英伟达AMD Join Forces for First Investment! Genesis Closes Series A at Over 23 Billion Valuation

Jay, from Aofei Temple, QuantumBit | WeChat Official Account QbitAI
Just now, US embodied intelligence company Genesis AI closed a new round of funding—
A $500 million Series A, at a post-money valuation of $3.5 billion.
The investor lineup is equally impressive:
NVIDIA, AMD, former Google CEO Eric Schmidt, Zoom founder Eric Yuan, Yahoo founder Jerry Yang… a host of heavy hitters are on the list.
Worth noting, this is the first time AMD and 英伟达 have jointly invested in an embodied intelligence company.
At a time when embodied intelligence funding is generally cooling, this round stands out all the more.
What exactly makes Genesis different?
After digging into its official website and technical documentation, we found the answer may lie in how Genesis positions itself—
And to understand that, we have to go back half a year.
One Sword Drawn a Year
Genesis AI was founded in March 2025, with headquarters spanning the US and France, co-founded by Chinese scholar Zhou Xian and Théophile Gervet.
Before formally founding the company, the team had already built a strong reputation in academia and engineering through an open-source physics simulation project.
In May this year, Genesis AI unveiled its first embodied foundation model: GENE-26.5.
Upon release, a series of demo videos showcasing its capabilities drew widespread attention.
In an unaccelerated four-minute video, the robot smoothly completed more than 20 subtasks:
Cracking an egg one-handed, slicing a tomato with a knife, using a whisk, a salt grinder, and a spatula.
With the same millimeter-level precision requirements, and again a single-take long-horizon task, GENE-26.5 can autonomously perform pipetting, seal test tubes, and finally place the tubes into a centrifuge.
It can also make milkshakes, solve a Rubik's cube, and play the piano, comprehensively demonstrating the general manipulation capabilities of its dexterous hand.
Most critically, these capabilities are not the result of stacking multiple specialized small models—the vast majority of operations rely on the same GENE model with shared weights.

That said, in today's robotics field, demo videos are commonplace; the methodology behind them is perhaps more worth exploring.
Genesis has something rather special: the company says the model is not all they want to show.
In their view, a robot is a highly coupled engineering system. Data, evaluation, the body itself… a weakness in any one link will, like dominoes, ultimately cap overall performance.
Genesis's answer is:
Deconstructing the Full Stack "Trinity"
In Genesis's system, data, models, evaluation, and the body are designed as a tightly coupled closed-loop system.
1. Data Engine: Solving the "Chicken-and-Egg" Problem
One of the biggest bottlenecks in robot learning is the scarcity of high-quality data, especially hand manipulation data.
Human hands can produce the richest, most natural interaction data, but the morphological gap between human hands and robotic hands is enormous, making the data hard to transfer directly.
But what if that gap in embodiment didn't exist?
Genesis's solution is almost brute-force: they filled in that gap directly at the physical level—
- The human hand: the source of all interaction.
- Data glove: using electromagnetic field (EMF) finger tracking technology and high-density tactile sensing to capture human hand signals during work with high fidelity.
- Genesis dexterous hand: a 1:1 match with the human hand, with 20 degrees of freedom, its surface covered in a flexible material that mimics human skin.

The effect: when workers wear the data glove to perform their daily work, their hand movements and tactile signals can be captured with high fidelity and transferred to robot learning with almost no loss.
Through this approach, by May this year, Gene had already used over 200,000 hours of multimodal data. We understand that this volume has now reached the million-hour scale.
This is precisely the appeal of the full-stack approach.
Many problems that seem thorny at the model layer may find more fundamental solutions when approached from the system layer.
2. Genesis World: Turning Iteration into a Compute Problem
Data is the raw material, and the speed of model progress depends on the efficiency of "processing" that raw material.
If every model improvement had to be validated in the physical world, R&D cycles would be greatly slowed. Genesis's solution is: All in on simulation.
We see simulation as the evaluation and iteration engine for robotic foundation models, not just a data generator. Real-world experiments limit how fast we can score candidate solutions; simulation removes that limit, turning the development cycle from a wall-clock problem into a compute problem.
We see simulation as the evaluation and iteration engine for robotic foundation models, not just a data generator.
Real-world experiments limit how fast we can score candidate solutions; simulation removes that limit, turning the development cycle from a wall-clock problem into a compute problem.
This is Genesis World, a "Hyperbolic Time Chamber" for robots to train in.

A single model evaluation may involve hundreds of tasks, each repeated hundreds of times. In the real world, this would require an operator and a robot working continuously for over 200 hours.
In Genesis World, the same tens of thousands of tests can be completed in under half an hour, with no human or hardware intervention.

