Today we're releasing @nvidia AlpaGym, our new open-source reinforcement learning (RL) framework for end-to-end autonomous driving. A key challenge for #Physical# #AI# is enabling policies to learn from the consequences of their actions. While supervised learning can teach a model to imitate behavior, robust autonomy ultimately requires learning through interaction with the environment. AlpaGym enables exactly that. Built on top of: - AlpaSim: our high-fidelity closed-loop autonomous driving simulator - Cosmos-RL: NVIDIA's distributed RL training and rollout infrastructure AlpaGym provides the glue that connects simulation, training, and driving policies into a scalable framework for post-training autonomous vehicle models in closed loop. With AlpaGym, researchers and developers can: ✅ Train end-to-end driving policies using reinforcement learning ✅ Run large-scale closed-loop simulations ✅ Experiment with new reward functions, policy architectures, and training strategies ✅ Benchmark models on public leaderboards 📖 Learn how it works: 💻 GitHub: 🏆 Open Challenges: - AlpaSim Closed-Loop E2E Driving Challenge: - Physical AI AV Reasoning Challenge: Learn more about the #Alpamayo# open platform: #PhysicalAI# #AutonomousDriving# #ReinforcementLearning# #Robotics# #OpenSource# #NVIDIA# #MachineLearning# @NVIDIADRIVE @NVIDIAAI

Momenta — the “First Physical AI Stock” Officially Lists Today, the “first Physical AI stock” Momenta officially listed on the Hong Kong Stock Exchange. Its opening share price rose more than 6%, making it one of the most talked-about tech IPOs in the Hong Kong market this year. My first reaction was to check its cornerstone investor lineup: 14 institutions joined forces to underwrite it, with total cornerstone subscriptions of approximately HK$3 billion (about US$376 million). The quality of this lineup is rare among recent Hong Kong IPOs. What is “Physical AI”?
Momenta’s core weapon is its world model — the R7 world model, which entered mass production for the first time in April this year. It supports scaled deployment in passenger cars, Robotaxi, and Robovan, and will later expand into Robotruck and embodied intelligence. Simply put: digital AI (such as GPT) competes in text and images, while Physical AI competes in the physical laws of the real world — and this race has only just begun. Supporting this narrative are solid scale metrics: mass-produced vehicles have surpassed 1 million units, with the fastest delivery of 100,000 units completed in under 40 days. The “flywheel effect” is already in motion. Tech companies are often criticized for “high R&D and difficulty in turning a profit,” but Momenta has achieved explosive revenue growth this time, with cash reserves exceeding RMB 10 billion on its books. This breaks the curse and gives it the confidence to meet the allocation standards of top-tier long-term capital. On the capital side: How crazy was this round of subscription?
Both the public offering and international offering saw oversubscription. In the cornerstone list, GIC and Fidelity International led strongly, each investing US$100 million; BlackRock, Oaktree Capital, and Franklin Templeton followed — with Franklin Templeton appearing on a Hong Kong cornerstone list for the first time in history. On the industrial side it is even more solid, with Mercedes-Benz and BYD both increasing their stakes. The domestic camp includes top players such as Gaoyi, Boyu, China Asset Management, GF Fund, and Pacific Insurance, all of which participated. Long-term capital is famously selective in its taste — it wants hard assets that can ride through cycles and have proven commercial models. So many top institutions betting on one company at the same time shows that the market already regards Momenta as a core, unavoidable target in the AI track. In closing
It is uncommon in the Hong Kong market for a company to simultaneously win the unanimous endorsement of sovereign funds, global top asset managers, industrial giants, and Chinese private equity, mutual funds, and insurance capital. Whether the “GPT moment” of Physical AI will truly arrive still needs time to verify, but at least from the capital reaction to this IPO, the market has already cast its judgment with real money. I’ll end by borrowing a line from inside Momenta: “We hope to write the Eastern Silicon Valley legend together with all Chinese AI companies, embrace the world, and let the world embrace China’s technology.” Wishing Momenta — and all friends following this track — the same luck as its stock code: 688, smooth and prosperous! #PhysicalAI# #HongKongIPO# #AutonomousDriving# #Momenta# #EmbodiedIntelligence#

Autonomous driving is amazing but the true dream will be when the car drops me off right where I’m going and goes off to deal with parking on its own, and then comes back and picks me up when I’m ready to leave

Final day at @Ai4Conferences with @teslaownersSV : from @Waymo’s journey in autonomous driving, safety and world models to enterprise AI, agent connectivity and ROI, the focus was clear—turning AI innovation into real-world impact. #ChinaAMC# #CNQQ# #AI4# #PhysicalAI# #AgenticAI# #EnterpriseAI# Risk Disclaimer: Investment involves risk, including possible loss of principal. Any forecasts, projections, or opinions contained herein are for reference only and are not guaranteed to occur. The information in this material reflects prevailing market conditions and our judgment as of the release date, which are subject to change without further notice.

Today in Caixin’s Business Brief: - Beijing issues mandatory national safety standards for autonomous driving - Consumer electronics prices rise on memory chip shortage, surging raw material costs

We have just released #Alpamayo# 2 Super — @nvidia’s frontier open reasoning model for #autonomous# #vehicles#. NVIDIA Alpamayo 2 Super is an open 34-billion-parameter #reasoning# vision-language-action (#VLA#) model designed to accelerate autonomous vehicle (AV) development. It combines the 32-billion-parameter NVIDIA #Cosmos# 3 Super Reasoner with a 2-billion-parameter diffusion-based Action Expert and is post-trained with #reinforcement# #learning#. Two aspects make Alpamayo 2 Super particularly exciting: 1. Open and commercially deployable: Alpamayo 2 Super is available on Hugging Face under OpenMDW-1.1, the Linux Foundation’s permissive license for open AI model distribution. The OpenMDW license is now being applied across the entire Alpamayo model family, enabling developers to deploy these models commercially without requiring additional permissions. 2. A multi-task foundation model for autonomous driving: Alpamayo 2 Super produces five tightly coupled outputs: - A trajectory describing the vehicle’s planned path. - A chain-of-causation (CoC) trace explaining the reasoning behind each driving decision, achieving benchmark-leading reasoning performance at frontier scale. - A meta-action (e.g., yield, lane change, stop) capturing the model’s intent. - Reasoning auto-labels that generate CoC annotations for training and validation data. - Visual question answering responses with 2D visual grounding, linking answers to specific regions in camera images. These multi-task capabilities enable developers to leverage a single foundation model across more of the development process, simplifying tooling and accelerating iteration. Resources: 🔹 Interview: 🔹 Video: 🔹 Blog: 🔹 Technical blog: 🔹 Hugging Face blog: 🔹 Model weights: 🔹 Inference code: As @JensenHuang has emphasized, open models help advance safety and security. We hope Alpamayo 2 Super will contribute to this vision by enabling researchers and developers around the world to build, experiment, and innovate in autonomous driving. We are excited to see what the community builds with Alpamayo 2 Super. @NVIDIADRIVE @NVIDIAAI