Charly Mwangi

Dad, Husband, Partner at Eclipse, Co-founder at Matter

[Theophile Gervet @theo_gervet]

Simulation is on the critical path to general-purpose robots.

Reality is too slow to scale robot interactions. Simulation lets us scale at the speed of software: evaluation today, reinforcement learning tomorrow.

Bridging the Sim2Real gap requires owning the full stack: system identification, control, compiler, physics, assets, and rendering.

Proud of the cross-functional collab across our global team on Genesis World 1.0 - and excited to open source it to help accelerate progress across the field.

More details:
https://t.co/EFUFMihvEa

[Zhou Xian @zhou_xian_]

Announcing Genesis World 1.0, the second release in our full-stack robotics tech suite.

After we released GENE-26.5 three weeks ago, many people reached out asking what was next for simulation.
Here we are.

Genesis has always been about building the best intelligence for robotics. Over the past year, we have moved a long way across the full stack, but simulation remains one of the most critical infrastructure layers for accelerating what comes next.

We have been heads down pushing Genesis World to a new level of realism, performance, and trustworthiness. Fully open sourced:
https://t.co/wZOU7ScSLP

[Charly Mwangi @charlythuo]

@gs_ai_ is back after 3 weeks! This time introducing Genesis World 1.0, their hyper-realistic simulation platform, which is a key part of their full-stack suite.

[Genesis AI @gs_ai_]

We are back again :) After three weeks of quiet building.

Introducing Genesis World 1.0, our latest simulation platform, the second release in our full-stack suite. Open-sourced.

Robotics is still bottlenecked by the 1× speed of the physical world. Every model, checkpoint, and data recipe eventually needs to be tested on physical hardware, slowly, expensively, and with limited coverage.

One hour in reality can become 100 days in simulation. That is how robotics model iteration moves from a wall-clock bottleneck to a compute problem.

To make this work, simulation has to be both fast and trustworthy.

Over the past year, we rebuilt the entire stack: a GPU-accelerated cross-platform compiler, penetration-free multi-physics contact solvers, unified rigid and deformable physics, and a photo-realistic renderer purpose-built for physical AI applications.

We built Nyx, a high-performance path-traced rendering engine for robotics application.

Genesis World 1.0 achieves near realtime performance with our latest development for penetration-free IPC solver, supporting various types of deformables beyond rigid bodies. It supports contact-rich, dexterous manipulation simulation across different embodiments: unitree, sharpa, wuji, genesis hand and various types of grippers.

Under the hood is Quadrants, our effort in pushing forward cross-platform GPU-accelerated computation. Quadrants started as a fork of Taichi, and we rebuilt most of the critical parts for optimizing simulation workloads, giving 10x faster launch time and up to 4.6x runtime performance compared to the initial Genesis release.

Together, they bring us to an unprecedentedly low sim-to-real gap, enabling zero-shot real-to-sim model evaluation and much faster iteration of GENE.

All available today.
Genesis World 1.0: https://t.co/aknCM3eqws
Quadrants: https://t.co/uXqPNI4cb6
Nyx: https://t.co/R8j0djqGnV

[Genesis AI @gs_ai_]

We are back again :) After three weeks of quiet building.

Introducing Genesis World 1.0, our latest simulation platform, the second release in our full-stack suite. Open-sourced.

Robotics is still bottlenecked by the 1× speed of the physical world. Every model, checkpoint, and data recipe eventually needs to be tested on physical hardware, slowly, expensively, and with limited coverage.

One hour in reality can become 100 days in simulation. That is how robotics model iteration moves from a wall-clock bottleneck to a compute problem.

To make this work, simulation has to be both fast and trustworthy.

Over the past year, we rebuilt the entire stack: a GPU-accelerated cross-platform compiler, penetration-free multi-physics contact solvers, unified rigid and deformable physics, and a photo-realistic renderer purpose-built for physical AI applications.

We built Nyx, a high-performance path-traced rendering engine for robotics application.

Genesis World 1.0 achieves near realtime performance with our latest development for penetration-free IPC solver, supporting various types of deformables beyond rigid bodies. It supports contact-rich, dexterous manipulation simulation across different embodiments: unitree, sharpa, wuji, genesis hand and various types of grippers.

Under the hood is Quadrants, our effort in pushing forward cross-platform GPU-accelerated computation. Quadrants started as a fork of Taichi, and we rebuilt most of the critical parts for optimizing simulation workloads, giving 10x faster launch time and up to 4.6x runtime performance compared to the initial Genesis release.

Together, they bring us to an unprecedentedly low sim-to-real gap, enabling zero-shot real-to-sim model evaluation and much faster iteration of GENE.

