Nvidia is kicking off 2025 with a bang with a stone of items that had cement its position as the leader in the areas of AI growth and games following a record-breaking 2024.
CEO Jensen Huang took the stage at Les in Las Vegas to present new hardware and software options spanning everything from specific AI mainframes to next-generation entertainment cards.
Nvidia’s biggest news: Project DIGITS, a$ 3, 000 individual AI computer that packs a petaflop of computing power into a desktop-sized field.
Built around the new—and up until now, secret—GB10 Grace Blackwell Superchip, this system is control AI models with up to 200 billion parameters while drawing energy from a regular store.
For heavier loads, people you link two devices to handle models up to 405 billion parameters.
For perspective, the largest Llama 3.2 design, the most advanced open-source LLM from Meta, has 405 billion criteria and cannot be run on consumer electronics.
Up until now, it required around 8 Nvidia A100/H100 Superchips, each one costing around$ 30K, totaling more than$ 240K just in processing hardware.
Two of Nvidia’s fresh consumer-grade AI supercomputers had cost$ 6K and be capable of running the same normalized design.
” AI will be used in every industry application. With Project DIGITS, the Grace Blackwell Superchip comes to millions of developers”, Jensen Huang, CEO of Nvidia, said in an official blog post. Every data scientist, AI researcher, and student can become more engaged and influence the development of the AI age by placing an AI supercomputer on their desks.
The GB10 chip represents a significant engineering breakthrough created as a result of a collaboration with MediaTek for those who enjoy technical details.
The system-on-chip combines Nvidia’s latest GPU architecture with 20 power-efficient ARM cores connected via NVLink-C2C interconnect.
Each DIGITS unit has up to 4TB of NVMe storage and 128GB of unified memory. Again, for context, the most powerful GPUs to date pack around 24GB of VRAM ( the memory required to run AI models ) each, and the H100 Superchip starts at 80GB of VRAM.
Nvidia’s plans to dominate AI agents
Nvidia has developed a new family of models that comes in three sizes, Nemotron, and it is one of the reasons it is expanding with two new models: Nvidia NIIM for video summarization and understanding and Nvidia Cosmos for Nemotron vision capabilities, or the ability to understand visual instructions. Companies are rushing to deploy AI agents, and Nvidia knows this.
Until now, the LLMs were only text-based. However, the models excelled at the following instruction: chat, function calls, coding, and math tasks.
They’re available through both Hugging Face and Nvidia’s website, with enterprise access through the company’s AI Enterprise software platform.
Again, for context, In the LLM Arena, Nvidia’s Llama Nemotron 70b ranks higher than the original Llama 405b developed by Meta. It also beats different versions of Claude, Gemini Advanced, Grok-2 mini and GPT-4o.
Nvidia’s agent push is now also related to infrastructure. The company announced partnerships with major agentic tech providers like LangChain, LlamaIndex, and CrewAI to build blueprints on Nvidia AI Enterprise.
These ready-to-use templates address specific tasks that make it simpler for developers to create highly specialized agents.
A new PDF-to-podcast blueprint aims to compete with Google’s NotebookLM, while another blueprint helps build video search and summary agents. Developers can use the new Nvidia Launchables platform, which enables one-click prototyping and deployment, to test these blueprints.
Gamers, rejoice! Performance-Defying Cards for the New GeForce RTX 5000 Cards
Nvidia saved its gaming announcements for last, unveiling the much-expected GeForce RTX 5000 Series. The RTX 5090, the industry’s top device, has 3, 352 trillion AI operations per second, which is twice the performance of the RTX 4090, which has 92 billion transistors in it. The entire lineup features fifth-generation Tensor Cores and fourth-generation RT Cores.
The new cards introduce DLSS 4, which uses AI to generate multiple frames per render and increase frame rates by up to 8x. Blackwell, the engine of AI, has arrived for PC gamers, developers and creatives,” Jensen Huang said”, fusing AI-driven 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 employ transformer models for super-resolution, promising highly realistic graphics and a lot more performance for their price—which is not cheap, btw:$ 549 for the RTX 5070, with the 5070 Ti at$ 749, the 5080 at$ 999, and the 5090 at$ 1, 999.
If you don’t have that kind of money and want to game, don’t worry.
AMD also announced today its Radeon RX 9070 series. The cards are based on the new RDNA 4 architecture, which employs a 4nm manufacturing process, and feature specialized AI accelerators to compete with Nvidia’s tensor cores.
While full specifications remain under wraps, AMD’s latest Ryzen AI chips already achieve 50 TOPS at peak performance.
Sadly, Nvidia is still the king of AI applications thanks to its CUDA technology, Nvidia’s proprietary AI architecture.
Over 100 enterprise platform brands will use AMD Pro technology through 2025 in order to address this issue, and AMD has established partnerships with HP and Asus for system integration.
In Q1 2025, Nvidia will likely launch the Radeon cards, which will give it an intriguing battle between gaming and AI acceleration.
Generally Intelligent Newsletter
A generative AI model’s generative AI model, Gen, tells a weekly AI journey.
