- AI Valley
- Posts
- Nvidia revealed $3000 AI supercomputer
Nvidia revealed $3000 AI supercomputer
PLUS: Google is building a virtual replica of Physical World
Together with
Howdy! It’s Barsee again.
Happy Wednesday, AI family, and welcome back to AI Valley.
In today’s edition:
NVIDIA announces World’s smallest AI Supercomputer
Google is building a virtual replica of Physical World
Plus trending AI tools, posts, and resources
Ready, set, go…
CES 2025
NVIDIA Announces World’s Smallest AI Supercomputer
NVIDIA has introduced Project DIGITS, a compact AI supercomputer designed for individual use by researchers, data scientists, and students. Unveiled at CES 2025, this device brings the power of AI directly to your desk.
Here's what you need to know:
Powered by the NVIDIA GB10 Grace Blackwell Superchip, Project DIGITS delivers up to one petaflop of AI computing performance, enabling efficient handling of large AI models and complex computations.
Despite its immense power, the system is roughly the size of a small book, making it suitable for desktop use without requiring extensive space.
It operates as a standalone PC and can also connect effortlessly to Windows or Mac systems, providing flexibility in various computing environments.
Equipped with 128GB of unified memory and up to 4TB of storage, it supports a wide range of AI workloads, from model development to deployment.
A single unit can manage AI models with up to 200 billion parameters, and linking two units increases this capacity to 405 billion parameters, accommodating both small and large-scale AI projects.
NVIDIA plans to launch Project DIGITS in May 2025, with prices starting at $3,000.
Why it matters:
Traditionally, running advanced AI models required access to large, remote data centers, limiting accessibility for many individuals and smaller organizations. Project DIGITS democratizes AI by providing a powerful, personal supercomputer that reduces dependence on cloud infrastructure, fostering innovation and experimentation in AI development at the individual level.
TOGETHER WITH SUPERHUMAN
The AI-Powered Inbox That Works for You
If you're anything like most AI professionals, your inbox isn't just full—it's never-ending chaos. Superhuman is your solution. That's why we built Superhuman—the most productive email experience ever made (and it works with your existing Gmail or Outlook setup).
Think:
AI summaries that let you skip the fluff.
Split Inbox to organize your email and focus on what actually matters.
Keyboard shortcuts so fast, you'll feel like you're coding your way through email.
This January, don't just clear your inbox—reclaim your time. Get one month free of Superhuman.
SIDE UPDATES
Sam Altman, CEO of OpenAI, predicts virtual employees (autonomous AI agents) could revolutionize workplaces as early as this year. OpenAI is developing its own agent, "Operator," to handle tasks like coding and travel booking autonomously. Early adopters, including Microsoft and McKinsey, see these tools transforming productivity, with McKinsey estimating up to 30% of U.S. work hours could be automated by 2030.
Nvidia has launched Cosmos, a family of foundational AI models specifically created to enhance the capabilities of robots to better understand and navigate the physical world. Trained on 20 million hours of human movement data, it generates 3D realistic video simulations to train robots for real-world scenarios. Companies in robotics, including Agility and Figure AI, are already integrating Cosmos into their systems.
NeuroAccess, a Chinese startup, has recently introduced a mind-reading device that turns thoughts into speech in real-time. In successful trials, the device was implanted into an epileptic patient’s brain, allowing it to not only translate complex Chinese but also control a robotic arm and interact with AI. It achieved 71% speech decoding accuracy and less than 100 milliseconds of latency just five days after implantation.
Anthropic is close to securing $2 billion in the latest funding round led by Lightspeed Venture Partners, which would triple its valuation to $60 billion, making it the fifth most valuable startup in the U.S. Amazon and Google continue to back the company, with Amazon increasing its investment to $8 billion recently, while Google has committed $1.5 billion in funding over time.
HeyGen just announced the integration of its new digital human model with OpenAI's Sora model. It allows virtual avatars to blend seamlessly into Sora-generated scenes and, in some cases, even outperform real actors. Unlike digital clones, these avatars are entirely AI-generated. Users can easily adjust their actions, expressions, and postures, and with no length limitations on videos, they now have the freedom to create longer projects.
WORLD MODELS
Google Is Building a Virtual Replica of Physical World
Google DeepMind recently announced the formation of a new team dedicated to developing generative AI systems known as "world models." These systems are designed to simulate physical environments, aiming to enhance AI’s learning and planning capabilities for a wide range of applications.