All available today.
Genesis World 1.0: https://t.co/aknCM3eqws
Quadrants: https://t.co/uXqPNI4cb6
Nyx: https://t.co/R8j0djqGnV

[Charly Mwangi @charlythuo]

In the Philippines with @UnderSecE and Foxconn Chairman & CEO Young Liu, visiting the forward-deployed industrial base. Super bullish on Pax Silica!

[Seth Winterroth 🤖 @Sethwinterroth]

In 2015, we started @EclipseVentures to back the boldest engineers building at the intersection of bits and atoms.

In 2016, @andrewdfeldman gave us one of the clearest pitches on a long-term vision I have ever heard in my career.

"The GPU architecture is fundamentally limited for AI. We are going to build a wafer-scale chip to replace it. It will take a decade."

A decade later, here we are. IPO Day.

A few lessons I've learned from our decade as investors in Cerebras:

The most important company building happens at the earliest stages. The technical bet, the architectural choices, the key manufacturing partnerships, the first 50 people on the team — all locked in before product-market fit, before revenue, before the capital markets care. Taking the time to get these foundational elements right is what lets you move with velocity later.

Talent is a compounding asset. From day one, Cerebras was obsessed with hiring the very best people in the world. Watching Andrew and the team obsess over this was a real lesson. Seeing Pierre and Lior work with the founders to build that team taught me a lot about the role a board member should play early in a company's history. One of the most formative things I've witnessed in venture. They treated every hire like it mattered, because it did.

Reputations are forged in the hardest moments, not the best ones. The reason Andrew first pitched Eclipse tells you everything: his several decade relationship with Pierre Lamond. Pierre had been on his board at SeaMicro and stood with him through the toughest stretches of building that company. When Andrew started Cerebras, he was adamant Pierre would be involved. The reputation you build in hard moment is what people remember. This is what brings you the next deal, the next hire, the next round. This is how we try to operate at Eclipse: true partners to founders through the hardest stretches of the journey.

It doesn't matter how good your technology is if you don't land the deals. No one remembers the company with the best chip and no customers. The last 24 months at Cerebras have been a clinic in commercial execution. Andrew, the GTM team, and the board (Lior, @ericvishria, and @vassallo) have been on a tear. Thousands of hours on planes to all corners of the world. Late nights. Holidays away from families. Hard decisions made under real uncertainty. Never losing sight of the mission. I've never seen anything like it. Great companies need a differentiated mission and world-changing technical execution — but they also need a shit ton of grit and a refusal to quit. Cerebras has all of it.

Eclipse is proud to have been there from the seed with @andrewdfeldman and the team. Watching Lior and Pierre steward this company for the last 10 years has been a masterclass.

Congratulations to everyone involved on reaching this milestone!

[Under Secretary of State Jacob S. Helberg @UnderSecE]

As @VP Vance said, the global market for critical minerals is failing. Supply chains are brittle.

We're fixing that. Pax Silica integrates the entire supply chain from raw minerals to advanced chips. We’re building a marketplace of trusted partners that reflects real costs, not coercion.

[WUJI TECH @wuji_global]

Huge milestone from our collaborator @gs_ai_ for their GENE 26.5 release.

We are proud that the original Wuji hand has been used in the development of Genesis’s proprietary dexterous hand. Genesis created a version unique to their system. We helped manufacture the Genesis Hand 1.0 based on their 1:1 human proportions and back-drivable design principles to support their human-level physical manipulation capabilities.

Proud to take this step together — from digital intelligence into real-world physical interaction.

[Hardik Kabaria @hardikk13]

Every major AI shift started with a new kind of world model.

Language. Vision. Code.

The next one is physics.

@saucentoss and I published a paper today defining what a real foundation model for physics has to be and why it enables Continuous Physics Reasoning.

If AI is going to help build the physical world, the bar has to be much higher.

Link to full paper: https://t.co/koNrDxrQPU

[Charly Mwangi @charlythuo]

LLMs made reasoning continuously accessible for language.

What would the equivalent look like for the physical world?

@VinciPhysics calls it Continuous Physics Reasoning.

@VinciPhysics is building a foundation model for physics. A model that reasons directly over geometry, materials, and manufacturing-scale physics.

They are democratizing physics! Changing Simulation from a slow, expert-driven verification workflow into an always-on intelligence layer embedded directly into design and manufacturing.

Honored to be partnering with @hardikk13 and his team.

https://t.co/50aoAmn7aK

[Hardik Kabaria @hardikk13]

Every major AI shift started with a new kind of world model.

Language. Vision. Code.

The next one is physics.

@saucentoss and I published a paper today defining what a real foundation model for physics has to be and why it enables Continuous Physics Reasoning.

