One of the great benefits of working in high-tech and living in Asia is the proximity to both the tech-forward consumer markets and the factories where so much of the world’s technology hardware is manufactured.
Living in Tokyo, taking a flight to Shanghai, Beijing, or Taipei was as normal and convenient to me as a flight from New York to Kansas City.
Occasionally, I’d even make day trips if I had an important meeting. Those were early mornings and very long days, but it is possible.
Travel in-region was almost a weekly routine for me. And I attended a large number of major conferences and trade shows every year.
They’re a fantastic way to keep the pulse of industries and see more local technology players and manufacturers than I’d ever see at a U.S. or European trade show.
I was reminded of this when reviewing some of the developments from the 2024 World AI Conference in Shanghai.
Given China’s stated national plan to become the world’s AI superpower by 2030, this event received a lot of attention.
And this year, there was one particular kind of Manifested AI that was most widely on display – humanoid robots.
Source: Xinhua News
Do Not Touch the Display
There were 18 different humanoid robots on display at the event.
It was an impressive sight, given that there are only a handful of contenders in the western markets that have been making progress, like Tesla, Figure, Boston Dynamics, Agility Robotics, 1X Technologies, and Sanctuary AI.
But the strong appearance didn’t tell the whole story. Those 18 robots all had a couple of things in common.
The first is that they were all in very early prototype stages with limited mobility. Any walking demonstrations were confined to vendor booths, and the robots were tethered from above to ensure that they didn’t fall over.
The second thing that they had in common is that “they” all tended to look like the humanoid robot shown below, which was also at the event.
Source: Xinhua News
Tesla choose to keep Optimus in a secure display case, safe from poking and prodding. Some chose to interpret that as a sign that Tesla “isn’t ready with the technology to demo it.” That would be the wrong interpretation.
After all, Tesla has nothing to prove and an invaluable amount of trade secrets to lose. Tesla is so far ahead of the competition, and it will be manufacturing Optimus in-house, it doesn’t need to demonstrate its technology at a trade show.
The last video update we had from Tesla was just five weeks ago, and Optimus is doing just fine.
It’s an interesting competitive dynamic in the industry right now.
Piecing together the hardware for a humanoid robot is relatively easy compared to developing the autonomous software necessary to create a functional robot with utility.
Many China-based companies have built a prototype as a sales tool, to show that they could be a good manufacturing partner for a technology company. After all, much of the tech industry relies on outsourced manufacturing.
The same will be true in robotics. Many robotics companies will focus on the really hard part – the hardware design and development of the artificial intelligence (AI) software – and leave the manufacturing to others.
Tesla has not only developed its own custom hardware but also the autonomous software built upon the software foundation that comes from its full self-driving software (FSD) for its electric vehicles.
The media rarely speaks of this, and yet FSD is Tesla’s single most valuable asset. And it’s also why Tesla will dominate the humanoid robotics industry for years to come.
That’s not to say that others won’t find their niche.
A Nascent Industry
Agility Robotics has been making progress by focusing on supply chain and logistics applications.
Below we’ll see a highly sped-up video clip of Agility’s Digit performing a single function of lifting bins and placing them onto a conveyor… over and over again.
One thing that should stand out to us in the clip above is that we see very few people. The shelves that deliver the bins to Digit are also on robotic platforms. And the logistics facility is highly automated.
While this is a relatively simple task to solve – compared to Tesla’s undertaking to make a general-purpose intelligent humanoid robot – it is still valuable, nonetheless.
There are currently around 600,000-700,000 unfilled supply chain and logistics jobs in the U.S. alone. And that number is expected to grow to 1.9 million by 2033 if the current trend continues.
The reality is that with each year that passes, it becomes more difficult to find labor willing to perform such backbreaking, repetitive, and monotonous work like what is shown above. Robotics powered by varying forms of artificial intelligence is the solution. It’s the answer to solving the labor shortage crisis.
This is not to say that China is far behind the U.S. and Europe in robotics technology. There are some bright spots.
It’s Not About the Fancy Hardware
Shown below is a great example of a stationary robot that is developed with a focus on fine motor skills. This is Astribot S1, developed by Stardust Intelligence.
We can see the natural dexterity of the arms and digits performing a complex task like folding a piece of clothing. This is certainly impressive…
But it’s not about the hardware.
The simple fact is that the entire industry is chasing Tesla’s Optimus right now. Some of the more advanced players can perform a subset of what Optimus is capable of, but none are close to Tesla’s autonomous technology and general-purpose functionality.
Yesterday, Tesla CEO Elon Musk provided an update on its plans for Optimus, mainly that Tesla will use Optimus for internal use this year and next and then open up production for external customers in 2026.
Critics, of course, suggested that Tesla has fallen behind or misled investors as a previously stated target was to allow for sales of Optimus robots by the end of next year.
My read on the announcement is different.
Tesla is already testing Optimus in Tesla factory and office settings. The company already has a functional humanoid robot with some autonomy, and they are focused on training for a wide range of tasks that will be useful on a factory floor, in an office, and even in a home.
Production is always small in the beginning. Tesla, which manufactures its own products, will benefit greatly from absorbing the initial production of Optimus. This will be a competitive advantage for the company.
And once Telsa has mastered production and can manufacture at scale, it will open up for external sales. Volume production will lead to a price point below $20,000.
At that price point, Optimus will not only be accessible to small businesses, it will be within range of many households that already spend similar amounts for help throughout an average year.
Having a home humanoid robot will become far more of a statement than having an iPhone or a Tesla EV ever has been.
Tesla has already mastered fully autonomous software built using neural network technology, as well as manufacturing advanced battery-powered electronics at massive scale.
Just because we can’t buy it yet, doesn’t mean that it doesn’t exist…