Who Will Win the Humanoid Race? The Battle Begins
- Sep 13, 2024
- 6 min read
The Short Version
Treat humanoids as R&D research platforms, not production-ready workers.
Add humanoids to your line only if you supply bleeding-edge tech to research markets, like Trossen Robotics.
Run actual use-case scenario research now; expect real ROI is still a few years out at minimum.
Watch the dozen-or-so contenders battle it out and note each company's commercialization approach.
Compare feature sets: servo configs, motor speed and strength, actuator types, and sensors.
Decide your bet: in-house novel servos and lifting strength versus off-the-shelf servos plus vision and hand-eye coordination.
Armchair quarterback the race and predict who thrives or dies before committing budget.
Who this is for
Robotics researchers and R&D labs
Nerdy entrepreneurs and founders
Product and hardware engineers
Investors watching the humanoid race
Small companies buying research platforms
Who will win the humanoid race? Nobody knows yet — and that's the point. The starting gun went off in 2024, and now a dozen or so companies are all sprinting at the same space, each betting on a different approach. I don't think we're anywhere near real ROI for humanoids in production — that's still a few years out, minimum.
But for actual use-case research? We're 100% here. Here's how I think this race plays out, and why I'll be watching with a bucket of popcorn.
Are humanoids actually ready — or is it still too early?
In the past few months, Trossen Robotics has begun discussing whether or not to add humanoids to our store. I have been kind of sh*t talking humanoids for a few years, saying that “It is WAY too early! These companies are way too early, and it needs another five years… blah, blah, blah.” But what do I know? I just work here.
I do believe it will be another few years at a minimum before we are anywhere near any realistic ROI for companies attempting to deploy humanoids in a production environment. BUT I think we are 100% here in terms of doing actual use-case scenario research.
Projects like Amazon testing Agility’s Digit in their warehouse are actually early stage R&D testing masquerading as “deployment” for flashy PR article headlines. Since Trossen Robotics supplies bleeding edge tech to research markets, the argument can be made that the technology is ready for small companies like ours to start adding humanoids to our line of research platforms. But you won’t be seeing humanoids doing actual useful work for a while yet.
I saw this Freethink article comparing all the top humanoids currently in development and/or close to commercialization, and it looks like a pretty decent line up of contenders for the great humanoid race of the 2020’s grand AI tech boom.
What happens when a dozen companies attack the same space?
Every new technology goes through a phase where a dozen or so companies all show up at once, out of stealth, to attack the same space — and then a bloody battle begins. Each one tries out a different approach to commercialization. It is a fight to gain market share and viability for their specific way of approaching the market. It is a race to generate enough sales to develop either self-sustained growth or more investment.
I always find this phase SUPER interesting because you get to watch a bunch of different ideas battle it out in the open market. The playbooks split roughly like this:
Product quality first: some companies obsess over building something great.
Marketing blitz: others eschew making something perfect and try to blitz their way into dominance, then worry about improving the product later.
Industrial focus: aim for just a few sales with deep-pocket customers, build partnerships, and run quick development cycles tailored to specific niches.
Commercial focus: make a more generalized product with broader appeal, hoping to hit lower price points through economies of scale.
Why do small design choices make or break a robot?
In Robotics, we see different kinds of feature sets being tried out. Different servo configurations, different choices in motor speed and strength, different actuator types, different sensors… There are many, many ways to make the same type of robot, and the right or wrong choice on something specific can make or break a product very fast.
Take humanoids. One company may focus on lifting strength while another focuses on vision and hand-eye coordination. Here's how those two bets look side by side:
Company A | Company B | |
Priority | Lifting strength | Vision + hand-eye coordination |
Servos | Researches and designs its own in-house, even patenting novel designs | Buys off-the-shelf servos |
Trade-off | Spends heavily on actuators | Settles for a lighter payload, puts robonerd engineers on camera optimization and arm/hand kinematics |
Both put their very different approaches into the market, and we get to sit back and see what happens. Who was right, and who screwed up?
If the market values humanoids picking up large boxes and carrying them around with a low need for high-precision placement, then Company A wins. If it turns out the market really wanted the humanoid to stand over a desk and sort things, then Company B made the right call — they won the majority of the market, got more sales, and got more investment, while Company A had a fire sale on their patents.
Can anyone predict who wins the humanoid race?
No one has a crystal ball, and it’s incredibly hard to predict what the market will do or what it wants. The reason is that the market itself rarely knows what it wants.
So the market tests new technology in all kinds of different ways until something clicks and an ROI is realized. That begins to fund the new technology to grow in those particular directions, and it influences the evolution of products. It is in this Darwinian capitalist way that companies and their products either thrive or die.
So, let the humanoid games begin
When I see a new lineup of humanoids like the ones in that Freethink article, what I hear is a starting gun going off. A fascinating race of a dozen companies all trying to beat each other to the finish line, where the survivors get to share in the new pot of gold when a market emerges.
It always starts with a large crowd of competitors and a bunch of new products hitting the market all at once. Then they drop off one by one until only two or three make it to the end to battle each other for the next decade.
As a nerdy entrepreneur, this is my version of the NFL. I like to grab a bucket of popcorn and armchair quarterback as we watch the game unfold and try to predict who we think will win and who will lose and why. So, let the humanoid games begin.
*Curious what bleeding-edge research platforms look like today? See what Trossen Robotics builds, and dig into the Trossen SDK docs to get started.*
Frequently Asked Questions
Who will win the humanoid race?
No one has a crystal ball. It starts with a dozen or so companies, then they drop off one by one until only two or three survive to battle for the next decade.
When did the humanoid race start?
The starting gun went off in 2024, kicking off the great humanoid race of the 2020's grand AI tech boom, and now the battle begins.
Are humanoids ready for real production work?
Not yet. Expect another few years at a minimum before any realistic ROI in a production environment. Projects like Amazon testing Agility's Digit are early-stage R&D masquerading as deployment for flashy PR headlines.
Why add humanoids to a store like Trossen Robotics now?
Because they supply bleeding-edge tech to research markets, the argument can be made the technology is ready for small companies to add them as research platforms, even if humanoids won't do useful work for a while.
Why do so many humanoid companies emerge at once?
It's a standard early phase of new tech adoption: a dozen or so companies come out of stealth to attack the same space, each testing different approaches to commercialization in a bloody battle for market share.
What design choices separate humanoid contenders?
Companies try different servo configurations, motor speed and strength, actuator types, and sensors. One may bet on lifting strength while another focuses on vision and hand-eye coordination.
How does the market decide the winner?
The market itself rarely knows what it wants, so it tests the tech until something clicks and an ROI is realized. In this Darwinian capitalist way, companies and products either thrive or die.