Jeff Brown calls it the 70X AI Agent. The pitch says Tesla’s Full Self-Driving software is a general-purpose AI agent, that the same neural network powers Teslas on the road and Optimus robots on the factory floor, and that one hidden supplier makes the optical hardware both depend on. The number in the headline is 70 times your money.
The first thing to understand about that number is where it comes from. Brown attributes the 70X projection to Elon Musk. Musk has said that Tesla could become a $100 trillion company. Brown’s materials cite Cathie Wood’s ARK Invest and the futurist Peter Diamandis making the same case. Brown’s role is to identify the supplier that benefits if any part of that thesis plays out. The 70X is the ceiling Musk paints, and Brown’s pick is the picks-and-shovels company underneath that ceiling. That distinction matters. A guru claiming 70X returns on a stock pick is one thing. A guru pointing at someone else’s 70X projection and saying “here is the supplier that benefits if he’s right” is a different structure. The risk in the second structure is whether the underlying projection is credible.
The $100 trillion Tesla claim in context
Global GDP is roughly $105 trillion. A $100 trillion Tesla would be a company worth nearly the entire annual economic output of the planet.
ARK Invest’s 2026 open-source model projects $2,800 per share by 2029 in the bull case, implying a market cap near $9 trillion at the upper bound. The $100 trillion figure comes from a longer-horizon projection that assumes Tesla captures the majority of a global autonomous ride-hailing market, replaces a substantial portion of human labor with Optimus robots, and dominates energy storage simultaneously. Those are three separate businesses, each requiring independent execution. The energy business is real and growing, but it is the smallest of the three in the projection.
ARK’s track record on specific Tesla price targets is mixed. They projected $4,000 per share by 2025 in a 2018 model. Tesla peaked at $414 post-split-adjusted. The methodology is transparent and the assumptions are published. The directional thesis on ride-hailing has held up better than the specific numbers.
An operator’s read: $100 trillion is a ceiling projection — the figure that emerges when every assumption in the bull case breaks the right way at once. The framework behind it is worth taking seriously as a map of what the Tesla thesis is actually claiming. The number itself marks the upper bound of that map.
Brown’s track record on Tesla specifically
Brown recommended Tesla in December 2018 when Bloomberg was running bankruptcy coverage and short interest sat near multi-year highs. He published a buy recommendation based on the thesis that Tesla was an AI and robotics company misclassified as a car company.
The stock returned roughly 1,510% from entry through its peak. Past performance does not guarantee future results.
That call was a category error identification, the same method Brown applied to Nvidia in 2016 and Bitcoin in 2015. The method: find the asset the market has miscategorized, identify what it actually is, and buy before the re-rating.
The Tesla call is the most documented of Brown’s big three. The timing is verifiable, the thesis is published, and the price action is public record. You can pull the chart and see the entry point and the peak.
What the Tesla call establishes is that Brown understands the difference between what a company does today and what its technology could become. That is the exact skill the 70X AI Agent thesis requires.
Is FSD actually a general-purpose AI agent
Brown’s framing is that Tesla’s Full Self-Driving system is a general-purpose physical AI agent that happens to have been trained on driving first. The neural network processes visual input, makes decisions, and executes physical actions. Those three capabilities, perception, decision, and actuation, are the components of autonomous agency.
The technical claim has support. Tesla’s FSD runs on a vision-only architecture with eight cameras providing 360-degree input. The same neural network that navigates a vehicle through traffic is, in principle, transferable to a robot navigating a physical environment. Tesla has demonstrated Optimus prototypes using the same vision stack.
The open question in the AI research community is how far a vision-only system can reach toward full autonomy — what the industry calls Level 5. Waymo uses lidar and radar as redundancy layers. Tesla’s bet is that vision alone, trained on enough data, is sufficient. FSD operates at Level 2, which requires a human driver ready to take control. Tesla has logged over 700,000 paid robotaxi miles across four cities, while Waymo has a multi-year head start and a fully driverless commercial service.
Brown’s thesis requires the architecture to be general-purpose and the demand for physical AI components to scale — Level 5 autonomy tomorrow is not a precondition. The robotaxi miles are real, the Optimus prototypes are real, and the question is timeline rather than direction.
The hidden supplier structure
This is where the 70X AI Agent pitch does something Brown has done before. In 2016, the hidden supplier was Nvidia itself, a $30 stock the market had misclassified as a gaming company. The investment was the supplier, not the brand.
The current pitch applies the same logic one layer down. Instead of buying Tesla, you buy the company that makes the optical hardware both Teslas and Optimus robots depend on. Brown’s report details the specific company, the ticker, and the reasoning for why the market has underpriced the component relative to the adoption curve.
Third-party pick-tracking sites have published candidate names spanning sensor manufacturers and camera module suppliers. Brown has not publicly named the pick outside the paid report. The structure, a picks-and-shovels supplier to a platform transition, is consistent with Brown’s documented method.
The variable in a supplier thesis is how long the platform company finds it cheaper to buy than to build. Musk has a documented preference for in-house manufacturing. Tesla builds its own seats, its own glass, its own battery cells. If the optical component is critical enough, Tesla may eventually bring it in-house. The supplier thesis has a window that depends on how long the build-versus-buy math favors buying the component from Brown’s pick.
The Thesis in Layers
Brown’s track record on category-error identification is documented and verifiable. The thesis that FSD is a general-purpose physical AI agent is technically defensible. The picks-and-shovels logic of buying the supplier instead of the brand is a proven structure in technology investing.
The 70X number is Musk’s ceiling projection, and the $100 trillion Tesla figure assumes three independent businesses all hitting scale simultaneously. The vision-only autonomy question is where the industry is still working — whether vision alone reaches full Level 5 without sensor redundancy is the variable that shapes the timeline. And the supplier thesis carries a build-versus-buy window, because Musk’s companies have a history of bringing critical components in-house once the volume justifies it.
Jeff Brown’s 70X AI Agent is a thesis built on a documented method applied to a technology transition that is actively unfolding. The number is someone else’s ceiling. The investment is the supplier underneath. The supplier’s trajectory depends on how the physical AI transition maps onto the timeline Musk projects, and how long Tesla keeps buying the component instead of building it.