🤖 Travis Kalanick’s ATOMS Raises $1.7B to Automate Physical Industry

• ●Travis Kalanick, Uber’s co-founder, has secured $1.7B at launch for ATOMS — one of the largest seed-stage robotics raises on record — targeting physical automation across industries including food, beverage, and mining.
• ●The scale of the raise signals institutional conviction that industrial robotics is entering a deployment phase, not just a research phase, with capital flowing before a single product has shipped at scale.
• ●Watch whether ATOMS pursues vertical integration (hardware + software + operations) as Kalanick did with Uber, or bets on a platform model — that choice will define its competitive moat.
• ●🔎 Read More →
• What matters: A $1.7B launch round for an unproven robotics startup is a market signal, not just a funding event — it tells you where the smart money thinks physical automation is heading.

🧪 Technology, Research & Innovation

🧠 VLM-IE3D Gives Vision-Language Models True 3D Spatial Reasoning from RGB Video

• ●Researchers introduced VLM-IE3D, a unified framework that injects both implicit and explicit 3D geometry into standard VLMs trained only on 2D RGB video — no depth sensors required at inference.
• ●Current VLMs fail on tasks requiring fine-grained spatial understanding; VLM-IE3D directly addresses this by learning geometry representations that persist across the model’s reasoning pipeline.
• ●If the approach generalizes, it could close a critical gap for robot manipulation and scene understanding without requiring expensive sensor upgrades on deployed hardware.
• ●🔎 Read More →
• What matters: Teaching VLMs to reason in 3D from RGB alone removes a major hardware dependency blocking their deployment in real-world robotic systems.

🚗 Knowledge Graphs + LLMs Are Being Tested as a Safety Layer for Autonomous Vehicles

• ●New research published via EurekAlert! examines how combining knowledge graphs with large language models can encode structured safety rules and edge-case reasoning for AV decision-making.
• ●The hybrid approach is technically significant because knowledge graphs provide verifiable, auditable logic — a property LLMs alone lack — which matters directly for regulatory certification of AV systems.
• ●As AV safety scrutiny intensifies globally, explainable, graph-backed reasoning architectures may become a prerequisite for commercial deployment approval rather than an academic curiosity.
• ●🔎 Read More →
• What matters: Pairing LLMs with knowledge graphs could give AV safety systems the auditability regulators demand without sacrificing the flexibility of learned models.

🚀 Product, Hardware & Model Launches

🧠 Generalist’s GEN-1 Foundation Model Expands to Multiple Robot End Effectors with a Single Base Policy

• ●Generalist has updated GEN-1 so a single foundation model can learn sensorimotor policies across different robot hands and end effectors — eliminating the need to retrain from scratch per hardware variant.
• ●This is architecturally meaningful: most manipulation policies today are tightly coupled to a specific gripper geometry, so a hardware-agnostic base model dramatically lowers the cost of deploying across robot fleets.
• ●The next test is whether GEN-1’s cross-effector generalization holds on novel, out-of-distribution hardware — that result will determine whether this is a genuine foundation model or a well-tuned multi-task policy.
• ●🔎 Read More →
• What matters: A single policy that transfers across end effectors is the manipulation equivalent of a universal adapter — it makes robot deployment economics fundamentally different.

🤖 AMD Launches Kria AI SoM and Ryzen AI Embedded X100 to Challenge NVIDIA in Real-Time Robot Control

• ●AMD unveiled the Ryzen AI Embedded X100 processor and Kria AI System-on-Module, both featuring unified memory architectures designed specifically for real-time robot control workloads.
• ●Unified memory is critical for robotics inference: it eliminates the latency penalty of shuttling data between CPU and GPU, which matters acutely in closed-loop control systems operating at high frequencies.
• ●AMD entering robotics silicon with dedicated hardware puts direct pressure on NVIDIA’s Jetson line — expect pricing and ecosystem competition to accelerate as both companies court robot OEMs.
• ●🔎 Read More →
• What matters: AMD’s robotics silicon push means the embedded compute market for robots is no longer NVIDIA’s to lose — competition is now real and hardware-level.

💰 Business, Startups & Investment

🧠 Hyundai Chairman Commits to Accelerating Physical AI Through NVIDIA and Google Partnerships

• ●Hyundai’s chairman has publicly pledged to fast-track the company’s physical AI strategy by deepening partnerships with both NVIDIA and Google — a dual-stack bet on compute and software platforms.
• ●Partnering with NVIDIA (likely for robotics inference and simulation) and Google (likely for foundation models and cloud AI) simultaneously suggests Hyundai is building a layered AI stack rather than betting on a single vendor.
• ●With Boston Dynamics already in its portfolio, Hyundai is positioning as one of the few OEMs with both the hardware assets and the executive mandate to compete in physical AI at scale.
• ●🔎 Read More →
• What matters: Hyundai’s dual-vendor AI commitment signals that legacy automakers are no longer observers in physical AI — they’re building infrastructure to compete directly.

🤖 Holiday Robotics Raises $105M for FRIDAY, a Wheeled Humanoid with Hot-Swappable Batteries

• ●Holiday Robotics closed a $105M round for its FRIDAY robot, which combines a wheeled mobile base with a dexterous upper body and hot-swappable batteries — a design optimized for continuous industrial operation.
• ●Hot-swappable batteries are an underappreciated engineering choice: they eliminate downtime for recharging, which is a critical uptime constraint in logistics and manufacturing deployments where robots must run multi-shift schedules.
• ●FRIDAY’s wheel-plus-manipulation architecture positions it against both AMR players and bipedal humanoid startups — the $105M gives Holiday runway to prove which market it can actually win.
• ●🔎 Read More →
• What matters: Hot-swappable batteries and a wheeled base aren’t flashy, but they’re the kind of operational pragmatism that separates robots that demo well from robots that actually work in warehouses.

⚡Capital concentration::ATOMS’ $1.7B launch round and Holiday’s $105M signal that robotics funding is shifting from seed experiments to deployment-scale bets.

⚡Hardware-agnostic AI::Generalist’s GEN-1 cross-effector policy and VLM-IE3D’s sensor-free 3D reasoning both point toward AI that adapts to hardware, not the other way around.

⚡Silicon competition::AMD’s Kria SoM entry into robotics compute breaks NVIDIA’s near-monopoly on embedded AI inference — pricing pressure and ecosystem fragmentation will follow.

⚡OEM commitment::Hyundai’s dual NVIDIA-Google partnership, backed by Boston Dynamics ownership, makes it the most credibly positioned legacy manufacturer in physical AI.

⚡The open question::With ATOMS unfunded-to-$1.7B overnight and wheeled humanoids raising $105M pre-revenue, the real test is whether 2026’s robotics capital wave produces deployments at scale by 2028 — or a correction.