​

The future of humanoids

The future of technology is turning sci-fi into reality. For two years, ChatGPT and other AI apps have ruled the digital world, with OpenAI leading the charge.

Since 2022, though, competitors like DeepSeek and Perplexity have emerged—some as wrappers, others as serious threats. But software alone can’t go the distance. Chatbots, limited to internet data and token caps, hit a wall without physical interaction.

Enter robotics: the next frontier for AI. While Tesla’s self-driving cars already showcase real-world intelligence, the future demands more—daily human-robot collaboration. This week, three breakthroughs brought that future closer. Here’s what’s happening.

What’s happening at Figure.ai

Brett Adcock is running the most exciting startup at the moment. It’s not often when you see young companies work to bring hardware to life. Especially when it comes to industrial robotics, which General Motors first deployed in 1961. Yes, America has been working on the robotics problem for six decades.

Well in the world of digital intelligence, Figure today introduced Helix, a new generalist Vision-Language-Action (VLA) model. This new model unifies perception and language in the real-world. Watch this video to see how Figure’s robots can complete any tasks on command.

​

The combination of hardware, software and engineering gives Figure a compounding advantage. I’ve watched several demonstrations by the Figure since last year. Each one continues to show significant progress within a few months.

Now if you have been paying attention to Figure, you will know that they recently canceled their partnership with OpenAI. This was shortly after DeepSeek launched their open-source model.

Figure achieved a "major breakthrough" in end-to-end robotic A.I.. This model was fully developed internally and is better tailored for humanoid robotics.

Outsourcing critical components like A.I. models could compromise Figure’s performance and approach to vertically integrated manufacturing. This decision reflects Figure AI's broader strategy to gain full control over its technology stack, akin to Apple's ecosystem model, and to position itself as a leader in the robotics industry.

Palmer Luckey’s back with VR at Anduril

I’m in the Palmer Luckey fan club since he ventured out to build Anduril. You may recognize his name as the founder of Oculus VR, which sold to Facebook (Meta) in 2014 for $2 billion. Initially this was a big bet for Meta to become the leader in virtual reality and the metaverse. But it also gave Palmer the funding to go into defense tech.

Now this is not your typical upstart. Anduril Industries is a product-first company, as opposed to the typical cost-plus contractor the U.S. government usually deals with. In fact, Anduril’s aspirations are so big that they want to compete head-to-head with the top five defense primes (Lockheed, Raytheon, etc).

Last week, I watched Palmer’s four-hour interview with Shawn Ryan. It was the first time Palmer described in great detail the global threats America faces today. To sum it up, America has lost its ability to make great military technologies. And we rely too much on China for our supply chain.

Since Anduril is a defense contractor, you probably aren’t aware they developed their own A.I. called Lattice. This system is designed to be the new command & control center for the U.S. military. Instead of relying on foreign competitors, Anduril wants to integrate American hardware with American software. I’m all for it.

Now Anduril is working on a $20 billion contract for the U.S. Government’s Integrated Visual Augmentation System (IVAS). In short, this is an advanced augmented reality headset. Our soldiers will become superhuman with Anduril’s hardware and software.

Building super soldiers is a new form of robotics. Think of it like Tony Stark’s Iron Man suit but without the constraints. One soldier will be able to command & control thousands of robotics with one headset. It is the perfect use case for combining hardware and software.

​

I also recommend watching Anduril’s new recruiting video.

​

Tesla new A.I. (Grok 3)

If you drove to work today, chances are you passed a Tesla on the way. It has become one of the most popular cars in the world. If you invest or follow Tesla, then you might have seen Optimus, their Tesla bot last fall as well. While Optimus doesn’t have full autonomy yet, the robot is getting there.

Today, Tesla cars have full self-driving which gives them access to real-world A.I. Instead of using LiDAR like Waymo, Tesla has intelligence cameras making quick decisions in real-time, like a human.

While developing this technology, Elon realized that Tesla’s technology could also be applied to robotics. Which is why he announced Optimus at its A.I. Day in 2021. On the surface, this looked like a typical product announcement but this week a lot has changed for Optimus.

If you use X.com, then you might be familiar with xAI’s Grok. xAI is Elon’s A.I. startup that competes directly with OpenAI to build super intelligence. On Monday, the xAI team released version 3.0 for X users. As for metrics, it has surpassed every app in the marketplace.

According to the benchmarks, it is the best A.I. in the world when it comes to coding, science and math. It’s world class and Elon’s team built Grok’s supercomputer in record. It took 122 days to build the first 100k Nvidia GPU cluster. Then it took another 92 days to build the next 100k Nvidia GPU cluster.

In fact, Grok 3 is now the #1 app on iOS. I’ve been using the free beta version for a week. It’s very impressive and much better than DeepSeek and ChatGPT. It is also probably the least known A.I. tool and I use it everyday. The only caveat is that it will cost $40 per month. I recommend watching the Grok release video on X.

Since Elon owns Tesla, X.com and xAI, he’s also mentioned that he will integrate Grok into his other companies. This means it will be available to everyone who owns a Tesla car and/or Optimus Bot in the future.