NVIDIA’s New Edge AI Chip Brings Frontier Intelligence to Drones and Robots.
Robotics AI Heats Up: Generalist Hits $3B Valuation as NVIDIA's Jetson Orin Nano 2 Pushes Physical AI to the Edge.
Robotics funding rounds and edge AI hardware launches signal the next phase of enterprise automation.
$3B: Generalist's New Valuation
$600M: Total Series B Raised
78 TOPS: Jetson Orin Nano 2 AI Compute
1: Generalist's Meteoric Rise in Robotics AI:
Robotics startup Generalist has reached a $3 billion valuation after closing an extension to its Series B funding round, underscoring how fast capital is flowing into the race to build a general-purpose AI “brain” for robots.
According to people with knowledge of the funding, the new round was led by 8VC. The fresh capital totals nearly $200 million, based on a regulatory filing, and is structured as an extension of the $400 million Series B that Radical Ventures led back in June, when Generalist was valued at $2 billion.
With this extension, the round's total funding now stands at $600 million — and the company's valuation has grown by 50% in under three months. Generalist and 8VC did not respond to requests for comment on the deal.
Founded in 2024, Generalist was started by former Google DeepMind researchers Pete Florence and Andy Zeng, alongside former Boston Dynamics engineer Andrew Barry. The founding team's background — spanning frontier AI research and hands-on humanoid and legged robot engineering — has helped the company attract a notable investor list from day one, including 8VC, Radical Ventures, Nvidia, Union Square Ventures, Bezos Expeditions, and AI researcher Fei-Fei Li. Until recently, Generalist operated quietly and with very little public visibility, a contrast to the high-profile marketing pushes of some of its better-known rivals.
At the center of Generalist's pitch is an AI foundation model built to generalize across many different robot bodies and tasks, rather than being trained narrowly for one machine or one job. The company's newly released Gen 1.5 model is reported to let robots learn entirely new tasks from video demonstrations as short as 3 to 12 seconds — a dramatic reduction in the amount of task-specific training robots have traditionally required.
Generalist is currently working with a handful of customers, using their real-world feedback to fine-tune the model for specific industrial and commercial use cases, according to a source familiar with the company's roadmap.
2: A Crowded, Fast-Moving Race to Build the “Brain” for Robots:
Generalist is far from alone in chasing a general-purpose robotics foundation model, and the sums of capital pouring into the space illustrate just how quickly investor conviction is building.
Physical Intelligence, one of the most closely watched competitors in the space, is reportedly valued at roughly $11 billion. SoftBank-backed Skild AI carries an even higher reported valuation of $14 billion.
Meanwhile, Genesis AI was in talks as of last month to raise capital at a $3 billion valuation — putting it roughly on par with Generalist's new price tag. Taken together, these numbers show a field where multiple, well-funded startups are racing toward the same goal from different starting points and technical approaches.
The funding surge reflects a broader bet among investors that robotics may be approaching its own “ChatGPT moment” — the point at which robots can perform general tasks without being explicitly trained for each individual one, much like large language models can handle a huge range of text-based tasks without task-specific fine-tuning.
However, that comparison comes with an important caveat that some VCs are quick to point out: robots cannot be trained on the entirety of the internet's data the way LLMs can. Physical, real-world data is expensive and slow to collect, which means a truly general robotics model capable of handling any task, in any environment, may still be years away, even as capital continues to flow into the sector at a rapid pace.
3: NVIDIA's Jetson Orin Nano 2 Brings Frontier AI to the Edge:
As robotics AI models grow more capable, NVIDIA is racing to put that intelligence directly onto the hardware that powers robots and drones. The company has unveiled the Jetson Orin Nano 2, an entry-level edge robotics computer built for physical AI.
NVIDIA is positioning the new board as an entry-level option for developers who want generative AI models running directly on a machine, instead of inside a data center. The company's argument rests on a broader shift in how well small and medium-sized AI models now perform: NVIDIA says models of this size have reached the same accuracy levels that only the largest frontier models achieved just a year earlier.
That leap in efficiency is what allows compact, low-power edge hardware to interpret language and images and act on that information in real time — a capability that robots, delivery drones, inspection drones, and vision AI systems all depend on, without drawing much power.
NVIDIA's Deepu Talla, VP of Robotics and Edge AI, frames the new board as putting that leap in model efficiency within reach of a huge developer base, pairing the performance and power efficiency needed for real-time reasoning at the edge.
On the specs side, Jetson Orin Nano 2 carries 78 trillion operations per second of AI compute, 8GB of memory, and an eight-core Arm CPU. NVIDIA built the board to deliver a significant jump in AI and video-processing performance while keeping cost and power draw low.
