Eventually, all that will matter to Tesla and others is changing civilization with Teslabot.
All that will matter is the race to improve and use AGI and humanoid bot. This will be the driving purpose and generator of value for civilization and human economy.
In chess – A forced move is one that a player must make to respond to a threat posed by an opponent. If a player’s king is in check, for instance, that player must make a move that gets their king out of check.
* Useful humanoid bots is easier than many believe
Evidence: Three Stanford students making useful robots using off the shelf hardware and part time work
Evidence: 50 repetitions of a task using co-training teleoperation to reach 95% success rate
Evidence: Rapid success with many competitors: Figure AI, Toyota Research etc…
* Value of human level or above human level humanoid bot is basically some multiple of current human labor === $100+ Trillion per year
Humanoid bot is 100X the value of robotaxi or robotruck
Companies have already spent $100+ billion trying to get robotaxi. Dozens of companies and many losing $2-3 billion per year for multiple years.
How can Tesla improve and speedup the Dojo AI training?
How can the AI make Teslabots more useful?
Critical milestones
* make Teslabot useful and valuable for some jobs or tasks in Tesla Gigafactories
* Have 5000 Teslabots and 20,000 factory workers do the work and match the production of 25,000 factory workers
* Move the 5,000 workers from each of 4 factories (Shanghai, Fremont, Berlin, Austin) to open another factory or enable three shifts to become four shifts at the same factory
* Have 10,000 or 20,000 Teslabots replace half of the human workers in a factory and increase production versus an all human factory
* Double overall Tesla production using Teslabots with human workers
* Increase the growth rate of Tesla production and speed the opening of new shifts, new factories while increasing margin
In 2024, Elon and Tesla must take the push for Teslabot and Dojo to world class levels. There is a 2025 Kuwait project to use 1 gigawatt of power for 700,000 Nvidia B-100 chips. Each B-100 will be about 5-10 times as powerful as the current Nvidia H-100. It costs $30,000 to $40,000 for one H-100. If a B-100 costs $50,000 each if ordered at the end of this year, then 700,000 B-100 would cost $35 billion.
NOTE – Nvidia 2023 revenue is about $27 billion.
The $35 billion is just the raw cost of chips. The full project energy, staff, facility and the chips could easily be $100 billion.
Tesla and Elon may only be spending $2-4 billion on the Dojo, FSD and Teslabot effort even with the 100 Exaflop AI training system.
Elon said AI compute resources and systems are increasing at 10 times every 6 months which is 100 times every 12 months.
In 2026-2027, China could be at a trillion for compute and diverting 20% of all national electricity for improving and training AI.
Gartner – Worldwide end-user spending on public cloud services is forecast to grow 20.4% to total $678.8 billion in 2024, up from $563.6 billion in 2025.
The spending on compute and energy and supply chain will lead the return in profits. It is critical to get the profit and revenue generation to start funding the flywheel of AI and humanoid bot progress.
Elon and Tesla solved the FSD profitability flywheel while Waymo, Cruise and others make 0.1% of the money back in revenue.
Elon and Tesla can get trillion(s) in value by using Teslabot to boost the productivity and efficiency of car and megapack production.
The 2026-2027 beast of AI and humanoid bot race will need Tesla to devote all of its megapack, solar, Dojo capabilities to be competitive and match China at a trillion in compute and 100+ gigawatts of power.
IF it is clear that the competition is going to 2025 (Kuwait project) and 2026-2027 (National China project) scale then what is inevitable for Tesla in 2024 with new Teslabot factories, and actuator factories and Dojo buildout?
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.
IMO, lots of blue-collar work, such as plastering, will be done by robots. If the hardware is good enough, then it is a software issue. Software can be pretrained and packed as modules; each occupation has one module, and a robot can download it. Enough smart intelligence will make use of all the theory and YouTube videos to train for specific occupations.
I think that solution to increased robot numbers will be to tax robots. There will be less human workforce, but robot’s work will be taxed instead human’s and that will give money to retirement funds,…
I agree Tesla bot has great potential. It depends how it will progress. Tesla has more than enough good financials to do it and other advantages, expertise as well.
