GoKawiil
Tech News
Nvidia posted quarterly sales of $96.2 billion, topping analyst estimates by roughly 4%, marking a new high for the chipmaker. Despite the strong results, shares dipped slightly as broader market anxiety over the sustainability of AI-related spending overshadowed the earnings beat.
Nvidia reported record quarterly revenue of $96.2 billion, up more than $10 billion from the prior quarter, driven by data center revenue that more than doubled year-over-year to $89 billion. Profits also more than doubled to $59.7 billion, and the company projects revenue will climb to $108 billion in the coming quarter.
Perplexity and Nvidia have released Portable Computer, a locally-run version of Perplexity's Computer AI platform that performs agentic workloads directly on a user's PC or workstation instead of relying on cloud servers. The app mirrors the original Computer interface, supports Qwen 3.8 and PPLX local models (both scaling to 27 billion parameters), and will soon add Nvidia's Nemotron 3.5 Lightning model, while asking users for permission before sending any data to cloud-based frontier models for harder tasks.
Researchers demonstrated a new Rowhammer variant called GPUThor that defeats NVIDIA's ECC memory protections on Ampere-class workstation GPUs, including the RTX A4000, A4500, A5000, and A6000. By using a non-uniform hammering pattern tuned to avoid triggering GDDR6's Target Row Refresh defenses, the attack generates far more bit flips than earlier concepts like GPUHammer, enabling denial-of-service conditions and root-level privilege escalation.
OpenAI revealed its first custom AI chip, called Jalapeño, claiming it delivers industry-leading speed and efficiency for inference workloads. The chip, co-developed with Broadcom, is set to be deployed in OpenAI's own infrastructure by year-end, with two more generations already in development.
Canada confirmed a new round of tariffs matching Washington's 50% duties, covering more than 700 American products worth roughly $20 billion, set to take effect September 8. The move follows the collapse of trade talks last week, with both governments blaming each other for last-minute demands. Ottawa specifically doubled its tariff on U.S. steel and aluminum to 50% in direct response to similar U.S. levies.
Perplexity has released Portable Computer, an agentic AI tool that runs entirely on local hardware instead of the cloud, built in partnership with NVIDIA. It relies on compact models like Qwen 3.8 and NVIDIA Nemotron 3.5 Lightning to handle complex tasks on-device, and only reaches out to cloud-based models when needed, with user permission. The feature is currently available to Perplexity Pro and Max subscribers running NVIDIA's DGX Spark hardware, with RTX GPU support planned.
Nvidia has updated Ray Reconstruction, its AI denoiser-upscaler for ray tracing, to the second-generation transformer architecture used in DLSS 4.5 Super Resolution. The company says the new model handles 35% more compute and 20% more parameters than before, improving image quality and reducing ray-tracing artifacts with little to no performance cost. GeForce RTX owners can access the update immediately.
Chris Malone, who led OpenAI's data center strategy after joining from Meta and Google in March 2024, has left the company. OpenAI confirmed it recently restructured its infrastructure organization, shifting oversight so that Malone no longer reported to President Greg Brockman but instead to VP Sachin Katti before his exit. The company says several other executives, including Uday Ruddarraju, Brent Mayo and Spas Lazarov, remain in charge of its data center efforts.
Elon Musk revealed on SpaceX's latest earnings call that the company's Starmind AI project will run on NVIDIA's Vera Rubin GPU architecture, both on the ground and eventually in orbit. Musk argued the NVL72 VR system design outperforms traditional rack-based computing setups and would be cheaper and more effective for SpaceX's needs.
Researchers identified a security bug in NVIDIA's OpenClaw tool that allows attackers to reach the local model server through the Ollama API without authentication. This unauthenticated access can be leveraged to tamper with the AI agent's underlying model, leading to persistent corruption of its outputs.
Perplexity and Nvidia have released an AI agent designed to run entirely on local hardware rather than through cloud-based inference, eliminating per-token usage charges. The system relies on on-device processing, letting users run agentic AI workflows without sending data to remote servers or paying ongoing API costs.