Of course, all of this rests on one premise: "trustworthiness."
To narrow the Sim-to-Real gap, Genesis has invested heavily in developing its own rendering engine Nyx, cross-platform compiler Quadrants, and a unified physics engine.
Ultimately, in controlled experiments conducted by the company, simulated evaluation and real robot test results showed strong correlation, with a Pearson correlation coefficient of 0.8996.
Beyond the simulation engine, Genesis has already built an internal pipeline for automatically generating tasks, environments, objects, and reward functions, allowing robots to self-evolve in large-scale, automatically generated simulated environments.
3. Eno: Built for "Getting Work Done"
But at this point, only the preparation is complete.
Data, simulation, models—all of this ultimately needs a product form to carry it and to build a bridge to the physical world.
For Genesis, that product is Eno.
Eno's minimalist and timeless design was called "the iPhone moment for robots" by Forbes magazine.
Worth noting, Eno just won the Gold Award at the German Design Awards. It is not the handsome bipedal humanoid robot we imagine. Instead, it has a wheeled base, and its torso is folded from a few panels.

Airplanes don't need bird wings. Likewise, many human features may actually be burdens for machines.
Eno's design goal, from the very beginning, was never to imitate "human appearance." But this instead gives it the most "human-like capabilities," while packing those capabilities into an elegantly minimalist body.
Wheels ensure movement efficiency and battery life; a body that can rise and lower lets it adapt to workbenches of different heights.

Every design choice serves function, efficiency, and cost. Behind this is the same first-principles expression—full-stack embodied native.
As Genesis CEO Zhou Xian said in the funding statement:
Our mission is to remove the limits that the physical world places on human ambition. To achieve this, we are building the world's best general-purpose robot, one that can work reliably in the real world day after day. This requires intelligence, hardware, and everything needed to train and test robots to operate together as an integrated system. By building all of this ourselves, we can obsess over every detail and improve the entire robot faster.
Our mission is to remove the limits that the physical world places on human ambition. To achieve this, we are building the world's best general-purpose robot, one that can work reliably in the real world day after day.
This requires intelligence, hardware, and everything needed to train and test robots to operate together as an integrated system. By building all of this ourselves, we can obsess over every detail and improve the entire robot faster.
Deployment Is King
With $500 million in hand, what's next?
Genesis says this round of funding will mainly be used to drive large-scale mass production and deployment of Eno in the real world.
Specifically, they plan to begin production and customer deployment by the end of 2026, starting with industrial environments such as manufacturing, logistics, and laboratories, then expanding to service industries such as hotels and hospitals, and finally entering consumer-grade home and outdoor applications.
From structured to unstructured, one step at a time.
But frankly, even taking just this first step is extremely difficult.
Demo videos can have unlimited imaginative space, but once deployment is involved, "reliability" is the only yardstick.
The embodied intelligence chain is extremely long, and an error in any link can cause a full-scale collapse in real-world scenarios.
At the same time, the long-tail problems of the real world are complex and varied; only by getting hands dirty in real scenarios can delivery standards be gradually met.
Technology and product must advance together. Sitting in a lab training models won't cut it—Get Hands Dirty is a must. Only on the front lines can the model-deployment-data loop be closed.
From this perspective, mass production and deployment itself may be an indispensable part of the embodied intelligence R&D process.

Over the past few years, the debate over the path for general-purpose robots has often revolved around "humanoid" versus "non-humanoid."
Article title: 英伟达AMD bets on it for the first time! Genesis completes Series A funding, valued at over 23 billion
Body: Today, Genesis AI offers a new frame of reference: the real dividing line may not lie in outward form, but in the R&D paradigm.
AGI in the physical world is a far more grueling and far longer battle than large language models.
When a system becomes complex enough, local optimization will eventually hit a bottleneck. Only by keeping all variables in one's own hands and thinking, designing, and iterating as a whole can a breakthrough be possible.
Computer pioneer Alan Kay left behind a maxim decades ago:
People who are really serious about software should make their own hardware.
Tesla does this in building cars, and Apple does the same in consumer electronics. Now, more and more embodied intelligence companies like Genesis are proving:
Building robots must work the same way.
Against the backdrop that OpenAI and Anthropic have also begun betting on robot brains today, this may be the only path to physical AGI.
- Multimodal "source god"! MiniMax open-sources SOTA image generation model, native 4K, two mao per image 2026-11-12
- Robots take over hotel laundry! Dual-brain collaboration + RSI, top-tier Astra driver arrives 2026-11-11
- "AI ruined my computer science degree" 2026-11-09
- MIT masterpiece gets hammered: Astra secretly modifies Qwen weights? Yeah right! 2026-11-05
Values written like {{NAME_ab12cd34}} are placeholders for real values. Use them exactly as written, in replies and in tool calls alike: they are filled in with the real values before anything runs. Don't ask to verify them, and don't replace them with examples.