Micron Technology, Inc. announced a strategic agreement with Anthropic that spans memory and storage AI architecture design, supply and demand, enterprise adoption of Claude across Micron and a strategic investment in Anthropic’s Series H funding round. The agreement directly links the demands of frontier AI models to how infrastructure is designed, supplied, and deployed at scale.
SK hynix is putting more weight on targeting the commodity DRAM market while adjusting the pace of expanding mass production of 6th-generation high bandwidth memory (HBM4). With HBM already accounting for more than 40% of revenue and establishing an overwhelming lead, the company says it is reallocating resources to secure additional revenue from commodity DRAM, where the supply shortage is severe, rather than engaging in excessive capacity expansion races.
Samsung Electronics recently held a three-day global strategy meeting. According to industry sources cited by Korean media, the meeting primarily focused on in-depth discussions regarding the expansion of High Bandwidth Memory (HBM) sales and the Long-Term Agreement (LTA) strategy.
According to industry insiders' analysis, Google is developing an upgraded v9 chip, likely codenamed Triggerfish, based on TPU v9 / Humufish, with MediaTek exclusively securing this new, higher-priced order. The key differences between this v9 variant and Humufish are: SRAM capacity is significantly increased to 2–3 times that of Humufish, a new simulation die is added, and memory is upgraded to HBM4E, versus HBM4 on Humufish. With Humufish lifetime shipments still estimated at 4-5 million units, Google is adding an incremental Triggerfish order of 1-2 million units, with production expected to begin in late 2027 and ramp in volume in 2028.
According to South Korean media reports, Samsung is currently collaborating with multiple partners to develop equipment for mass-producing its seventh-generation 10‑nanometer (1d) DRAM, with plans to roll it out in the second or third quarter of next year. Given the time required for actual mass-production preparation, initial volume production of 1d DRAM is expected to commence as early as the end of next year. Samsung's 1d DRAM is anticipated to be used as the core chip for the ninth-generation high-bandwidth memory (HBM5E), which aims to commercialize in 2029.
According to media reports, ASUS launches ET900N G3 workstation,which is already listed at several retailers in the U.S. The price starts from $99,999. ET900N G3 is equipped it with an Nvidia Grace Blackwell Ultra desktop superchip, which features 72 ARM Neoverse V2 processor cores and a Blackwell Ultra graphics chip. The processor is paired with 496 GB of LPDDR5X RAM with a bandwidth of 396 GB/s, while the graphics chip can access 252 GB of HBM3e VRAM with an impressive bandwidth of 7.1 TB/s. The memory pool, which totals 748 GB, can be combined to process larger AI models.
HBM4E is the successor to HBM4 (6th generation), which is currently in mass production. It is expected to be used in NVIDIA's next-generation AI accelerator, "Rubin Ultra," slated for release next year.
According to South Korean media citing SK hynix and fire authorities on the 12th, a fire broke out around 9:55 a.m. (local time) on the second floor of Building M15X, located within SK Hynix's Cheongju plant in Heungdeok District, Cheongju, North Chungcheong Province. After the fire broke out, employees from M15 and M15X were immediately evacuated outdoors. Reports indicate that the company's firefighting team extinguished the fire in its early stages. Cheongju issued a disaster alert, advising nearby residents to avoid the area and vehicles to take detours due to concerns over potential fluoride leaks. According to the latest news, the fire department stated that the investigation found no gas leakage. SK hynix stated that initial control has been completed, and the company is investigating the specific circumstances, adding that production equipment is operating normally. A fire also broke out at SK hynix's Cheongju Site 4 on the 1st of this month.
According to Korean media reports, Samsung Electronics is pushing ahead with the construction of an advanced semiconductor packaging (back-end process) plant in Gwangju, marking the company's first semiconductor production base in the Honam region (its current facilities are mainly concentrated in the Chungcheong region). This will be Samsung Electronics' first new packaging facility in 35 years since the establishment of its Onyang plant. The project is also expected to include front-end production lines for DRAM and NAND flash memory. The investment plan is reportedly scheduled for discussion on the 29th during a meeting chaired by President Lee Jae-myung, which will include heads of major corporate groups including Samsung.
According to Korean media reports citing industry sources, thanks to Samsung Electronics' advanced 2nm process technology and increased output of HBM chips, coupled with improved yield rates and large-volume orders, the foundry division is expected to turn a profit as early as the third quarter of this year. This moves up its profitability target from the original timeline of late this year or next year. After posting trillions of won in cumulative losses since 2022, the division may return to profitability for the first time in roughly four years.
With its eighth-generation TPU, Google breaks from the tradition of single-chip iteration by splitting training and inference into two distinct chips: the TPU 8t designed for large-scale training, and the TPU 8i optimized for inference and AI agents. This enables precise matching of the different demands of AI scenarios.
Chey reiterated that the memory supply bottlenecks triggered by the widespread adoption of AI are expected to persist until 2030. He pointed out that not only is global capital pouring into AI data center construction, but Nvidia's upcoming AI PC architecture will also create rigid demand for high-capacity memory, providing long-term growth momentum for the memory market.
After roughly 18 months of development, Crescent Island is expected to begin customer sampling later this year. Intel is also exploring manufacturing options that could leverage its own fabs, potentially enhancing cost competitiveness through greater vertical integration.
Samsung Electronics announced that it has begun shipping the industry’s first 12-layer HBM4E samples to major global customers, further strengthening its leadership in the next-generation HBM market.
Alper Ilkbahar, CTO and EVP of Sandisk, said in a recent media interview that the global AI race is becoming increasingly memory-centric rather than compute-centric, which could exacerbate an unprecedented tight supply of memory chips. He also provided an update on HBF, revealing that the wafer is currently being designed, samples of the HBF chip will be available by the end of this year, and the complete product with a controller is expected to launch next year.