SemiAnalysis预计ASML将上调长期指引
Related
SemiAnalysis: Is TTFT overrated? The bottleneck of large model inference lies in the trade-off between interactivity and throughput.
According to an article published by SemiAnalysis, as reported by Odaily Odaily, the "time-to-first-token" (TTFT) parameter, frequently cited in large-scale model inference, may be overemphasized by the market. The core dimension of an inference system is not "the faster the better," but rather a trade-off between "single-user interaction speed (tok/s/user)" and "overall throughput efficiency (tok/s/GPU)." In most scenarios, TTFT has a limited impact on the user experience, but the token generation speed is the key variable. SemiAnalysis concludes that only about 10%–20% of inference tasks are truly limited by latency, while most other scenarios rely more on throughput and cost-efficiency optimization.
SemiAnalysis: Nvidia's system memory spending is projected to exceed 40% by 2027.
Odaily Odaily reports that SemiAnalysis, in a post on its X platform, stated that the proportion of memory in Hyperscaler capital expenditures has sparked considerable discussion, especially after MU released its financial report last week. Some market participants were surprised by the level this proportion might reach next year. SemiAnalysis released its initial view in late February, and many clients disputed its 30% figure. In May, after pricing increases exceeded expectations, SemiAnalysis responded that, after combining DRAM, NAND, and HBM, memory spending in Nvidia systems would exceed 30% by the end of 2026 and surpass 40% in 2027. SemiAnalysis expects this dynamic to become more clear in the coming months.
SemiAnalysis: SPHBM4 shifts the complex engineering burden of AI chips to the substrate layer
According to a Odaily by SemiAnalysis on the X platform, SPHBM4 will shift the complex engineering burden of AI chips. Chip manufacturers will no longer purchase the extremely expensive proprietary "silicon interposer + ABF substrate" combination, but will instead turn to purchasing ultra-large, high-layer-count ABF substrates, or adopt glass substrates ahead of time when performance requirements are directly pushed to the substrate layer. The substrate boom has just begun.
Goldman Sachs raises ASML's target price from €1,770 to €2,000.
Odaily Odaily reports that Goldman Sachs has raised its target price for ASML from €1,770 to €2,000. (Jinshi)
Citrini vs. SemiAnalysis? SemiAnalysis founder challenges Citrini researcher, claiming he revealed information about the Rubin Ultra 12 Hi to clients back in March.
According to Odaily Odaily, a new debate has erupted between two well-known investment research firms, Citrini and SemiAnalysis. Citrini researcher Zephyr cited sources claiming that "Rubin Ultra HBM performance has actually dropped to the 12 Hi level, and they haven't even started using HB technology yet." They will not be using 16-Hi chips in 2027. Currently, hybrid bonding technology is very expensive for HBM (its yield is extremely low). BESI's biggest advantage in the short term lies in TSMC's use of hybrid bonding technology in its chip manufacturing and packaging processes, enabling the integration of EICs and PICs, thereby improving CPO production efficiency. In response, Citrini researcher Jukan commented with a related tweet, "I told you this back in April." In response to this, Dylan Patel, founder of SemiAnalysis, forwarded a screenshot from Discord and said: "In March, customers using the SemiAnalysis Memory and Accelerator models already knew that the Rubin Ultra 16 Hi would be replaced by a 12 Hi version." Previously, Citrini researcher Jukan cited reports that "Nvidia plans to drastically cut its next-generation rack memory configurations," causing a sharp drop in US and South Korean memory stocks. SemiAnalysis founder Dylan Patel clarified that the report was taken out of context and was sensationalist.
MGBX will launch trading pairs for BMNRUSDT and ASMLUSDT perpetual contracts.
According to official Odaily, MGBX will launch the BMNRUSDT and ASMLUSDT perpetual contract trading pairs on June 18, 2026 at 18:00 (SGT). Trading will open at 18:00 (SGT) on June 18, 2026. Leverage ratio: Maximum support 20x BMNR: BitMine Immersion Technologies Inc. is a Bitcoin and Ethereum network company focused on long-term cryptocurrency accumulation, whether through its Bitcoin mining operations or proceeds from capital raising transactions. ASML: ASML is a leader in lithography systems, holding a 90% market share, used in the manufacture of semiconductors.