SpaceX Has Another $150 Million Per Month Deal

SpaceX is adding another $900 million of revenue in 2026 and $1.8 billion of revenue per year with another AI rental.

This is with Reflection. It adds from July 2026 to 2029. This means $900 million more in 2026 and a runrate of $2.35 billion per month in Q4. $28 billion per year in 2027 with Anthropic, Google and Reflection deals. SpaceX building more data center means more AI rental revenue. Cursor at $4 billion per year revenue and doubling every 5-6 months. New Databricks deal can add another $500M to $1 Billion per year in 2027 and more beyond.

SpaceX can keep building new ai data center as fast they can. As fast as they can is about. 0.3 to 0.5 GW per quarter now and scaling to 1 GW per quarter by mid 2027. Rubin chips for the new data center is rentable at $100 B per year per gigawatt while Blackwell chips is rentable at $50 B per year per gigawatt.

$28 billion per year in 2027 with Anthropic, Google and Reflection deals. SpaceX building more data center means more AI rental revenue. Cursor at $4 billion per year revenue and doubling every 5-6 months. New Databricks deal can add another $500M to $1 Billion per year in 2027 and more beyond. $34-40 billion per year runrate in Q4 for AI only. $25billion runrate for launch and Starlink in Q4. operating profit $20 billion per year.

Broadcom is worth about $2 trillion and they made $68 billion per year last year and are at about $88 billion per year now.

Screenshot

SpaceX is making the $150 million per month or $1.8 billion per year renting about 18000 Blackwell chips. They installed 100,000 GPUs in 30 to 60 days which is down form 90-120 days per 100,000 in the first 200,000 they installed. I has been 17 days since the Google $11 billion per year rental was announced. SpaceX can average a $2 billion per year rental twice a month. while renting about half what they are adding

2 thoughts on “SpaceX Has Another $150 Million Per Month Deal”

  1. Cursor was growing really fast, but I doubt it will continue to do so for so many years. They say the bottleneck was compute and the largest expense so that will be solved.
    Revenue is nice, but there will be expenses running all that!

  2. These vast sums like $28billion a year are not coming from consumers or mid tier and lower companies, they are coming from big tech companies and investment and hedge funds. Those funds need to pay back the investors and generate money above that to keep funding these huge compute farms. What is the payback period for this compute farms? 5 years? Maybe longer? The heaviest use of AI I have seen is to write code, emails and schedule manufacturing – do those use cases justify those sums?

Comments are closed.