Nvidia just reported record Q2 revenue of $46.7 billion, up 56% year-over-year, alongside $26.4 billion in net income. Growth was spearheaded by data center demand, with its Blackwell chips alone driving $27 billion in sales. However, regulatory uncertainty left the company with zero sales of its China-focused H20 chip to Chinese customers.
Nvidia just made its Jetson Thor robotics computing platform generally available. It delivers 7.5x the AI compute and double the memory of the previous Orin chip, letting physical AI agents run generative reasoning models directly on edge devices. Boston Dynamics and Agility Robotics are already putting it inside Atlas and Digit, with developer kits starting at $3,499.
NVIDIA just dropped the first major update to Project G-Assist, its experimental on-device assistant for PC gamers. A new lightweight model cuts VRAM usage by 40%, bringing local voice and text controls to any RTX card with at least 6GB VRAM. It also integrates a mod.io plug-in hub you can browse via chat.
Nvidia just dropped Nemotron-Nano-9B-v2, a hybrid Mamba-Transformer model slimmed down to fit on a single A10 GPU. Developers can toggle reasoning traces on or off using simple /think tokens and dial in exact thinking budgets. It delivers up to 6x faster inference than similar transformers while outperforming Qwen3-8B across key benchmarks.
NVIDIA just released Granary, an open-source multilingual speech dataset spanning around a million hours of audio across 25 European languages. Alongside it, the company launched Canary-1b-v2 for high-accuracy translation and Parakeet-tdt-0.6b-v3 for fast, real-time transcription on Hugging Face.
NVIDIA just released the FLUX.1 Kontext [dev] NIM microservice, making advanced local image editing much easier on RTX PCs. Quantization slashes the 24GB model down to 7GB for RTX 50 series, while TensorRT delivers over 2x speedups with direct ComfyUI support.
U.S. authorities are secretly planting location tracking devices in select shipments of advanced AI chips and servers suspected of being illegally diverted to China, according to Reuters. Sources in the server supply chain reported finding hidden devices inside packaging and even within Dell and Super Micro systems powered by Nvidia and AMD hardware. Why it matters: It reveals the extraordinary lengths Washington is taking to police semiconductor export restrictions.
Stability AI and NVIDIA just rolled out a Stable Diffusion 3.5 NIM microservice to streamline enterprise image generation. Running TensorRT optimizations on H100 GPUs, it delivers a 1.8x speedup over standard PyTorch while packaging SD3.5 Large alongside Depth and Canny ControlNets in a single container.
Nvidia and AMD just agreed to pay the U.S. government 15% of their revenue from AI chip sales to China. President Donald Trump said he originally demanded 20% before Jensen Huang negotiated a 5% discount. The deal walks back recent bans on chips like the H20 and opens the door for a trimmed-down Blackwell model to enter China.
Nvidia just launched a major robotics push headlined by Cosmos Reason, a 7-billion-parameter reasoning model built to serve as the physical brain for autonomous machines. Alongside new Omniverse neural rendering libraries and Blackwell servers, the system lets developers generate hyper-realistic synthetic training data and run complex simulations. Why it matters: By closing the sim-to-real gap in software, building next-generation robots no longer requires breaking expensive hardware in the physical world.
Donald Trump just revealed he initially planned to break up Nvidia after learning about its near-monopoly, but backed off after an advisor warned it would take rivals a decade to catch up. Jensen Huang later won him over in person, convincing the administration to let Nvidia sell H20 chips to China.
OpenAI just struck a deal with Oracle to add 4.5 gigawatts of data center capacity in the U.S. for its Stargate infrastructure project. Combined with its Texas site, Stargate now has over 5 GW under development to house more than 2 million chips, with initial Nvidia GB200 racks already spinning up early frontier workloads.
The UK just fired up Isambard-AI, a £225M national supercomputer packing 5,400 Nvidia superchips in Bristol. Ranking 11th globally, the nuclear-powered beast will give public researchers the raw horsepower to tackle everything from skin cancer detection on darker skin tones to dairy farming and crowd safety. Why it matters: It gives Britain a shot at AI sovereignty without relying entirely on private US tech giants.
Former OpenAI CTO Mira Murati just announced a massive $2B funding round led by a16z for her startup, Thinking Machines Lab, driving its valuation to $12B. The team is building collaborative multimodal AI and plans to ship its first product in the coming months, complete with a major open-source component for researchers.
Nvidia just applied to resume sales of its H20 AI chips to China after securing U.S. government assurances, reversing an export ban from April. Commerce Secretary Howard Lutnick confirmed the move is tied to a deal on Chinese rare earth shipments. Chinese tech giants are already scrambling for stock, though U.S. lawmakers are pushing back hard.
Meta is pouring hundreds of billions into massive multi-gigawatt data centers to power its new Superintelligence Labs. The first cluster, Prometheus, arrives next year, followed by Hyperion—a 5-gigawatt behemoth that triples OpenAI's planned footprint. Zuck's goal is straightforward: give his newly poached researchers the most compute per head on the planet.
Intel's new CEO Lip-Bu Tan reportedly conceded to employees that the company has fallen out of the top 10 semiconductor makers and that it is "too late" to catch Nvidia in AI model training. Facing massive losses and layoffs, Intel plans to abandon the training race and redirect its focus to edge AI on PCs and autonomous agents.
NVIDIA just unveiled Helix Parallelism, a technique co-designed with its Blackwell architecture to fix memory bandwidth bottlenecks in multi-million-token decoding. By decoupling how GPUs handle attention and feed-forward networks, it boosts concurrent users by up to 32x at the same latency budget. It basically makes encyclopedia-sized prompts practical for real-time serving.
NVIDIA just rolled out TensorRT optimizations for Black Forest Labs' FLUX.1 Kontext [dev] model, bringing over 2x faster image generation and editing to RTX GPUs. Quantization cuts model size down to 12GB for FP8 and 7GB for FP4, allowing creators to run step-by-step visual edits locally without complex workflows.
OpenAI just confirmed it is testing Google's TPUs, but shut down rumors of any large-scale rollout. While the company continues to rely heavily on Nvidia GPUs alongside AMD and CoreWeave, it is also pushing ahead with an in-house chip expected to tape out later this year.
