for HBF are narrow, targeting only AI inference, especially low batch and long context MoE models, and it is not a substitute for HBM. The actual bandwidth target is about 1.6 TB/s, which is at the HBM3E level, suitable for sequential reads to load model weights, more aligned with the capacity needs of a small number of GPUs in local or private enterprises, rather than ultra-large-scale bandwidth scenarios. Heat dissipation reliability has not yet been resolved; flash memory will degrade faster at high temperatures next to GPUs, and mitigation measures such as UCIe separation and daily refresh have yet to be validated.In terms of manufacturing, Sandisk/Kioxia has experience with 3D NAND, and SK Hynix complements HBM-style stacking capabilities, but mass production is still to be confirmed. Overall, storage is shifting towards a specialized layered market, with NAND shortages expected to continue until 2028, and HBF may further impact supply and demand.

According to eight people familiar with the discussions, despite warnings from tech companies that this move could affect the United States' dominance in the field of artificial intelligence, the Trump administration is still considering imposing a new round of comprehensive tariffs on semiconductors.

According to SemiWiki, WSTS data shows that the global semiconductor market sales reached $368 billion in the second quarter of 2026, a quarter-on-quarter increase of 35% and a year-on-year increase of 104%, setting a new historical high. The growth is mainly driven by AI demand, with significant quarter-on-quarter increases from memory manufacturers, SK Hynix and Sandisk at 51%, Kioxia at 76%; Nvidia's AI processor business grew 18% quarter-on-quarter, and Broadcom is expected to grow 33% quarter-on-quarter.Kioxia and Sandisk rose to eighth and ninth place globally, respectively, with both companies' revenues growing more than four times year-on-year. Excluding major AI suppliers, the other top 20 semiconductor manufacturers only saw a quarter-on-quarter increase of 9%, with guidance indicating a total quarter-on-quarter growth of about 6% for the third quarter of 2026. The growth expectations for memory manufacturers in the third quarter have clearly slowed down.Based on the latest forecasts, the semiconductor market growth rate for the entire year of 2026 is expected to reach or exceed 90%, with Semiconductor Intelligence predicting a growth of 101% and RCD Advisors predicting 112%; in the storage sector, WSTS expects a growth of 249%, while Gartner predicts 280%. Most institutions expect the growth rate to fall back to the mid-20% range in 2027.