An AI agent processed SpaceX’s 226MB S-1 filing in 12 minutes, purchased live market data using USDC on Base, and produced an investment committee memo after just 6 paid API calls. The full run cost $1.87. According to the source material, this was not a demo workflow but a record of actual paid API usage, and it did not require an API key.
An end-to-end research workflow ran without a human analyst team
The key shift described in the report is not just faster document reading. The AI agent selected its own research path, paid for the datasets it needed, and assembled a finished memo that included bull and bear cases, valuation work, a risk matrix, and a decision card. The final stance in that memo was “hold and watch.” Work of that scope would normally take a team of analysts days rather than minutes.
The article also says the peer comparison table was built in real time. It used Jintel’s GraphQL endpoint to pull batch fundamentals for five comparable companies, with that data request priced at $0.10. The full analysis package was generated through agentic.market and used Base-based x402 payments for paid calls.
The memo’s view on SpaceX mixed dominant assets with heavy liabilities
On the bullish side, the memo pointed to three areas: launch, Starlink, and vertical AI integration. The figures cited in the filing say SpaceX has accounted for more than 80% of global mass to orbit since 2023, while Falcon mission success has exceeded 99%. Starlink was described as having 10.3 million paying users across 164 countries, with 2025 revenue of $11.4 billion and segment-adjusted EBITDA of $7.2 billion.
On the bearish side, the memo focused on capital intensity, debt, and governance. It argued that SpaceX’s real long-term debt load was about $42 billion, above the $29.1 billion shown on the capitalization cover page. That total included a roughly $20 billion bridge loan due in September 2027. For the AI segment, the filing figures cited in the article show $3.2 billion in 2025 revenue against an operating loss of $6.4 billion and capital expenditures of $12.7 billion.
Cheap paid data access may change how Wall Street research gets done
The broader point of the article was less about whether SpaceX is a buy and more about how research can now be produced. The AI agent was not used as a summarization tool alone. It accessed paid data sources directly, linked information across documents and datasets, and delivered a memo in a format close to institutional research output. The article contrasted that cost with a Bloomberg terminal, which it said runs about $24,000 per year, versus $1.87 for this chain-based workflow.
Based on the details provided, the process already covered filing review, data acquisition, financial comparisons, and risk framing. The source does not claim this replaces Wall Street analysts outright. What it does show is that a large block of standardized investment research work now has a low-cost, automated execution path that can operate in real time.