The new board reaches twice the inference performance of the existing Jetson Orin Nano Super, a gain NVIDIA attributes to improved Tensor Cores and higher memory bandwidth, all packed inside the same compact form factor as its predecessor. Running in 15-watt mode, Jetson Orin Nano 2 uses 40% less power than the Orin Nano Super while matching its performance level — a meaningful efficiency gain for battery-powered robots and drones.
The board runs on NVIDIA's open software stack, alongside Jetson agent skills and the wider Jetson AI ecosystem. NVIDIA says Jetson Orin Nano 2 is built to run large language models and vision language models optimized for memory-efficient inference at the edge, naming its own Cosmos and Nemotron models as examples, alongside third-party models like Gemma 4 and Qwen 3.
4: Early Partners Test Real-World Physical AI Applications:
Cognex, Doosan Bobcat, and Matic sit among the first companies NVIDIA names as adopting and exploring Jetson Orin Nano 2. NVIDIA says more than three million developers already build on its robotics stack, and the company expects partners to bring edge AI into home robots, vision AI systems, delivery and inspection drones, carrier boards, hardware systems, and reference designs.
Wing, the drone delivery subsidiary of Alphabet, already runs Jetson Orin Nano Super and NVIDIA's software stack across its delivery drone fleet. The company plans to evaluate Jetson Orin Nano 2 to push further into real-time AI perception and reasoning, with the goal of making deliveries from local businesses to residential yards faster and safer.
Dinuka Abeywardena, Head of Perception at Wing, said drone delivery depends on AI that enables fast, reliable understanding of the real world, and that Wing is exploring the new board as a path toward more responsive, energy-efficient drones. Notably, Wing's existing fleet runs on the separate Jetson Orin Nano Super product, and there is currently no public timeline for a move from evaluation to production use of Jetson Orin Nano 2 on delivery flights.
“Home robots need to understand people, map spaces, and clean autonomously in dynamic environments,” says Matic Robots CEO Navneet Dalal.
Matic Robots, a consumer robotics company, is adopting Jetson Orin Nano 2 for its home cleaning robots. NVIDIA says the board will let Matic add conversational AI, gesture detection, precision mapping, and semantic understanding of the home, layered on top of its existing autonomous cleaning behavior.
Beyond these early adopters, NVIDIA named a large roster of Jetson hardware partners building around the new board. AAEON, ADLINK, Advantech, and Aetina sit among the manufacturers building carrier boards and hardware systems for it.
A further group — Antmicro, Aptiv, Auvidea, and AVerMedia — is working on customized AI software and reference designs, alongside Chuanglebo, Connect Tech, ForeCR, and JWIPC. Rounding out the list, NVIDIA named Neurealm, Plink, Realtimes, RidgeRun, RS, Seeed Studio, Tauro Tech, Twowin, TZTEK, and YUAN as partners working to help customers reach the market faster.
“Frontier intelligence has reached the edge,” says NVIDIA's Deepu Talla — last year's data-center-only models now run in real time on entry-level Jetson hardware.
5: Why This Matters for Enterprise AI Adoption:
Taken together, these two developments point to the same underlying trend: AI is moving out of pure software and into the physical world, and the infrastructure to support it — both capital and compute — is arriving faster than expected. Billion-dollar robotics rounds like Generalist's, alongside rivals Physical Intelligence, Skild AI, and Genesis AI, show investor conviction that general-purpose physical AI is coming.
At the same time, NVIDIA's edge hardware shows that the compute needed to run frontier-level intelligence is becoming small, cheap, and power-efficient enough for everyday deployment — not just for robotics labs, but for drones, vision systems, and home devices already reaching real customers.
For enterprises, the lesson isn't that every business needs a robotics strategy — it's that AI capability is becoming more accessible, more real-time, and more embedded in day-to-day operations across every industry, from logistics and manufacturing to home services and customer-facing operations. The pace of investment and hardware innovation in physical AI is a strong signal of where enterprise AI adoption is headed next: out of the browser tab and into real-world workflows.
Physical AI Is Moving Fast — Enterprise AI Doesn't Have to Be Complicated.
From billion-dollar robotics rounds to AI chips small enough to fit inside a drone, the message is the same: intelligence is moving out of the lab and into everyday operations. Enterprises don't need a robotics lab or an in-house AI research team to keep pace — they need a platform that turns AI strategy into working systems.
Otherworlds AI's Agent+ Business AI Platform gives teams automated, Google Opal-powered workflows starting at $297/month — no robotics PhD required. For organizations with more specific needs, Otherworlds AI also builds custom enterprise AI solutions tailored to your operations.
Learn more at otherworldsai.com
