AI/data-center regulation shock is hitting risk assets, but the larger claim is that SpaceX and Tesla are still on a path to dominate compute, transport, and eventually space-based manufacturing if approvals, power, and chip supply keep compounding. Texas and Arizona permitting, launch cadence, and political control are treated as immediate bottlenecks, while the longer arc is 2027-2030 scale-up in data centers, Cyber Cabs, FSD, and Optimus. The video’s bull case culminates in huge standalone and merged valuations for TSLA and SpaceX, plus a much larger combined vertical stack if chips, power, launches, and compute all integrate. The close is still a risk-off warning: the tape is a bloodbath, margin is dangerous, and cash matters.
Markets are being trashed as the AI trade gets smashed on fears the left will choke data-center and AI growth with regulation; the risk-control rule given is never go above 20% of margin, because cascading selling can hit heavily leveraged retail fast.
STRC is cited as the kind of name that can be sold in a heartbeat in stressed conditions, while the tape is framed as a bloodbath with only early signs of bottoming.
Texas permitting is tied to the Jan. 19 governor transition: Abbott is due out, the new governor is described as a Democrat, and Texas actions do not take effect before Jan. 19; the requested changes include no NDAs, union jobs, subsidized electricity, no 20-year tax holidays, and paying for electricity and water.
Abbott’s moratorium is said to last until mid-December, creating a window to lock in approvals for SpaceX before the political transition; of the 10 GW Texas wants approved, 70%+ is already permitted and supposedly hard to unwind once terms are agreed.
Arizona is next in the pipeline: SpaceX is having a county/city meeting Wednesday to develop about 1.5 miles / 774 acres for up to 6 GW, with drawings showing about a 3.5 GW layout for power, cooling, and data centers; the blue area is guessed to be a detention facility/prison already there.
State competition for data centers is described as brutal: if a jurisdiction asks for too much, you get nothing; Louisiana is still subsidizing and is also where SpaceX launch facilities are being built; the estimate is that 10% to 40% of data centers could be iced, with a rough 41% state-pause risk cited.
SpaceX is at $124, back to levels last seen on Aug. 7, 2026; TSLA bottomed around $345 and is moving up, but not in a V-shape.
Orbital compute is framed as the answer to impossible Earth constraints: it is said to be impossible to get enough solar energy, cooling, solar hardware, and rockets into space—yet Musk is confident it can be done; the model asks how many launches and satellites are needed per GW.
Musk is said to have tweeted he will pass on monthly terrestrial compute in late 2028; to beat 2 GW and then 4 GW on Earth, the estimate is about 100 launches for 200-ton payloads or 200 payloads for 4 GW, implying 7 flights per day versus a current 3-4 flights per day baseline.
The launch cadence math says a year of Falcon 9 is 160 launches, so the needed monthly pace would exceed current annual cadence; current approvals are said to total 140 launches across Florida and Texas, enough for about 1 GW, while Texas has a filing in for 456 launches and Florida has not yet filed for the increase.
A 2027 target is laid out as ~100 launches for 2 GW, ~220 launches for roughly 2027 scale, and ~300 launches if execution is great; 1,200 launches per year are said to be needed even after Trump’s push for 1,000 launches per year.
Starship v3 is the key catalyst: v3 engines are described at 270 tons of thrust each, with a next engine at 300 tons and a next rocket planned for 2027; if thrust rises from 100-120 tons to 270, payload scale jumps materially.
Cost is the moat: launch cost is described as falling from $100 million to under $1 million, possibly $250,000, roughly a 400x drop; that makes orbital compute economically plausible.
SpaceX’s revenue growth is described as 860% quarter over quarter, with profit for the first time and a run rate from $18.5 billion to $100-120 billion; valuation models use P/E around 100 or 30-50 depending on assumptions.
The compute thesis is that new data centers are already adding 3 GW of compute, three times the second-largest data-center category, and SpaceX has coherent compute with homogeneous chips unlike others.
The next-year allocation model says about 0.5 GW may go into space, while 17 GW stays on the ground split between 10 GW of fully done SpaceX data centers and 7 GW where Microsoft or Amazon provide chips and memory while SpaceX supplies the buildings, power, and cooling.
