Micron (MU) delivered one of the most dramatic AI-driven turnarounds in recent memory. In calendar 2024 / early fiscal 2025, quarterly revenue sat in the $5–8B range while the stock traded mostly between $80–$150. By mid-2026, quarterly revenue had exploded past $15B (hitting $23.86B in one quarter) and the stock surged above $1,000 and the company went to $1.15 trillion up fro November 2024 $110 billion.
SpaceX (SPCX) is in a remarkably similar inflection point right now.
Micron moved from $6B per quarter to $25 billion per quarter over 2 years.
SpaceX was at $4.6B in Q1 of 2026. Q4 the SpaceX revenue is
$2.2B per month from Anthropic and Google. $0.7B per month from Cursor. Grok Revenue $0.1B per month. This $9B per quarter. Cursor can overperform and Grok can and another deal could add $3B in the quarter.
$6-7B per quarter from Starlink and DTC and launch. $15B per quarter
Revenue per GW Multipliers — The Real Upside Story
This is where SpaceX has more leverage than Micron ever did.
If SpaceXAI rented out the remainder of Colossus 2 (345K Nvidia B300 chips at the Google price of $11 billion per year for 110k B200 chips) this would be $34 billion per year. 550K B300 chips is $55B/year.
In Q4 2026 without much more rental SpaceX would be at a $60 billion per year runrate that would be similar to Broadcom.
However, building another 440K chips of data center over 12 months could rent out the rest of Colossus 2 while moving to new Rubin chips. The Satellite revenue and launch revenue is also still growing. Mid 2027, $100 Billion per year runrate. AI rental revenue is at 70-80% margins.
At $200 billion per year of revenue, SpaceX is at Nvidia revenue and profit levels. This should happen in 2027 or mid 2028.
SpaceX has 20% allocation of Rubin chips.
Rubin chips (next-gen Nvidia), Rubin chips have 10X the inference efficiency of B300 chips. This should be 2-3X in rental. If 2X that would be $22 billion per year per 100k R100 chips. 550K R100 Chips would be $110B/year to $160B/year. Renting out 550K Rubin chips next year is possible. Adding 110-200k per quarter is possible. Adding 100K chips per quarter is $22 billion per year per quarter of new revenue.
Feynman (after Rubin in 2028) 4-6X from B300. $220B/year to $500B/year
Cursor composer 3 and beyond, Grok 5 and Grok 6 Code. This could get to the higher Jensen quote of $300b-400B per year from 1 GW for Rubin Chips. +3x on top.
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.
I still have unanswered skepticism:
1. Aside from Anthropic and Google, who else has the money to rent compute from XAI? Pretty sure OpenAI would rather gouge their eyes out than work with Elon on anything. Microsoft and Amazon don’t seem to want anything beyond Azure and AWS. France/Europe are actively legislating against American tech, so Mistral seems out.
2. The only one I can think of is Meta/Facebook. Right now, they’re in such disarray that I’m not sure Zuck will to blow yet another wad of cash with no assured payback. I could be wrong.
3. AI is very cool. It will be a gigantic money machine. But adoption, outside of coding and chatbots, might take a while. The day of big companies willy-nilly throwing unlimited stacks of cash (by means of token usage) on AI usage is being curtailed. Upper management and board of directors want to see a return on their AI usage.
Of course, I could be wrong…