Company
NVIDIA
NVIDIA is a U.S. accelerated computing company. After establishing itself in graphics processors, it now develops GPUs, systems, networking and software for artificial intelligence, data centers, scientific computing, gaming and robotics.
News
-
Eight NVIDIA DGX Sparks, 1TB of memory: giant AI models reach the desktop, but how fast?
A 753-billion-parameter AI model can now run across eight compact NVIDIA DGX Spark computers. On October 22, 2026, NVIDIA published benchmarks covering large-model inference, multi-model orchestration and distributed training on its desktop-scale platform. With approximately one terabyte of combined unified memory, the cluster can accommodate models previously impractical for small local installations. But the results also expose the limitations of distributed computing: eight machines do not automatically deliver eight times the performance.
-
NVIDIA and Schneider Electric test digital twin to model AI data center power and cooling in Portugal
Adding powerful AI accelerators to a data center requires more than available rack space. Electrical distribution, liquid cooling and workload characteristics must all be considered. On October 21, 2026, Schneider Electric, NVIDIA and several partners introduced a digital-twin pilot at Portugal's SINES Data Campus, designed to model these dependencies before infrastructure changes are made. The project could reduce deployment risks, although no independently verified performance or energy savings have yet been published.
-
NVIDIA Vera doubles Modal's AI agent throughput in early tests, challenging Intel and AMD on CPUs
NVIDIA's push into server CPUs is gaining traction with AI infrastructure providers. On October 19, 2026, Modal published early benchmarks showing that its first Vera server could complete up to twice as many agent-oriented tasks under heavy concurrent load as a comparable Intel Granite Rapids system. Modal intends to introduce the processor alongside its existing Intel and AMD hardware, although the performance results remain company-run measurements rather than independently validated production benchmarks.
-
NVIDIA faces a reported $35 billion China GPU order surge that production may not meet
A workstation GPU has become one of the most sought-after AI processors in China. Reuters reports that ByteDance, Alibaba and Tencent have placed orders worth more than $35 billion at list prices for NVIDIA's RTX PRO 5500. The reported demand extends far beyond the company's expected supply, creating an unusual situation in which a professional graphics card is being pursued as an alternative to restricted data-center accelerators.
-
Nvidia DLSS 5 now runs on AMD Radeon GPUs through a mod — but there's a catch
Nvidia's DLSS technologies are no longer entirely confined to GeForce hardware, at least for enthusiasts willing to experiment. Developer Daniel Blanco has expanded DLSS-NR-on-AMD to support the four major DLSS modules on Radeon RX 7000 and RX 9000 GPUs. The new alpha 0.8.0 release, covered on October 18, 2026, also introduces broader feature management through NeuralHub. But getting the software to run is not the same as matching official Nvidia performance or compatibility.
-
DLSS 5 adds a neural pass after the game engine finishes rendering
NVIDIA has published new technical details on DLSS 5's 3D-Guided Neural Rendering. Unlike Super Resolution or Multi Frame Generation, this feature is less about reconstructing pixels or producing extra frames than altering the final rendered image itself. NBA 2K27 is the first commercial implementation and already exposes both sides of the approach: fine-grained engine integration and a heavy Tensor workload.
-
Nvidia is worth more than $5.5 trillion: the huge valuation gap inside AI
There is no single useful number for "the value of AI." By late September 2026, Nvidia is worth more than $5.5 trillion on the public market while CoreWeave remains below $50 billion. Between them sit cloud giants, chipmakers and software companies financing the same computing boom.
-
Nvidia just made $89 billion from data centers in a single quarter
Nvidia's quarterly numbers have become so large that percentages can obscure the actual scale. Its latest Data Center figure does the opposite: $89 billion in three months from the business supplying the infrastructure behind the AI buildout.
-
NVIDIA's Rubin transition is already showing up in revenue before Blackwell has slowed down
New GPU generations normally create an awkward handover. Customers know faster hardware is coming, suppliers need to retool, and the previous architecture risks losing momentum before the replacement reaches volume. NVIDIA is attempting that transition while Data Center revenue is still growing 117% year over year.
-
H200 is old by NVIDIA's roadmap and still valuable enough to fight over in China
Calling H200 a last-generation GPU is technically fair and practically misleading. Chinese AI companies are not shopping from NVIDIA's complete 2026 catalog. Export licenses decide which generations can become products at all.