If AI is going to help build the physical world, the bar has to be much higher.

Link to full paper: https://t.co/koNrDxrQPU

[Eric Schmidt @ericschmidt]

This is a major advancement in robotic reasoning and control. See the video !

[Zhou Xian @zhou_xian_]

The past year has been the happiest period of my life.

Genesis is a place of miracles.
We set countless ambitious goals.
None of them seemed possible with existing technology.
One by one, we made them real.

The density of breakthroughs coming from such a small team, across so many directions at once, feels almost rare in human history.

What you see here is fully autonomous, 1x speed, run on the exact same model.

[Charly Mwangi @charlythuo]

Physical AI is entering a new era.

Super excited to see @gs_ai_ unveil GENE-26.5 - a major step toward general-purpose robotics and human-level capability.

What makes this team unique is that they are rebuilding the entire stack from the ground up:

- A robotics-native foundation model.
- A 1:1 robotic hand with human-like dexterity
- most scalable, capable and least intrusive data collection glove for motion, force, and touch.
- A high fidelity simulator that accelerates experimentation

At @EclipseVentures, we believe the biggest AI opportunity ahead is bringing intelligence into the physical world.

Proud to be partnering with the @gs_ai_ team - @zhou_xian_ & @theo_gervet - as they push the frontier of physical AI and embodied intelligence forward.

The future of robotics will not be narrow and task-specific. It will be general-purpose, adaptive, robust and scalable.

And this is only the beginning.

#PhysicalAI #Robotics #EmbodiedAI #AI #Automation

[Genesis AI @gs_ai_]

We are back. After one year of quiet building.

Introducing GENE-26.5, our first robotic brain that takes a major step toward human-level capability.

For years, robotics has struggled to learn from the world’s largest and valuable data source: Humans.

Solving it means rethinking the whole stack from the ground up:

- A robotics-native foundation model.
- A 1:1 human-like robotic hand.
- A noninvasive data collection glove for motion, force, and touch.
- A simulator that turns weeks of experiments into minutes.

GENE-26.5 is trained across language, vision, proprioception, tactile, and action. We designed a set of tasks to test how far we can go with this new paradigm.

Fully autonomous, 1x speed, one model, same weights. (Enjoy with sound on)

We are approaching the endgame for robotics.

And this is just a beginning.

[Genesis AI @gs_ai_]

4/

Rubik’s Cube solving has been a long-standing challenging benchmark for robotic manipulation. The task requires fine-grained control under the geometric and kinematic constraints imposed by the cube itself. Prior state-of-the-art is still the single-handed solver from OpenAI’ in 2019. For the first time, we can solve a Rubik's Cube using two hands together, powered by GENE.

[Genesis AI @gs_ai_]

3/

Carries laboratory experiment with mm-level precision and complex tool usage.

[Genesis AI @gs_ai_]

2/

GENE-26.5 cooks in an unsimplified, real-world setting with more than 20 subtasks.

[Genesis AI @gs_ai_]

We are back. After one year of quiet building.

Introducing GENE-26.5, our first robotic brain that takes a major step toward human-level capability.

For years, robotics has struggled to learn from the world’s largest and valuable data source: Humans.

Solving it means rethinking the whole stack from the ground up:

- A robotics-native foundation model.
- A 1:1 human-like robotic hand.
- A noninvasive data collection glove for motion, force, and touch.
- A simulator that turns weeks of experiments into minutes.

GENE-26.5 is trained across language, vision, proprioception, tactile, and action. We designed a set of tasks to test how far we can go with this new paradigm.

Fully autonomous, 1x speed, one model, same weights. (Enjoy with sound on)

We are approaching the endgame for robotics.

And this is just a beginning.

[Charly Mwangi @charlythuo]

Vibe-manufacturing is coming to US.
Just watched @makematterco demo MatterOS - the entire manufacturing workflow on one seamless platform:
• Pre-SOP: DFM, process planning, simulation, sourcing,…
• Post-SOP: MRP, shop floor orchestration, cell config, genealogy, quality, Andon,…

The kicker? ZERO purchased software. No Dassault. No SAP. No Rockwell.

Built end-to-end in-house, in 5 months, by a handful of ninjas (Ex-@Tesla, @Rivian , @Apple , @GoogleDeepMind with deep domain expertise. This is what happens when the team building it has actually run factories.

MatterOS is built to enable autonomous manufacturing. The Tesla FSD playbook applied to manufacturing - and it’s already working.

@adiprasad27

[Charly Mwangi @charlythuo]

Let’s talk about Vibe-Manufacturing…yes, it’s going to be a thing in the US.