Nvidia Corporation Stock Price, News & Analysis

NVIDIA (NASDAQ:NVDA) announced the NVIDIA Inference Context Memory Storage Platform, powered by the BlueField-4 data processor, a purpose-built AI-native storage platform for long-context, agentic AI workloads announced Jan 5, 2026.

The platform extends GPU memory with cluster-level KV cache, enables high-bandwidth sharing across rack-scale systems, and claims up to 5x improvement in tokens-per-second and up to 5x greater power efficiency versus traditional storage. Key components include BlueField-4 hardware acceleration, NVIDIA DOCA, NIXL, Dynamo software, and Spectrum-X Ethernet. Major storage vendors are building systems with BlueField-4, which will be available in the second half of 2026.

NVIDIA (NVDA) launched the Rubin platform on Jan 5, 2026: a codesigned rack-scale AI system built from six chips — Vera CPU, Rubin GPU, NVLink 6 switch, ConnectX-9 SuperNIC, BlueField-4 DPU and Spectrum-6 Ethernet switch — to accelerate training and inference.

Key claims: up to 10x lower inference token cost and 4x fewer GPUs for MoE training versus the Blackwell platform; Rubin GPU delivers 50 petaflops NVFP4; NVLink 6 offers 3.6TB/s per GPU and NVL72 rack 260TB/s. Rubin products enter production with partner availability in H2 2026.

NVIDIA (NVDA) unveiled new physical AI models, open-source simulation and orchestration frameworks, and Jetson and Thor robotics compute at CES on Jan 5, 2026.

Key releases include Cosmos Transfer 2.5, Cosmos Predict 2.5, Cosmos Reason 2, the GR00T N1.6 VLA model, Isaac Lab-Arena, and OSMO, plus integrations with Hugging Face and LeRobot. NVIDIA also announced the Jetson T4000 module (1,200 FP4 TFLOPS, 64GB) priced at $1,999 at 1,000-unit volume and availability of IGX Thor for industrial edge AI.

NVIDIA (NVDA) on Jan 5, 2026 at CES unveiled the Alpamayo family: open-source reasoning AI models, simulation tools and datasets to speed development of safe, reasoning-based autonomous vehicles.

Key components: Alpamayo 1 — a 10-billion-parameter chain-of-thought vision-language-action model on Hugging Face with open weights and inference scripts; AlpaSim — an open-source, end-to-end simulation framework on GitHub; and Physical AI Open Datasets — 1,700+ hours of driving data covering rare edge cases. Industry partners cited include JLR, Lucid, Uber and Berkeley DeepDrive. NVIDIA positions these tools as teacher models for fine-tuning, distillation and closed-loop validation under its Halos safety system.

NVIDIA (NVDA) launched the Nemotron 3 family of open models, datasets and libraries on Dec 15, 2025, targeting agentic multi-agent AI systems.

The lineup includes Nemotron 3 Nano (30B), Nemotron 3 Super (~100B) and Nemotron 3 Ultra (~500B). Nemotron 3 Nano is available today and claims 4x token throughput vs Nemotron 2 Nano, a 1M-token context window and up to 60% reduction in reasoning-token generation. NVIDIA released 3 trillion tokens of training/post-training/RL data plus NeMo Gym, NeMo RL and NeMo Evaluator; Super/Ultra expected H1 2026.

NVIDIA (NASDAQ: NVDA) and Synopsys (NASDAQ: SNPS) announced a multi-year strategic partnership on Dec 1, 2025 to integrate NVIDIA accelerated computing, agentic AI, and Omniverse digital twins into Synopsys engineering and simulation tools.

Key facts: NVIDIA invested $2.0 billion in Synopsys common stock at a purchase price of $414.79 per share. Initiatives include CUDA-X acceleration across Synopsys applications, integration of AgentEngineer with NVIDIA’s agentic AI stack, cloud-ready GPU engineering solutions, and joint go-to-market activities using Synopsys' global sales network.

NVIDIA (NASDAQ: NVDA) will present at the UBS Global Technology and AI Conference on Tuesday, Dec. 2, 2025 at 6:35 a.m. Pacific time. Interested parties can listen to the live audio webcast at investor.nvidia.com. Webcast replays will be available for 90 days after the presentation.

NVIDIA (NASDAQ: NVDA) reported record Q3 fiscal 2026 revenue of $57.0 billion, up 22% sequentially and 62% year-over-year, driven by a record Data Center quarter of $51.2 billion (+25% Q/Q, +66% Y/Y). GAAP and non-GAAP diluted EPS were both $1.30. GAAP gross margin was 73.4%; non-GAAP gross margin was 73.6%. Operating income was $36.0 billion (GAAP). During the first nine months of FY26 the company returned $37.0 billion to shareholders, with $62.2 billion remaining on the buyback authorization. NVIDIA issued Q4 FY26 guidance of $65.0 billion ±2% revenue and target non-GAAP margin ~75.0%.

NVIDIA (NASDAQ:NVDA) announced that NVQLink, a universal interconnect for quantum-classical systems, is being adopted by more than a dozen global supercomputing centers and national labs to link quantum processors with NVIDIA GPUs via the CUDA-Q platform.

NVQLink delivers 40 petaflops AI performance at FP4, a GPU–QPU throughput of 400 Gb/s and latency under 4 microseconds. Quantinuum’s Helios QPU used NVQLink to demonstrate the world’s first scalable real-time decoder with a 67 microsecond reaction time (32x faster than a 2 ms requirement).