Basically if the hardware is good, reliable enough, then it is a software problem. Software can be upgraded.
So human robots in home, they will cook top notch meal, clean, take care of some housework, cut lawn,… That is just at home,… Most likely affordable to the ordinary Joe.
I think that demand will be extremely high. People buy all sort of crap and useless toys, but robots can actually prove great.
I agree, at least for the foreseeable future, companies (big & small) will be taxed on how many robots they employ. But this isn’t just for retirement accounts, it’s for everyone. UBI will become law, I don’t love the government being in control of my finances, but I don’t see a choice.
Bipedal locomotion might be the though part as it take about a year for a human to walk.while grasping things and moving them is a lot faster..
I agree: the impact of the Teslabot on humanity could exceed all his other projects put together. By a large factor. Even the colonization of Mars may be accelerated more by the Teslabot than by SpaceX. (Though both are helpful).
These bots will enable automated factories that will make every product they build cheaper. And the bots will build bots. So the bots themselves become cheaper. A virtuous cycle.
The price of products ultimately becomes mostly a function of the price of raw materials and energy. But mining bots will make raw materials so cheap, that the cost may end up being mostly a function of energy costs. The bots may make it cheaper to build and install solor panels, windmills, fission reactors, and possibly fusion reactors. So even the cost of energy may drop as a result of this.
The hype is for a reason, a tsunami is building, and few seem to realize the water is receding.
People who pay attention know that the work on AI and humanoid robots are finally converging. Having embodied AI smart as a person, changes everything.
The growth will be HUGE.
Global humanoid robot numbers:
2023 – ~300
2024 – ~5,000
2025 – ~ 150,000
2026 – ~ 500,000
2027 – ~ 3,000,000
2028 – ~20,000,000
2029 – ~ 100,000,000
2030 – ~300,000,000
In 20 years we will have more humanoid robots, then people on Earth.
Lets keep in mind this is just humanoid robots, so it doesn’t count self driving vehicles, and the biggest job in the US…is people driving.
Also doesn’t account for all the office jobs where AI doesn’t need a physical body. Accounting, Secretary, Lawyer, Tax preparer, Telemarketer, Customer Service Reps, and many other data entry jobs.
I think humanoid bots will be mostly leased at rates that are extremely profitable to both OEM and user in developed world factories and commerce. A human worker equivalent bot could replace 4 workers in a 24/7 operation because it can cover 3 shifts plus replacement for vacation and illness. Since workers wages and benefits cost more than $50k/year even at the low end, that’s $200k per year in earnings. The bot would pay for itself in a few months and keep pulling in revenue for years. The employer/user would get a low cost package covering everything – and huge savings over prior labor costs.
There are vast numbers of jobs within the skill range Teslabot has already shown. For example, Car company warehousing and distribution has a function called “unitizing” – which is basically just putting parts in boxes for retail and repair shops. Most parts built are shipped raw and used in assembly. Tesla bot in videos has already shown the dexterity and speed for the job and could stand by a belt or packaging machine and put parts in boxes all day long – standing on an inductive charging pad. It could walk to different locations for different but similar tasks. These jobs now are cost $100k+ per worker and one Teslabot could replace 4 workers.
Replace low-skilled agricultural workers that need to be brought it and you could have the ‘political’ momentum for mass countryside adoption…
I think the rough estimate is $5k – $10k to support an agricultural worker (room/board on site) for a warm crop season at 50 – 80hrs/ wk – might be a tough number to crack… depends on total cost of Bot ownership per year… lease? rent-to-own?…
Do they work well in mud?
There are already agricultural robots but they tend to be attachment on vehicles. I suspect that will continue to be the way to go in certain industries like agriculture but replacing one arm with 2 arms and conventional blind idiot programming with neural nets will make them far more flexible and practical.
Humanoids and other mobile embodied AI may have some places though the early years will be all about getting training data. However, fixed AI robots will be a major competitor to Optimus and Google’s Mobile Aloha for industrial applications.