Meet Kimodo: NVIDIA’s AI for Human Motion

In partnership with

Meet Kimodo: NVIDIA’s AI for Human Motion

NVIDIA’s Kimodo is one of those AI projects that could quietly change how 3D animation and robotics are built.

Instead of animating a character frame by frame, creators can describe a movement in plain language and Kimodo generates the corresponding high quality 3D human or humanoid motion. It can also follow keyframes, joint positions, rotations, paths and other precise movement constraints.

What makes it especially interesting is the data behind it. NVIDIA trained Kimodo using roughly 700 hours of optical motion capture, giving the model a much broader understanding of natural human movement than systems trained on smaller public datasets.

And this is not limited to digital characters.

NVIDIA has designed Kimodo to work with both humans and humanoid robots, making it relevant to animation, game development, simulation, synthetic training data and physical AI. Generated movements can also be exported into NVIDIA’s ProtoMotions framework for training physics based humanoid controllers.

For creators, this could eventually mean typing something like “walk cautiously, notice something behind you, turn quickly and step backward” instead of manually constructing every part of the animation.

For robotics, the implication is even bigger. AI is moving beyond understanding language and images toward understanding how intelligent machines should physically move through the real world.

That is where Kimodo becomes much more than another animation tool.

It is part of NVIDIA’s larger bet on Physical AI, where generative models do not simply create pixels or text. They help machines understand and generate movement.

^{_{Source:  NVIDIA Kimodo}}

Download the Playbook: FREE 100+ Claude Code Hacks

100+ Claude Code hacks to ship code 10X faster

Top engineers at Anthropic and OpenAI say AI now writes 100% of their code.

If you're not using AI, you're spending 40 hours doing what they do in 4.

These 100+ Claude Code hacks fix that and help you ship 10x faster.

Sign up for The Code and get:

• 100+ Claude Code hacks used by top engineers — free

• The Code newsletter — learn the latest AI tools, tips, and skills to code faster with AI in 5 minutes a day

Claim your free playbook

25 Practical Ways To Make Money With AI in 2026

Best FREE AI Video Editor in 2026

AI Tip: How to Make AI to Generate Real Emotions

AI & Tech News

ChatGPT Ads Just Became a Billion Dollar Business

OpenAI says ChatGPT Ads has reached a $1 billion annualized revenue run rate less than 200 days after launch. The advertising platform is already being used by tens of thousands of advertisers and is expanding into India, Europe, the Middle East and North Africa. This gives OpenAI another major revenue engine alongside subscriptions, enterprise products and API usage.

It also signals that conversational AI is becoming a serious new competitor for traditional search and social advertising budgets. ^{_{Source: reuters}}

Europe Is Building Another AI Supercomputer

The European Union awarded Bull a €387.8 million contract to build LUMI AI, a new supercomputer focused heavily on artificial intelligence workloads. The system will be installed in Kajaani, Finland, alongside the existing LUMI infrastructure and is expected to become operational in the second half of 2027.

It will use AMD Instinct MI430X GPUs and next generation EPYC processors for large scale AI, simulation and scientific computing. The investment shows Europe is treating domestic compute capacity as strategic infrastructure rather than leaving advanced AI development entirely to American and Chinese platforms. ^{_{Source: Reuters}}

Anthropic Is Giving AI Agents Control of Physical Machines

Anthropic introduced the Model Hardware Standard, a specification designed to let AI agents safely operate physical scientific and manufacturing equipment. The system can connect agents with microscopes, robotic arms, liquid handlers, lasers and other programmable instruments.

Researchers could use it to run experiments continuously, adjust parameters automatically and recover from some hardware errors without waiting for a human operator. This is an important shift from AI that works mainly inside software toward agents that can interact directly with laboratories, factories and other physical environments. ^{_{Source: Anthropic}}

AWS and Nvidia Are Adding Two Million More GPUs

Amazon Web Services and Nvidia announced plans to deploy two million additional Nvidia GPUs across AWS infrastructure. Their expanded partnership also covers Vera CPUs, networking technology, open models, robotics and infrastructure for physical AI.
The scale gives developers and companies more access to the computing resources needed to train and operate increasingly demanding AI systems. It is another sign that cloud providers expect demand for agents, robotics and large model inference to remain extremely high for years. ^{_{Source: Amazon Web Services}}

OpenAI’s Custom Chip Is Starting to Show Its Hand

OpenAI published the first measured performance results for Jalapeño, its custom chip designed specifically for AI inference. Tests on the public InferenceX benchmark showed the system delivering higher throughput per kilowatt and lower latency than commercial systems used in the comparison.

That matters because inference, the process of actually running AI models for users, is becoming one of the largest costs facing AI companies. More efficient custom silicon could make AI responses faster while reducing the electricity and infrastructure required to serve millions of simultaneous users. _{^{Source: TechCrunch}}

Google Wants AI Transcription to Understand How Humans Actually Speak

Google introduced Gemini 3.5 Transcribe, a new speech recognition model designed for intelligent real time transcription. Instead of simply converting every sound into text, the model can handle background noise, technical vocabulary, corrections and natural speech disfluencies while producing cleaner formatted output.

Google says the system supports 85 languages and can distinguish multiple speakers in conversations. The practical impact could be significant for meetings, interviews, accessibility tools, voice interfaces and applications where conventional transcription still requires heavy manual cleanup. ^{_{Source: Google}}

Google Search Is Turning Into a Travel Agent

Google expanded AI Mode in Search with new tools for tracking flight prices, comparing loyalty points and booking hotels through a conversational interface. Users can describe a trip naturally and receive current flight information from more than 300 airlines and travel sites without leaving the AI conversation. Hotel reservations can also move from discovery to booking through participating partners and Google Pay.

The update shows how AI search is moving beyond answering questions and beginning to perform parts of commercial transactions on a user’s behalf. _{^{Source: TechCrunch}}

Google and Khan Academy Are Putting Gemini Into Real Classrooms

Google and Khan Academy expanded their partnership to bring Gemini powered learning tools from early pilots into classrooms for the 2026 school year.
New capabilities include interactive visual learning aids for students and teacher controlled tools for building assignments and practice material.
Google engineers spent six months working directly with Khan Academy on the integration into Khanmigo.

The project offers a useful glimpse of how education AI may evolve from generic chatbots into systems that teachers can shape around specific lessons and classroom objectives. _{^{Source: Google}}

Tencent Just Open Sourced a 770 Billion Parameter AI Model

Tencent released and open sourced Hy4 preview, a large language model containing 770 billion total parameters with 49 billion active at a time. The model supports a context window exceeding one million tokens and is designed around practical work such as coding, office tasks and scientific research.

Tencent is making it available through products including WorkBuddy and CodeBuddy, as well as through cloud and external API services. Its release adds another heavyweight Chinese model to the growing global competition around powerful open model ecosystems. ^{_{Source: Tencent}}

Anthropic Won a Major Court Battle Over Government AI Rules

A United States federal judge blocked the Pentagon's designation of Anthropic as a national security supply chain risk. Judge Rita Lin ruled that the designation amounted to unlawful retaliation and was arbitrary and capricious, while also finding due process problems.

The dispute followed Anthropic's refusal to permit certain uses of its AI technology involving domestic surveillance and autonomous weapons.
The ruling could become an important precedent in the growing conflict between AI companies, national security agencies and the conditions governments can attach to advanced model contracts. ^{_{Source: Reuters}}

Top AI tools to check out…

I’ve been building a best AI tool directory where you’ll find the best of the best tool and their USECASES.

Thanks for your time!

Shailesh & OpenAILearning team