Here's what you need to know:
The initiative will be led by Tim Brooks, who previously led OpenAI’s Sora project and joined DeepMind in October.
The development process will leverage large-scale datasets, including video and multimodal content, to train the world models.
They will work alongside existing Google AI projects, such as Gemini AI models, Veo video generators, and Genie (a prior world model for simulating playable 3D environments in real-time.)
The primary goal is to create AI models capable of generating realistic, interactive simulations of the physical world. These simulations can be applied to video games, movies, and training scenarios for robots and AI systems, thereby enhancing realism and interactivity.
Are there any other players in this space?
Several competitors already have a head start over DeepMind’s world modeling project, including Nvidia’s Cosmos platform for advancing physical AI, autonomous vehicle, and robot development, and the World Labs startup created by “the godmother of AI,” Fei-Fei Li.
Introducing 🧞Genie 2 🧞 - our most capable large-scale foundation world model, which can generate a diverse array of consistent worlds, playable for up to a minute. We believe Genie 2 could unlock the next wave of capabilities for embodied agents 🧠.
— Jack Parker-Holder (@jparkerholder)
2:24 PM • Dec 4, 2024
TRENDING TOOLS
Journals AI > 30 day journals tailored to you. (link)
Socra > A context-aware AI that collaborates with you and your team for enhanced productivity and decision-making. (link)
Project Ambience > Get deep work done with AI-tailored ambiance mixes. (link)
WikiTimeline > Turns any Wikipedia article into an interactive timeline for easy visualization and comparison. (link)
VocAdapt > Automatically adjusts content to your understanding level, making it easier to read, watch, and listen to content in a new language without difficulty. (link)
COOL FINDINGS / RESOURCES
CONTENT CORNER
1/
This guy is unbelievable. Every generation has a few. Henry Ford. Bill Gates and Steve Jobs. And now Jensen Huang.
2/
This man is literally telling us how to MAKE MILLIONS OF DOLLARS and everyone is half asleep in that crowd.
I am about to have the BIGGEST YEAR OF MY LIFE! $NVDA
— DekmarTrades via TradeCaster (@DekmarTrades)
3:33 AM • Jan 7, 2025
AI agents will be the next multi-trillion-dollar opportunity. And another big thing will be software-coding AI that will help every software engineer code like never before.
3/
$NVDA announces partnership with $TM to develop next-gen autonomous vehicles 👀
— Shay Boloor (@StockSavvyShay)
3:58 AM • Jan 7, 2025
Nvidia partners with Toyota to develop an autonomous vehicle. Considering the size and quality of both companies, this could be the best partnership ever.
4/
CES 2025 just started, and we’re already seeing incredible tech.
Here're 20 most impressive reveals of CES 2025 so far:
1. Halliday Glasses: 3.5-inch AI smart glasses
— Hasan Toor ✪ (@hasantoxr)
8:41 PM • Jan 7, 2025
This year's CES already shows the technological leap that is underway. With the development of AI and ASI, the evolution will be impressive in the coming years. We are getting closer and closer to the conquest of space and immortality.
5/
Nvidia CEO Jensen Huang tonight on robotaxis and self-driving: “I predict that this is going to be the first multi-trillion dollar robotics industry.”
— Sawyer Merritt (@SawyerMerritt)
5:12 AM • Jan 7, 2025
Nvidia CEO Jensen Huang predicts that robotaxis and self-driving cars are going to be the first multi-trillion dollar robotics industry.
6/
Meet The Enron Egg, the world’s first at-home nuclear reactor. Watch the historic reveal.
— Enron (@Enron)
8:33 PM • Jan 6, 2025
This is a joke (which recently went viral), but it's actually kind of a good idea. Home nuclear reactor to power your home compute cluster and to run your private, local AI.
THAT’S ALL FOR TODAY
That’s all for today’s issue, folks.
💡 Help me get better and suggest new ideas at [email protected] or @heyBarsee
👍️ Like what you see? Subscribe here
Thanks for being here.
HOW WAS TODAY'S NEWSLETTER |
REACH 100K+ READERS
Acquire new customers and drive revenue by partnering with us
Sponsor AI Valley and reach over 100,000+ entrepreneurs, founders, software engineers, investors, etc.
If you’re interested in sponsoring us, email [email protected] with the subject “AI Valley Ads”.