Autonomous systems need both accurate sensing and reliable connectivity. #FMCW# #LiDAR# delivers precise motion and distance data, while high-frequency #wireless# enables real-time communication and response. See how Sivers Semiconductors is powering next-generation sensing and connectivity: #AutonomousSystems# #Robotics# #EdgeAI#

How do we make robot policies robust to rare but high-impact failures? Video #World# #Models# (WMs) are rapidly becoming a powerful tool for robotics, enabling policy evaluation and improvement by "imagining" future outcomes. But there's a catch: these imagined futures are typically nominal samples, making it easy to overlook the rare yet safety-critical events that matter most. In our new paper, StressDream: Steering Video World Models for Robust Policy Evaluation and Improvement, we explore a simple but powerful idea: 💡 Instead of passively sampling futures, actively steer world model imaginations toward high-impact yet still plausible scenarios. StressDream optimizes the initial diffusion noise at inference time, allowing us to generate targeted stress-test scenarios without retraining the world model. This enables: - More robust policy evaluation by exposing failure modes that random sampling often misses. - Improved policy optimization by training against challenging but realistic imagined futures. As generative world models become a foundation for #Physical# #AI#, the ability to systematically probe their "long tail" of plausible futures will be increasingly important for building reliable and trustworthy autonomous systems. 📌 𝖯𝗋𝗈𝗃𝖾𝖼𝗍 𝖯𝖺𝗀𝖾: 📄 𝖯𝖺𝗉𝖾𝗋: Work led by Junwon Seo, with a great set of collaborators: Sushant Veer, Thomas Ran Tian, Wenhao Ding, Apoorva Sharma, Karen Leung, Edward Schmerling, Andrea Bajcsy. @NVIDIADRIVE @NVIDIAAI #Robotics# #WorldModels# #PhysicalAISafety# #AISafety# #AutonomousSystems# #RobotLearnin#

"The Age of Autonomous Systems" isn't a forecast. It's the stack being built right now: agents that act, transact, and get things done on your behalf. The premise underneath all of it is a trust layer, which is exactly what Kite is building. We are pleased to share that our Co-Founder & CEO @ChiZhangData will speak in the Tech Track at GTLC 2026 Bay Area Chinese Tech Leaders Summit, hosted by TGO Silicon Valley. She shares the stage with founders and researchers including: ▷ Yuandong Tian (@tydsh), Founder of Recursive Superintelligence, the $4.65B self-improving AI lab. Former Research Director at Meta FAIR, and earlier on Google X's self-driving car team (now Waymo). ▷ Yangqing Jia (@jiayq), creator of Caffe and Co-Founder & CEO of Lepton AI (acquired by NVIDIA). Former VP at NVIDIA, VP at Alibaba Group, and AI Director at Facebook. ▷ Junchen Jiang (@JunchenJiang), CEO of Tensormesh and Professor at UChicago, co-creator of the open-source LMCache. ▷ Bo Li, Co-Founder & CEO of Virtue AI and Professor at UIUC, a leading scholar in AI safety and trustworthy ML (IJCAI Computers and Thought Award, Sloan Fellow). ▷ Ryan Hanrui Wang, SVP Applied AI at Nebius and former Co-Founder & CEO of Eigen AI (acquired by Nebius for $643M), out of MIT's HAN Lab. ▷ Ding Zhao, Professor at CMU and director of the Safe AI Lab, focused on trustworthy and safe AI. ▷ Weiyan Shi (@shi_weiyan), Professor at Northeastern and AI2050 Fellow, who built the negotiation agent behind Meta's CICERO. 📅 July 17, 2026 · Tech Track 2:00 PM 📍 Q Bay Center, San Jose If you're in the Bay Area, come find us. 🪁 🔗

"America must now build autonomous systems at both the quality and quantity required to win. We have the talent advantage. We are losing the production race. The stakes are whether the United States arrives at the next conflict with overwhelming superiority in autonomy, or whether we cede that advantage to adversaries who are already building it." a16z's Daniel Penny, Alex Oliver, and Zach Chen on the moral case for autonomous warfare:

Agentic AI introduces a new model where autonomous systems can coordinate, make decisions, and engage in commerce. That raises an important question: what financial infrastructure best supports those interactions? In this clip, I share how we’re thinking about that shift and why TRON is the answer.

We are hiring! The Autonomous Systems and Physical AI Research (ASPIRE: group at @nvidia is looking for talented PhD Research Interns to join us in advancing the frontiers of #autonomous# #systems# and #Physical# #AI#. We work across a broad range of research areas, including #reasoning# models, generative simulation, #agentic# AI workflows, and Physical AI #safety#, with applications spanning autonomous vehicles and a broad range of Physical AI systems. Interested in pushing the limits of what’s possible? Apply now:

🚀 Panel Session Ethereum Meets Hardware: From Devices to Autonomous Systems 🚀Featuring: Leo Lin @linyao01, Co-founder
of Arkreen @arkreen_network Ben Zhai, CEO of @roboaiio Wallace Leung, General Manager (Hong Kong) at StarFive @StarFiveTech Nijika, Head Product Manager at Leaptic Tech Kelly Luo, VP of Faith Technology ⏰ Apr. 22 | 11:30 - 17:30 📍 STAGE 2, HK WEB3 FESTIVAL 🔗Livestream: #AppGathering# #Ethereum# #Web3Festival# #EAG#

