Tech News
← Home  ·  All topics

Nvidia

452 GoKawiil briefs on this topic

Explainer: Ray tracing's performance cost still limits adoption on mid-range GPUs

Ray tracing is a rendering technique that simulates realistic lighting, shadows and reflections by tracing light rays, in contrast to older rasterization and screen-space reflection methods used in non-RT titles. According to Steam Hardware Survey data cited in the piece, NVIDIA's RTX 5070 is currently the most-used graphics card, indicating many PC gamers now own hardware capable of handling ray tracing. The author, who uses an RTX 5090 with DLSS 4.5 Multi Frame Generation, says gamers on older cards like the RTX 3060 or Radeon RX 6600 XT should disable ray tracing and prioritize stable frame rates instead.

Broker Ting-Wei Sun pleads guilty in Super Micro chip-smuggling case

Ting-Wei 'Willy' Sun, a third-party broker charged alongside Super Micro co-founder Yih-Shyan Liaw and sales manager Ruei-Tsang Chang, has pleaded guilty to conspiring to illegally reroute Nvidia AI chips to China. Sun's plea is tied to an agreement with prosecutors whose cooperation terms haven't been disclosed, and he's scheduled for sentencing on September 8, 2027. Investigators allege the group disguised real Super Micro servers as dummy units to deceive export inspectors before routing them through Thailand to Alibaba.

SpaceX Spectrum Deal and Nvidia Chip Plans Add to Volatile AI-Driven Market Week

Major indexes hit fresh records before a sharp tech selloff on Thursday, with the S&P 500 still finishing the week up 1.2% and the Nasdaq up 0.6% after Friday's rebound. Separately, SpaceX reportedly seeks to raise $40 billion to expand its Nvidia-powered AI compute business and announced a deal to acquire a nationwide wireless spectrum portfolio for Starlink.

Nvidia reportedly ends GeForce RTX 5090 production, redirects GB202 chips to AI cards

According to leaked factory notices cited by hardware leakers @hongxing2020 and @Zed__Wang, Nvidia has stopped producing GB202 chips for consumer RTX 5090 graphics cards, including Chinese 'D' variants. The notice claims the RTX 5080 24GB will become the new top consumer GeForce model, with the RTX 5080 16GB being discontinued. Nvidia has not confirmed the claim and typically does not comment on rumors.

Nvidia's GPU customer base widens as neocloud providers multiply past 300

Nvidia's stock hit a record this week, pushing its market cap near $6 trillion, as the company projects $108 billion in October-quarter revenue, up 89% year-over-year. SemiAnalysis counted 323 Nvidia GPU providers as of September, up from 209 eleven months earlier, while Nvidia's own filings show five clients each accounted for at least 10% of accounts receivable in the July quarter, up from three in January. Companies now access Nvidia chips through hyperscalers like Amazon, Microsoft and Google, specialized 'neoclouds' such as CoreWeave, online marketplaces, or direct hardware purchases.

Modder fits 360mm AIO radiator inside Metabox gaming laptop, cuts temps 60°C

YouTuber TrashBench rebuilt the cooling system of a Metabox P950EP laptop (Nvidia GTX 1060, Intel Core i7-8750H) by installing custom waterblocks, a new pump, and a cut-down 360mm AIO radiator mounted behind the screen. The mod lowered CPU and GPU temperatures by 60°C during benchmarks and stress tests, and allowed an average 14% overclock gain. The builder used 3D-printed mounts, epoxy, and thermal pads to fit the radiator and attach heatsinks to the RAM and VRMs, making the modification permanent.

Mistral Large 4 ranks top outside US/China, lags Chinese open models overall

Mistral released Large 4, a trillion-parameter mixture-of-experts model, in a research preview this week, with full weights due at the end of October. Independent benchmarking firm Artificial Analysis scored it 38 on its Intelligence Index, making it the top model from outside the US and China, though several Chinese open models including DeepSeek V4.1 Flash and MiMo-V2.6-Pro scored as high or higher overall. Large 4 scored 50 on Artificial Analysis's Cyber Index, tying GLM-5.3-Flash, and hit 82% on the CyberGym-E2E-AA cybersecurity test, its best individual result.

Nvidia's Jensen Huang awarded National Medal of Science by Trump

President Trump presented Nvidia CEO Jensen Huang with the National Medal of Science for his work turning GPUs into general-purpose computing platforms used across AI and scientific research. The ceremony, held at the US Institute of Peace in Washington, D.C., also honored AMD's Lisa Su, Google's Sergey Brin and Elon Musk with the same medal, while Michael Dell and Satya Nadella received the National Medal of Technology and Innovation. At the event, eleven tech companies pledged $2.4 billion combined in AI tools and computing credits to the government's Genesis Mission, with Nvidia committing $1 billion over five years.

Firmus Grid cancels planned IPO targeting $30 billion valuation

Australian cloud-computing company Firmus Grid, backed by Nvidia, withdrew plans for an initial public offering that had aimed for a $30 billion valuation. The company currently operates only two data centers.

Nvidia-backed Firmus pulls planned $5 billion Australian IPO

Firmus, the Nvidia-backed Australian AI data center operator, has withdrawn its planned initial public offering, which would have raised about $5 billion and ranked as Australia's second-largest share sale ever. The company said market volatility and unfavorable offer terms led its board to scrap the listing, and it will instead seek funding through private markets while weighing other public or private options.

Nvidia-backed Firmus pulls planned $30bn Australian stock market listing

Firmus, an AI data centre operator backed by Nvidia, Blackstone and Jane Street, has abandoned plans for a stock market debut that would have valued it above $30bn, citing market volatility and investor interest. The company said it will instead seek funding through private markets while weighing other public and private options. The withdrawal follows UniSuper and other institutional investors declining to join the IPO over valuation concerns.

OpenAI's annualized revenue was $50 billion, not $68 billion, AI stocks fall

OpenAI told investors its annualized revenue stood at roughly $50 billion at the end of September, below the $68 billion figure widely reported last month, CNBC confirmed. A person familiar with the matter said the higher figure had included gross revenue from OpenAI's partners, used for comparison with rival Anthropic. Shares of Nvidia, Oracle and CoreWeave fell after the revised figure emerged, alongside reported growth rates of 77% overall and 107% in enterprise business.