What changed
2026-08-06 07:25 UTC → 2026-08-06 07:26 UTC ·
added
removed
In mid‑July Since July 2026 the rollout of new fabs and advanced‑node production has accelerated. Tata Electronics partnered with Taiwan’s Powerchip to start India’s first large‑scale and Powerchip’s Gujarat wafer fab in Gujarat, fab, now delayed pushed to mid‑2028 mid‑2028, will start with an initial a 90 nm node. Within a week, Samsung announced a second Texas fab slated for targeting completion by in 2026 and a 1.4 nm product line for 2030; its Texas 2 nm volume production target is slated for 2030, while end‑2026 with a second Texas plant planned. TSMC accelerated is fast‑tracking its own 1.4 nm line in Taiwan for mid‑2028 volume and expects its 3 nm wafer starts to reach 180 000 per month by Q4 2024, with 2 nm output of 80‑100 k wafers per month by mid‑2028 and year‑end. Huawei outlined a 1.4 nm‑equivalent roadmap for 2031. By August 2026 Samsung said its wafer‑fab capacity would be reported 100 % utilized in the second half utilization of the year, a level that could bring its non‑memory division to break‑even fab capacity in H2 2026, driven by Q3‑Q4. AI and HPC demand for HBM4, and confirmed progress toward full‑scale 2‑nm production. At the Future of Memory and Storage conference the company Samsung unveiled V10 BV‑NAND, LPDDR5X‑PIM, the zHBM architecture and a new architecture, V10 BV‑NAND storage solution. and LPDDR5X‑PIM AI‑memory. SK Hynix, together Hynix with Marvell, Marvell introduced the CMM‑Ax CXL‑based memory module that integrates high‑performance DRAM with Marvell’s Structera A engine, delivering up to 5.5 × higher throughput than a single GPU for GPU‑level AI data‑center workloads. throughput. LG Electronics announced presented a multi‑tier three‑tier AIoT chip platform (1.5 TOPS, 10 TOPS, 30 TOPS+), and (1.5‑30 TOPS). HanYu Semiconductor revealed showcased an AI‑driven ultrasonic full‑die inspection system and a second‑generation placement machine for detecting defects in multilayer ceramic capacitors used in automotive and AI server applications. capacitors. Samsung also reported that full‑capacity utilization is driven by demand Display revealed Center Etched Thin Glass technology for HBM4, AI and high‑performance computing, confirming volume production of its Texas 2 nm fab by the end of 2026 and the start of a second Texas plant. TSMC projected a rapid ramp‑up of its 3 nm line (180 000 wafer starts per month in early Q4 2024) and 2 nm output (80‑100 000 wafers per month by year‑end), pushing combined foldable OLEDs. Combined 2‑nm/3‑nm capacity above across TSMC and Samsung now exceeds 260 000 wafers monthly. per month.