Who owns the AI agent? As autonomous systems take on more decisions, the question of ownership, IP, and policy will become one of the most consequential debates in tech. @ysiu will join @WebX_Asia on July 13 in Tokyo for a panel on AI ownership and policy, followed by a fireside chat on what the rise of agentic AI means for the open digital economy. #WebX2026#

Excited to share the latest expansion of the @nvidia #Alpamayo# open platform for reasoning-based autonomous vehicles. Since its launch earlier this year, Alpamayo has seen rapid adoption across industry and academia, with its reasoning models surpassing 400,000 downloads and earning a #COMPUTEX# 2026 Best Choice Award. As announced by Jensen Huang during his #COMPUTEX# keynote, we are now introducing several major additions designed to accelerate the development of next-generation AV systems (more details here: 🚗 Alpamayo 2 Super — a new 32B-parameter driving foundation model with: • Full 360° surround-view perception • Advanced reasoning capabilities and chain-of-causation outputs • Meta-actions such as lane changes, yielding, and stopping • Reasoning auto-labeling and visual grounding for scalable data annotation • State-of-the-art performance across reasoning, prediction, and alignment tasks 🔄 AlpaGym — an open-source framework for closed-loop reinforcement learning, enabling AV models to learn from the consequences of their actions in simulation and helping bridge the gap between training and real-world deployment. 📊 New Open Benchmarks — including challenges for closed-loop driving and long-tail reasoning to help the community measure progress and drive innovation. 🛠️ Alpamayo Recipes — a centralized repository of end-to-end workflows covering supervised fine-tuning, reinforcement learning, quantization, and model customization. Reasoning models and closed-loop training are becoming foundational technologies for autonomous systems. Our goal is to provide the open tools, models, infrastructure, and benchmarks needed to accelerate progress across the entire AV ecosystem. A huge thank you to the many researchers, engineers, and community members whose feedback helped shape this release. Resources: • Overview of the latest Alpamayo release (note: some components will be released over the coming weeks): • @nvidia announcement: #AutonomousVehicles# #PhysicalAI# #Robotics# #AI# #MachineLearning# #ReinforcementLearning# #OpenSource# #NVIDIA# #Alpamayo# @NVIDIADRIVE @NVIDIAAI

5 Early-Stage Names With the Cash to Back Up the Story $ONDS $ABCL $SMR $JOBY $QS $ONDS — The defense and autonomous systems story here has been one of the fastest-scaling in the space, and the balance sheet backs it up. Ondas ended Q2 with roughly $1.4 billion in cash, cash equivalents, restricted cash, and short-term investments, alongside a $757 million backlog that grew 66% sequentially. Revenue is scaling fast too — Q2 came in at $83.8 million, up more than 13x year-over-year, with full-year guidance raised to $525–$550 million. The bear case worth knowing: a chunk of that cash pile has already been earmarked for recent acquisitions (DZYNE, CyberHawk), so the fortress balance sheet is partly a function of equity raises funding a roll-up strategy — worth watching how disciplined that stays. $ABCL — A biotech name with a genuinely clean balance sheet: over $565 million in cash and marketable securities, plus access to roughly $110 million in committed government funding, giving management a stated runway of at least three years. Two new partnerships with Jazz Pharmaceuticals and Vertex added over $110 million in non-dilutive upfront cash this year alone — a good sign that the platform is monetizing without constant capital raises. The lead program, ABCL635, already delivered its Phase 2 catalyst on August 10 — a single dose cut moderate-to-severe hot flash frequency by 83% versus 33% for placebo at week 4, hitting statistical significance with a clean tolerability profile. Next catalyst to watch is ABCL575 Phase 1 data, expected Q4 2026. $JOBY — This is the standout on pure balance sheet strength: roughly $2.3 billion in cash and short-term investments as of the end of Q2. That's real ammunition heading into what's shaping up to be the most important stretch yet — first eVTOL passenger flights targeted for later this year, alongside continued FAA certification progress. The offset: cash burn is heavy, with the company using about $202 million in the quarter, so the runway is strong but not infinite. $SMR — Liquidity here has ballooned to about $1.9 billion in cash, cash equivalents, and investments, up roughly $900 million in a single quarter. Revenue is essentially nonexistent right now (a byproduct of project timing, not demand), so this is a pure binary-catalyst setup — the whole thesis hinges on ENTRA1 closing a definitive agreement with the Tennessee Valley Authority. If that lands, the cash position gives NuScale the ability to execute immediately without needing to raise into a potential re-rate. $QS — Total liquidity of $859 million, split between cash/equivalents and marketable securities, funds continued scaling of the Eagle Line production process and expansion into new verticals including AI data center batteries (QSDC) and defense/aerospace (QSAS). Management has guided full-year Adjusted EBITDA loss of $250–$275 million, so the cash pile is a multi-year runway rather than a war chest for aggressive expansion — still, it removes near-term financing risk while the company works toward commercialisation. each of these names can fund its own roadmap for years without going back to the well, which takes one major risk off the table for early-stage exposure. That doesn't remove execution risk — cash doesn't guarantee contracts, certifications, or clinical data — but it does buy time for the thesis to play out without shareholders getting diluted along the way. Not financial advice.