GoKawiil - Tech News

31.

China drafts $295 billion plan to build national AI data center grid running on 80% homemade silicon — projected 2028 timeline could run into limits of local chip production (tomshardware.com)

2026-06-10 | by Luke James | get AI Data Center Kit → | tags: accelerators, china, chips

32.

AMD's Helios MI455X AI platform breaks cover, initial systems use UALink-over-Ethernet interconnects — AMD's Vera Rubin rival surfaces, but the downsides of Ethernet could hamstring performance (tomshardware.com)

2026-06-04 | by Anton Shilov | get AMD Ryzen 9 7950X → | tags: accelerators, ethernet, helios

33.

Frore shows off LiquidJet Nexus coldplate for Nvidia Vera Rubin, other AI accelerators — offers up claimed 10% token generation boost over rival liquid-cooling solutions (tomshardware.com)

2026-06-04 | by Anton Shilov | get LiquidJet Nexus Coldplate → | tags: cooling, credit, frore

34.

Astera Labs showcases 320-lane PCIe 6.0 switch for vendor-agnostic scaling in data centers — up to 80 accelerators can be scaled up using PCIe alone (tomshardware.com)

2026-06-03 | by Anton Shilov | get PCIe 6.0 Switch → | tags: accelerators, astera, astera labs

35.

SK hynix unveils 'iHBM' thermal architecture that cools AI memory at the source — integrated cooling elements inside HBM interface cut thermal resistance by 30%, target next-gen HBM5 accelerators and dense AI data centers (tomshardware.com)

2026-05-26 | by Etiido Uko | get SK hynix iHBM Cooling Kit → | tags: sk hynix, ihbm, hbm5

36.

Jensen says Nvidia now has 'zero percent' market share in China — says US export policy 'has already largely backfired' (tomshardware.com)

2026-05-03 | by Anton Shilov | get Nvidia GeForce RTX 4090 → | tags: nvidia, jensen huang, china

37.

Samsung and SK hynix warn AI-driven memory shortages could last until 2027 and beyond, as HBM demand explodes — customers already reserving supply years ahead, while the wider DRAM market begins to tighten (tomshardware.com)

2026-04-30 | by Etiido Uko | get Samsung 32GB DDR5 RAM → | tags: samsung, sk hynix, hbm

38.

Inside Google's TPU V8 strategy, delivering two chips for two crucial tasks at incredible scale — network scales up to 1 million TPUs per cluster, an advantage over Nvidia AI accelerators (tomshardware.com)

2026-04-27 | by Luke James | get Google TPU V8 Developer Kit → | tags: google tensor processing units, tpu 8t, tpu 8i

39.

TSMC's details next-gen CoWoS roadmap: over 14-reticle packages and 48x leap in compute power expected by 2029 — massive size enables 24 HBM5E stacks and additional memory bandwidth jump (tomshardware.com)

2026-04-27 | by Anton Shilov | get High-Performance Computing Memory Kit → | tags: tsmc, cowos, hbm5e

40.

SpaceX says it is going to begin manufacturing GPUs — $1.75 trillion IPO listing reportedly includes in-house GPU production (tomshardware.com)

2026-04-23 | by Anton Shilov | get NVIDIA GeForce RTX 4090 → | tags: spacex, gpus, tesla

41.

Meta commits to 1 gigawatt of custom chips with Broadcom as Hock Tan decides to leave board (cnbc.com)

2026-04-14 | by Katie Tarasov Jordan Novet | get Broadcom Chip Development Kit → | tags: meta, broadcom, ai accelerators

42.

Meta commits to 1 gigawatt of custom chips with Broadcom as Hock Tan agrees to leave board (cnbc.com)

2026-04-14 | by Katie Tarasov Jordan Novet | get Broadcom Custom Chip Kit → | tags: meta, broadcom, ai accelerators

43.

Meta commits to one gigawatt of custom chips with Broadcom as Hock Tan agrees to leave board (cnbc.com)

2026-04-14 | by Katie Tarasov Jordan Novet | get Broadcom Chip Development Kit → | tags: meta, broadcom, ai accelerators

44.

Intel and SambaNova team up on heterogenous AI inference platform — different hardware performs different workloads (tomshardware.com)

2026-04-08 | by Anton Shilov | get Intel Xeon Processor → | tags: intel, sambanova, xeon 6

45.

