After a Record breaking 2024Nvidia is starting 2025 with a bang, as it unveils a slate of products that could cement its dominance in the areas of artificial intelligence and gaming development.

CEO Jensen Huang took the stage at CES in Las Vegas to showcase new hardware and software offerings that include everything from AI-powered super PCs to next-generation gaming cards.

Nvidia's biggest announcement: Project DIGITS, a $3,000 personal AI supercomputer that packs petaflops of computing power into a desktop-sized box.

Built around the new – and so far secret – GB10 Grace Blackwell Superchip, the machine can handle AI models with up to 200 billion parameters while drawing power from a standard outlet.

For heavy workloads, users can connect two modules to process models with up to 405 billion parameters.

For context, the largest Llama 3.2 model, Meta's most advanced open source LLM software, contains 405 billion parameters and cannot run on consumer hardware.

Until now, it required about 8 Nvidia A100/H100 Superchips, each one costing About 30 thousand dollarstotaling over $240,000 just in hardware processing.

Two of Nvidia's new consumer AI supercomputers will cost $6,000 and will be able to run the same quantum model.

“AI will be prevalent in every application for every industry. With Project DIGITS, the Grace Blackwell Superchip reaches millions of developers,” Nvidia CEO Jensen Huang said in an official statement. Blog post. “Putting an AI supercomputer on the desks of every data scientist, AI researcher and student empowers them to participate and shape the AI ​​era.”

For those who like technical details, the GB10 chip represents a significant engineering achievement born from the collaboration with MediaTek.

The system-on-chip combines Nvidia's latest GPU architecture with 20 power-efficient ARM cores connected via NVLink-C2C interconnect.

Each DIGITS module has 128GB of unified memory and up to 4TB of NVMe storage. Again, for context, the most powerful GPUs to date have about 24GB of VRAM (the memory required to run AI models) each, and the H100 Superchip starts at 80GB of VRAM.

Nvidia plans to take over AI agents

Companies They are rushing to deploy AI agentsNvidia knows this, and that's probably why Nemotron developera new family of models that come in three sizes, announced its expansion today with two new models: Nvidia NIIM for video summarization and understanding and Nvidia Cosmos to give Nemotron Vision capabilities — the ability to understand visual instructions.

Until now, LLMs have been text-based only. However, the models excelled in the following instructions: chat, function calls, programming, and mathematical tasks.

It is available through Hugging Face and Nvidia's website, with enterprise access through the company's AI Enterprise software platform.

Again, for context, in LLM ArenaNvidia's Llama Nemotron 70b ranks higher than the original Llama 405b developed by Meta. It also outperforms the Claude, Gemini Advanced, Grok-2 mini, and GPT-4o variants.

Nvidia proxy payment is now also tied to infrastructure. The company announced partnerships with leading proxy technology providers like LangChain, LlamaIndex, and CrewAI to build blueprints on Nvidia AI Enterprise.

these Templates ready to publish Addressing specific tasks makes it easier for developers to create highly specialized agents.

A new scheme for turning PDF files into podcasts aims to compete with Google's NotebookLM, while another scheme helps create search and video summary operators. Developers can test these schemes through the new version Nvidia launchable The platform, which enables the creation and deployment of prototypes with one click.

Players, rejoice! The new GeForce RTX 5000 cards are a performance beast

Nvidia saved its gaming announcements for last, revealing what was long overdue GeForce RTX 5000 series. The flagship RTX 5090 features 92 billion transistors and delivers 3,352 trillion AI operations per second – twice the performance of the current RTX 4090. The entire lineup features 5th generation Tensor Cores and 4th generation RT Cores.

The new cards offer DLSS 4, which can boost frame rates by up to 8x using AI to create multiple frames per display. Blackwell, the AI ​​engine, has arrived for PC gamers, developers, and creators He said"By integrating AI-based neural rendering and ray tracing, Blackwell is the most significant computer graphics innovation since we introduced programmable shading 25 years ago."

The new cards also use adapter models for superior resolution, and promise ultra-realistic graphics and a lot more performance for their price — which isn't cheap, BTW: $549 for the RTX 5070, 5070 Ti at $749, 5080 at $999, 5090 at $1,999 .

If you don't have that much money and want to play, don't worry.

AMD too Announce today Radeon RX 9070 series. The cards are built on the new RDNA 4 architecture using a 4nm manufacturing process and feature custom AI accelerators to compete with Nvidia's tensor cores.

While full specifications are still under wraps, AMD's latest Ryzen AI chipset is already achieving 50 TOPS at peak performance.

Unfortunately, Nvidia is still the king of AI applications thanks to its capabilities CODA technologyNvidia's AI architecture.

To address this issue, AMD has partnered with HP and Asus for system integration, and more than 100 enterprise platform brands will use AMD Pro technology through 2025.

Radeon cards are expected to hit the market in the first quarter of 2025, giving Nvidia an interesting fight in both gaming and AI acceleration.

Modified by Sebastian Sinclair