🤖 NVIDIA Jetson Thor: The Ultimate Robot Brain

ADVANCEMENTS IN ROBOTICS

1. 🧠 NVIDIA Jetson AGX thor - The new “Robot Brain”

In a single leap, NVIDIA just supercharged robotics. The Jetson AGX Thor developer kit priced at $3,499 packs the ground-breaking Blackwell GPU, delivering up to 2,070 FP4 teraflops of AI compute and a massive 128 GB of memory, all within a 130 W envelope. That’s a jaw-dropping 7.5× performance boost and 3.5× better energy efficiency over the aging Jetson Orin platform.

This isn't just raw power; it’s a smart, edge‑ready brain for robots. Imagine autonomous machines that host multiple generative AI models right on board: humanoid assistants navigating unpredictable environments, precision surgical bots responding in milliseconds, and logistics automatons making split‑second decisions without relying on the cloud.

Big names like Amazon Robotics, Boston Dynamics, Caterpillar, Meta, and Figure are already diving in, while industry giants John Deere and OpenAI are evaluating the platform. NVIDIA’s software stack - Isaac, Metropolis, Holoscan, and GR00T makes it a one‑stop shop for physical AI creators.

In short: the Jetson Thor is less a chip and more like an on‑robot intelligence engine unlocking a future where every corner of the physical world can think, decide, and act.

2. 📦 Charming: PAT (by Keren Sasson) is a delivery robot that feels more like your friendly neighborhood postman than a cold machine. With the capacity to haul multiple packages in one go and a smiling “face,” it lightens sidewalk traffic while bringing a touch of warmth to urban logistics. Think of it as city living getting a friendly handshake with efficiency built in.

3. 🐕 Protective: Hong Kong is rolling out robot dogs armed with insecticide sprayers to battle the rising tide of Chikungunya fever especially in tricky, hard‑to‑reach zones. These agile quadrupeds help ease the burden on frontline workers while delivering targeted mosquito control with precision and speed. Could this be the future of smart, humane public health defense?

4. 🧬 Microscopic: Researchers just introduced “light‑driven lattice soft microrobots” (LSMR), tiny hydrogel machines that dance with laser guidance hopping, rotating, and squeezing through tight gaps with fluid grace. They’re nearly 30× faster than earlier designs and use a fraction of the energy, opening new doors in bionics and biomedical navigation. Imagine swarms of smart, gentle robots exploring the micro‑world inside us.

5. 🎓 Immersive: RealMan has unveiled a sprawling 3,000 m² Humanoid Robotics Data Training Center in Beijing, packed with 108 robots and ten lifelike environments from eldercare to smart catering. It’s a sandbox where robots learn real‑world skills, generating over a million high‑quality data points per year to power embodied AI. It’s where robots are not just built, they grow intelligent.

ROBOTICS ON SOCIAL MEDIA THIS WEEK

1. 🤖 Breakthrough: Boston Dynamics’ ATLAS gets a brain upgrade while Unitree unveils its next-gen flagship humanoid.

2. 🤖 Insane: A jaw-dropping demo shows robot hands mastering fine control, proof we’ve entered a whole new robotics era.

3. 🌐 Vantage: Inside SCO’s media hub in Tianjin where AI humanoid robots, drones, and diplomacy converge on the global stage

4. 😮 Hyper-Real: Peek beneath the skin of a lifelike humanoid robot that blurs the line between human and machine.

5. 🏓 Smash: A humanoid robot rallies 100+ table tennis shots with human-like speed and precision.

6. 🤖 Flip: Boston Dynamics pushes robot dogs to their limits with RL stunts to build stronger machines.

7. 👾 Playtime: Kids in China casually play with a humanoid robot as if it’s one of them.

8. 🤖 Mentality: A humanoid robot shows off its kitchen skills with flawless, tireless precision.

SUNDAY SCHOOL

Is Google’s Nano Banana the future of editing?

SPECIAL

🧠 What bees teach us about smarter AI

Sometimes nature is the best computer scientist. A fascinating University of Sheffield study reveals that bees sharpen their vision not with bigger brains but with smart movement. Their flight patterns refine neural signals, letting them recognize visual patterns with astonishing accuracy. And the best part? This happens without massive computing power.

Researchers modeled this behavior in a digital bee brain and stumbled on a potent insight: AI and robotics could be radically more efficient by using motion not brute force to simplify perception tasks. Think robots that move to learn, not just process robots that use motion as a lens for seeing, sensing, and understanding the world not just GPUs.

Imagine drones that adjust their flight paths to “see” better, or sensor-driven machines that reposition themselves to interpret data more clearly. It's the difference between sending endless data to the cloud and using physical context to reduce complexity.

This isn’t sci-fi, it’s a call to rethink how we design AI-driven systems. Movement as perception aligns elegance with efficiency. Could the next leap in AI be inspired by something as small as a bee’s flight?

PROMPT OF THE WEEK

Prompt in more detailed on building a sensitivity analysis for valuation based on terminal growth rate.

I want to create a sensitivity analysis table for the valuation of a company using the Discounted Cash Flow (DCF) method, where the terminal value is calculated using the Gordon Growth Model.

Please include the following in your analysis:

Base case assumptions for:

▫️ Free Cash Flow (FCF) for the last forecasted year

▫️ Discount Rate (WACC)

▫️ Terminal Growth Rate

Generate a data table or matrix showing how changes in the terminal growth rate (e.g. from 1% to 5%) affect the intrinsic value per share. You can include 5–7 scenarios.

Clearly state the formula used for terminal value:

Terminal Value = (𝐹𝐶𝐹)𝑛 × (1+𝑔)
                   (𝑊𝐴𝐶𝐶 − 𝑔)

Include instructions or an example on how to calculate the equity value from enterprise value by subtracting debt and adding cash, and dividing by number of shares to get intrinsic value.

Optional: Include a second dimension of sensitivity based on WACC variations (e.g. 8% to 10%) for a 2D sensitivity table.

Please present the output in a well-labeled, professional table, and explain insights that can be derived from the table. The tone should be suitable for use in an investor presentation or valuation report.

If you find the contents of this email useful, subscribe now & share with your friends.

Your Opinion Matters!

Got more feedback or just want to get in touch? Reply to this email and we’ll get back to you.

Thanks for reading.

Until tomorrow!

Shinky & the Hanoomaan AI team