Memory will consume 30% of hyperscaler AI data center spending this year, a 4X increase over 2023 — Nvidia gets preferential supply terms well below standard market rates, says analyst firm (tomshardware.com)

2026-04-03 | by Luke James | get Nvidia GeForce RTX 4090 → | tags: nvidia, hyperscaler, dram

46.

Why Nvidia just poured $2 billion into AI ASIC competitor Marvell — NVLink Fusion turns into soft ecosystem lock-in (tomshardware.com)

2026-04-02 | by Luke James | get Nvidia NVLink Bridge → | tags: nvidia, marvell, nvlink

47.

Ollama adopts MLX for faster AI performance on Apple silicon Macs (9to5mac.com)

2026-03-31 | by Marcus Mendes | get MLX AI Developer Kit → | tags: apple silicon, mlx, hugging face

48.

US gov't revokes controversial AI hardware export rule that would mandate investments from foreign companies — new export rules are still in the works, though (tomshardware.com)

2026-03-14 | by Anton Shilov | get AI Hardware Export Kit → | tags: ai accelerators, u.s. commerce department, nvidia gb300

49.

FLASH radiotherapy's bold approach to cancer treatment (news.ycombinator.com)

2026-03-07 | by Tom Clynes | get Radiation Therapy → | tags: cancer treatment, linear particle accelerators, cern

50.

UALink roadmap plots course to optimized AI data center interconnects — examining the open standard designed to combat vendor lock-in while offering cost and performance optimization (tomshardware.com)

2026-02-20 | by Anton Shilov | get Optimized AI Data Center Interconnects → | tags: accelerators, ethernet, hardware

51.

HetCCL makes clustered Nvidia and AMD AI accelerators play nice with each other via RDMA — vendor-agnostic collective communications library removes an obstacle to heterogeneous AI data centers (tomshardware.com)

2026-02-04 | by Bruno Ferreira | get Nvidia Accelerator → | tags: gpus, hetccl, networking

52.

Nvidia reportedly boosts Vera Rubin performance to ward hyperscalers off AMD Instinct AI accelerators — increased boost clocks and memory bandwidth pushes power demand by 500 watts to 2300 watts (tomshardware.com)

2026-01-21 | by Anton Shilov | get GPU → | tags: higher, memory, nvidia

53.

Furiosa: 3.5x efficiency over H100s (news.ycombinator.com)

2026-01-15 | get GPU → | tags: accelerators, furiosa, nxt rngd

54.

SK hynix shows 16-Hi HBM4 memory for AI accelerators — 48 GB at 10 GT/s over a 2,048 interface (tomshardware.com)

2026-01-14 | by Anton Shilov | get NAND → | tags: backside, memory, power

55.

AMD touts Instinct MI430X, MI440X, and MI455X AI accelerators and Helios rack-scale AI architecture at CES — full MI400-series family fulfills a broad range of infrastructure and customer requirements (tomshardware.com)

2026-01-06 | by Anton Shilov | get AI accelerator → | tags: accelerators, helios, instinct

56.

Here’s why visiting museums between Christmas and New Year’s makes me a better leader (feeds.feedburner.com)

2025-12-25 | get Art Museum → | tags: accelerators, accelerators techstars, awards

57.

Huawei's AI chip capabilities still pale in comparison to American silicon — report from U.S. council details that despite fears, Nvidia continues to lead by a wide margin (tomshardware.com)

2025-12-18 | by Luke James | get GPU → | tags: accelerators, advanced, ascend

58.

The RAM pricing crisis has only just started, Team Group GM warns — says problem will get worse in 2026 as DRAM and NAND prices double in one month (tomshardware.com)

2025-12-01 | by Anton Shilov | get Memory → | tags: accelerators, commodity, dram

59.

China's hybrid-bonded AI accelerators could rival Nvidia's Blackwell GPUs — top semiconductor expert hints at 'fully controllable domestic solution' (tomshardware.com)

2025-11-27 | by Anton Shilov | get AI Accelerator → | tags: china, hardware, hypothetical

60.

AMD's Linux kernel patches suggest enablement of next-gen Instinct MI400-series AI GPU accelerators (tomshardware.com)

2025-11-21 | by Anton Shilov | get GPU → | tags: generation, gpus, instinct