AMD is reportedly planning another consumer GPU price increase of 10-15% in the second half of 2026, driven by rising memory costs as AI datacenter demand reallocates memory capacity away from consume…
A developer used Gemini Flash Lite to transcribe a video explaining High Bandwidth Memory (HBM) technology. The video details how HBM achieves high bandwidth through an extremely wide bus width, contr…
Micron Technology is set to report fiscal Q3 2026 earnings on June 24, projecting record revenue of $33.5B and a non-GAAP gross margin of 81%, more than double from a year ago, driven by surging deman…
Worldwide revenue for Data Center IT Semiconductors and Components surged 116% year-over-year in Q1 2026, driven by rising memory prices, NVIDIA's Blackwell platform ramp, and hyperscaler custom accel…
The "Memory Wall" bottleneck, originally coined in the 1990s to describe the speed gap between processors and DRAM, has re-emerged in the AI era as DRAM and high-bandwidth memory (HBM) struggle to mee…
A team running a 70B Llama 3 fine-tune at 200 requests per second cut median time-to-first-token from 380 ms to 140 ms on the same hardware by implementing speculative decoding. The technique addresse…
Micron Technology is emerging as a frontrunner in the AI-driven memory supercycle, with surging demand for its DRAM and HBM products driving record revenue and margin expansion. Strategic customer agr…
Modern GPUs spend most of their time during AI inference waiting for data, as memory bandwidth cannot keep pace with compute throughput. This fundamental bottleneck has driven the AI hardware market, …
SK hynix on Tuesday unveiled iHBM, a new high bandwidth memory technology that embeds a proprietary cooling element directly inside the HBM package to reduce heat generation in AI computing environmen…
Memory's share of AI chip component costs rose from 52% to 63% between Q1 2024 and Q4 2025, according to an analysis of Nvidia, AMD, Google, and Amazon chips. Total component spending on AI chips grew…
According to a leaked image of Intel's upcoming "Crescent Island" AI GPU, the chip features a massive Xe3P core and bypasses the ongoing global HBM shortage by utilizing 160GB of LPDDR5X memory instea…
Cerebras Systems has secured a 750MW compute deal with OpenAI, positioning the company for its upcoming IPO as demand for fast token generation surges. The wafer-scale chip maker's speed advantages, p…