š The Key Takeaways from Nvidia's GTC 2025 Keynote and Why We Stay Bullish
Jensen Huang unveils Nvidiaās next wave of innovations: from new AI chips to robotics and networking.
Jensen Huang shared Nvidiaās NVDA 0.00%ā key plans and developments for AI chips, robotics, and networking infrastructure during the GTC 2025 keynote yesterday. The market didnāt react strongly, but weāre very excited about Nvidiaās future. Weāll dive into that later in this article, but letās first take a look at the most important announcements from the keynote.
Key announcements
Nvidiaās GPU roadmap for the future
Nvidia has some exciting chips coming in the next few years. Here are the key ones to watch:
Blackwell Ultra (H2-2025): Coming at the end of 2025, the Blackwell Ultra will deliver 20 petaflops of computing power. The biggest upgrade is in memory, going from 192 GB to 288 GB, making it perfect for AI and data center tasks.
Vera Rubin (H2-2026): Expected in late 2026, this chip combines a Vera CPU and Rubin GPU, offering 50 petaflopsāover double the Blackwell Ultraās power.
Rubin Ultra (H2-2027): Arriving in 2027, the Rubin Ultra will deliver a huge 100 petaflops, making it the most powerful chip for AI and complex applications.
Feynman (2028): The Feynman generation, arriving in 2028, has no details yet, but itās set to be a major step forward.
Dynamo: an open-source AI system
Nvidia introduced Dynamo, an open-source AI platform. Dynamo will help companies quickly scale AI models without being tied to closed systems. By offering Dynamo, Nvidia can reach a larger market and provide more flexible AI solutions.
Blue: Nvidiaās humanoid robot
Nvidia unveiled Blue, a humanoid robot powered by Nvidiaās AI and physics engine. Blue can adapt and learn, marking a big step forward for robotics and showing that Nvidia's technology goes beyond computer chips.
DGX personal AI computers
DGX Personal AI Computers are powerful desktop systems designed for smaller companies and developers. Powered by Blackwell Ultra chips, these systems allow businesses to build powerful AI models without needing access to massive data centers.
Partnership with General Motors
Nvidia is partnering with General Motors to develop AI solutions for self-driving cars and smart factories. This partnership shows how Nvidia is branching out into industries like automotive and robotics.
Quantum-X and Spectrum-X network chips
Nvidia introduced Quantum-X and Spectrum-X, two network chips that use photonic technology to connect GPUs across the world. These chips make networks faster and more energy-efficient, which is essential for the growing demand for AI applications.
Isaac GR00T N1: The new robotics platform
The Isaac GR00T N1 platform helps robots learn and adapt independently, making them more useful for industrial and commercial applications.
Why We Are Still Bullish on Nvidia
Nvidiaās stock fell by 3% during the keynote, as investors expected more specific details or faster progress on new technologies. However, after hearing the keynote, weāre even more excited about Nvidiaās future. We believe the company has strong potential, and hereās why we think Nvidia will keep growing in the years ahead.