Microsoft and Amazon are said to be unable to finish data centers because of chip delays, so they would take chips and pay SpaceX to host them; idle chips are framed as a $2 billion-a-year short on $10 billion of chips sitting idle over five years, and dead chips are compared to a $500 billion dark-fiber hole.
The moat is speed: SpaceX data centers can be built in 66 days, while CoreWeave takes 2-2.5 years for one; recent events allegedly made compute scarcer over the last 48 hours, and the conservative estimate is that 20% of data centers will be iced.
Tesla Cyber Cabs are already in Texas and Florida: 1,400 unsupervised Cyber Cabs are said to be in those states, with 1,244 registrations as of yesterday and 150-200 in Florida; the tracker only covers 1% or less of miles, so higher tracked miles imply growth.
Cyber Cab scale targets are 2,000 this month, 10,000 in January, about 1,000 in Austin, 4,000 for the Austin metro area, and up to 40,000 across five or six cities; the suggested safe zone to avoid crushing competitors is about 2,000, definitely 1,000.
FSD has a little over 1.7 million paid users now, with 6 million targeted for 2027; existing vehicles are about 30% adopted and are expected to reach 80-90% in the US, while new vehicles are already at 75% and could go to 90-100%.
Cyber Cab economics are laid out as $50,000 a year per car in expensive U.S. taxi markets; 1 million robotaxis equals $50 billion, and a robot cyber cab can be produced for about $15k, making the one-time cost versus recurring revenue the core math.
Optimus is framed as a major industrial ramp: Fremont is described as built for 20,000 a month, initial scale may be 200,000 by end of next year, and the largest Texas facility could reach 10 million; some units go to Tesla factories first, then suppliers.
The home-use case for Optimus is mainly the AI5 chip as edge compute, described as about a pedawatt of compute, 20x to 50x a laptop or small server; hand use is just a bonus, while the real value is the local AI/edge stack.
The merger/valuation section says Marvel’s New Lum deal timing is causing ARB-related shorting and no premium; the broader market message is that AI earnings are still great, the AI trade is not slowing, and cascading liquidations are happening.
TSLA standalone is projected at $10 trillion by 2030, equal to 6.6x from here or about $2,530/share; SpaceX standalone is projected at $15 trillion by 2030, or 6.1x from a $122 baseline.
The merged entity is projected at $25 trillion with no synergy, implying 7.3x for SpaceX holders and 7.6x for Tesla holders; a three-for-one SpaceX-to-Tesla merger ratio is the preferred outcome, though four is said to be possible.
A bull case puts SpaceX at $15 trillion by end-2027, about $1,000/share; the longer-range view is 80 GW on Earth, possibly 100 GW, with 50 billion per GW and 30% margin producing over $1.3 trillion of profit and more than $4 trillion of revenue.
The extreme upside case runs to 200 GW and about $20 trillion of revenue, $10 trillion of profit, and a $300 trillion market cap under a big PE multiple; Peter Diamantis is cited as calling this the first $100 trillion company.
The thesis is that compute prices will stay high because Earth-side supply is constrained and demand is effectively infinite; if compute stays scarce, the answer is to move more of it to space.
The combined-company scenario says Earth and space deployment can both double if Tesla and SpaceX share solar, batteries, megablocks, megapods, turbines, and chips; the optionality across Earth and orbital compute is described as something nobody else has.
The far end of the model includes a solar-system buildout of Earth, Moon, and orbits, a billion satellites, moon development, and mass drivers on the Moon; it also ties to billion-scale bots, shared data, and future medical and materials breakthroughs.
Medical claims are aggressive: cancer is said to be cured in 3-5 years for many cancers, organs are expected to be produced in space via starfall bioreactors, and meniscus replacement for spinal and shoulder issues is already being discussed.
Space bioreactors are said to enable drugs, better fiber including optical fiber, and hundreds of tons of biological materials coming back down over time, because microgravity helps tissues grow without falling apart.
The final warning is explicit: be very careful with margin and always have cash on the sidelines, even if the long-term SpaceX/TSLA compute story remains intact.