GoKawiil
Tech News
Biren Technology reported first-half 2026 revenue of $183.9 million, up nearly 2,000% year-over-year, as Chinese buyers shifted to domestic AI accelerators after Nvidia halted GPU shipments to China under US export controls. Despite the revenue jump and a 42.7% gross margin, Biren posted a $56.2 million loss due to heavy investment in new chips, software, and rack-scale systems.
Ajinomoto build-up film, the insulating material used in nearly every high-end CPU, GPU, and AI accelerator package, is supplied almost entirely by one Japanese company that controls over 95% of the global market. As AI data centers scale into the gigawatt era with millions of accelerators needed, this narrow supply chain of substrate makers and a single film source is becoming a critical constraint on chip production.
Nvidia has started using Palantir's software platform to run its supply chain operations, making itself the first user of a system the two companies plan to jointly market to other businesses and government agencies. The partnership positions Nvidia as a live demonstration of Palantir's capabilities amid the broader AI infrastructure buildout.
Nvidia has launched GeForce Graphics Driver 616.92, adding Game Ready optimizations for the path tracing update in 007 First Light alongside support for Active Matter, Aniimo, WARDOGS and a DLSS Super Resolution update for PUGS. The release also patches several bugs, including browser flicker, virtual display creation failures, and black screens during Remote Desktop sessions introduced by the prior 616.56 driver.
A modder known as Perseval_BLR, aided by Claude AI, built an app called NeuralScreen that overlays Nvidia's leaked DLSS 5 neural rendering technology onto any windowed application, not just games. Demonstrations show it applied to YouTube gameplay footage, a scene from Interstellar, and Doom 3 running in a window, with customizable presets and key-bound controls for switching, recording, and comparisons.
Brazilian YouTuber Fabian (fmklab) built a working version of the RTX 3070 Ti 16GB that Nvidia never shipped, by fitting an RTX 3070 Ti GPU onto an RTX 3070 PCB and swapping in higher-capacity Samsung GDDR6 chips. After reconfiguring the board's memory straps using Nvidia's BIOS Reader tool, the card correctly recognized the full 16GB pool in GPU-Z, matching the specs of a prototype that had only ever surfaced in database listings and photos.
Jensen Huang has for the second time this year proclaimed that artificial general intelligence has been achieved, this time tying his claim to OpenAI's newly launched GPT-6 Astra model. He highlighted that Astra was trained on more than 100,000 Nvidia Blackwell chips, with 400,000 more coming online soon, even as independent benchmarks show Astra trailing rival Claude in some tasks.
Marvell Technology CEO Matt Murphy told CNBC that the company's ability to earn trust with major cloud computing customers over ten years has driven its shares up about 241% in the past year, far outpacing rival Broadcom's 6.6% gain. He pointed to new AI partnerships with Nvidia and Google, along with custom chip deals with all four major U.S. hyperscalers, as evidence of that trust paying off.
Inception has released Mercury 2.5, an update to its diffusion-based language model that the company says is 40% more intelligent than Mercury 2 while keeping the same low-latency, low-cost performance. The model runs at over 1,100 tokens per second on standard NVIDIA GPUs, supports a 260K-token context window, and launches at an 80% discount, priced at $0.04 per million input tokens and $0.15 per million output tokens.
NVIDIA's new DLSS 5 neural rendering technology launches with NBA 2K27, adding finer skin texture and improved shadow detail to player and crowd models. Testing on an RTX 5090 at 4K showed the feature nearly halved frame rates, dropping from 240fps to 150fps, while offering little visible improvement in wide-angle gameplay shots.
The Department of Energy has unveiled the Genesis Mission, a national initiative combining AI systems with top-tier supercomputers to tackle 26 designated challenges of national importance. The program brings together major technology firms including AMD, Nvidia, and Microsoft alongside smaller startups pursuing novel computing approaches, such as dataflow-based AI processors.
A hardware comparison finds that Nintendo's Switch 2 and Valve's Steam Deck are closely matched in handheld mode, with the Switch 2's Nvidia Ampere GPU offering a slight graphical edge over the Deck's AMD RDNA 2 chip. The gap widens dramatically when the Switch 2 is docked, boosting its GPU output from about 1.71 to 3.7 TFLOPS, enabling 4K/60fps in titles like Metroid Prime 4, while the Deck sees no docked performance boost. Meanwhile, the Steam Deck's Zen 2 CPU can hit higher clock speeds and it also carries more unified memory, 16GB versus the Switch 2's 12GB.