According to CNA, NEO Semiconductor — an AI memory company founded by Silicon Valley entrepreneur Fu-Chang Hsu (許富菖) — announced a new generation of AI chip embedded-memory technology called X-SRAM on August 6, 2026, at Future Memory and Storage (FMS 2026), the annual global memory and storage industry conference. The company simultaneously launched a companion platform, NEO.AI. TechNews reported the same announcement, describing the pairing of X-SRAM and NEO.AI as intended to address insufficient embedded memory capacity in AI chips.
CNA reported that traditional SRAM (Static Random-Access Memory) is approaching its scaling limits at advanced process nodes, which constrains growth in on-chip cache capacity. This scaling ceiling is the specific problem NEO Semiconductor's announcement targets: the company frames X-SRAM and NEO.AI as new technology aimed at solving AI chips' embedded-memory capacity shortfall, a challenge both CNA and TechNews cited in their coverage of the FMS 2026 launch.
Both CNA and TechNews reported the same set of figures from NEO Semiconductor. The company said X-SRAM uses a new memory architecture that provides up to five times the density of traditional SRAM. It said this could raise AI chip embedded-memory capacity from a current range of roughly 200MB to 400MB up to a range of 1GB to 2GB. NEO Semiconductor also projected that X-SRAM could cut High Bandwidth Memory (HBM) access power consumption by an estimated 50% to 80%.
| Metric | Current (per NEO Semiconductor) | With X-SRAM (per NEO Semiconductor) |
|---|---|---|
| Density vs. traditional SRAM | 1x (baseline) | Up to 5x |
| Embedded memory capacity | ~200MB–400MB | 1GB–2GB (projected) |
| HBM access power consumption | Baseline | Estimated 50%–80% reduction |
These figures are NEO Semiconductor's own stated projections, as reported by CNA and TechNews; both outlets attribute the numbers directly to the company's own statements at the announcement.
Acer founder Stan Shih (施振榮) commented on the announcement through a press release issued by his Stan Shih Foundation (智榮基金會). Per CNA, Shih said AI is rapidly transforming global industries and that memory will be a key pillar supporting the next stage of AI development — a statement TechNews reported in near-identical wording. CNA further quoted Shih saying that NEO Semiconductor's continued rollout of innovations such as X-SRAM and NEO.AI offers a new direction for breaking through AI memory bottlenecks, and that he hopes these innovations bring more possibilities for the semiconductor industry in Taiwan and globally.
The two threads in this coverage line up: CNA's reporting on SRAM's approach to its scaling limits at advanced nodes explains why an embedded-memory architecture claiming up to five times the density — and a jump from roughly 200MB–400MB to 1GB–2GB of capacity — would matter to AI chip design. Stan Shih's comments, issued through his foundation rather than through NEO Semiconductor itself, add a Taiwan semiconductor-industry voice to a launch whose density, capacity, and power figures currently rest on the company's own statements as relayed by CNA and TechNews.
X-SRAM is a new AI chip embedded-memory technology from NEO Semiconductor, announced at FMS 2026. According to CNA and TechNews, the company says it uses a new memory architecture that can provide up to five times the density of traditional SRAM.
NEO Semiconductor was founded by Silicon Valley entrepreneur Fu-Chang Hsu (許富菖), according to CNA's report on the company's FMS 2026 announcement.
Per CNA, Acer founder Stan Shih said, through the Stan Shih Foundation, that he hopes NEO Semiconductor's innovations such as X-SRAM and NEO.AI bring more possibilities for the semiconductor industry in Taiwan and globally.
According to Intel's newsroom announcement, the company is investing €5 billion ($5.7 billion) to expand its Leixlip campus in Ireland, producing Intel Xeon 6 and next-generation Xeon processors on the Intel 3 node. Intel says the capital programme began earlier this year and builds on more than €30 billion invested in Ireland since 1989.
According to TechCrunch and VentureBeat, Meta released Muse Code — a beta terminal-based AI coding agent for large repositories — alongside the Muse Spark 1.2 model on August 5, 2026. Muse Code installs with a single command, runs persistent background agents, and is entirely proprietary, putting Meta in direct competition with Anthropic's Claude Code and OpenAI's Codex.
NVIDIA has moved Alpamayo 2 Super into commercial use, according to NVIDIA's official blog. The autonomous-driving reasoning model runs on Cosmos 3 Super Reasoner, holds three times the parameters of its 10-billion-parameter predecessors, and ships under the OpenMDW-1.1 license, which ITHome reports NVIDIA has extended across the entire Alpamayo model family for fine-tuning and commercial redistribution.