The subpoena confrontation comes to an end: SpaceX (SPCX.US) was forced to appear in court over AI safety, and space capital fanaticism now has a regulatory rift

The Zhitong Finance App learned that after several weeks of subpoena confrontation, Musk's SpaceX (SPCX.US) finally agreed to attend a New York City Council Technical Committee hearing this Friday to be questioned about artificial intelligence security risks. This belated testimony brought SpaceX's AI sector into the spotlight of global regulation, and also revealed the space giant's situation with a market capitalization of over 2 trillion US dollars between capital frenzy and compliance pressure.

From the “sole absentee” to being forced to appear in court

In mid-September, New York City Assembly Speaker Julie Mening issued voluntary invitations to testify to five of the world's largest AI companies. Anthropic, OpenAI, Google, and Meta confirmed their attendance, but SpaceX's AI department did not respond. On September 28, Meining officially issued a subpoena to SpaceX. This is the first time she has used the city council's subpoena power.

The hearing on October 5 lasted nearly ten hours. Representatives of the four major AI companies were intensively questioned by lawmakers after being sworn in, but SpaceX did not attend. Meining said on the spot that the company was in a state of “direct violation of the subpoena.” The city council later informed SpaceX that it plans to request the state Supreme Court to enforce the subpoena. Two days later, SpaceX voluntarily requested a meeting and finally agreed to attend the hearing this Friday.

This hearing did not “go through the motions.” The City Council is considering 10 proposed AI regulatory legislation. Core provisions include: all AI systems sold or deployed in New York City must be equipped with a human-operated “kill switch”; violators are fined $25,000 each; mandatory third-party verification; and the nation's first AI whistleblower bonus system — whistleblowers can get a share of the fines. The city council is also considering a proposal to create private litigation rights for individuals harmed by AI.

The deep background of this investigation is the long-term stagnation of AI legislation at the US federal level. The National Assembly has held numerous AI hearings over the years, yet it has never passed a substantial AI security bill. Meining's judgment on this was unequivocal: “In view of the importance of the matter, these companies have an obligation to come to Parliament and take an oath to answer questions.”

New York City isn't the only actor, but its legislative ambitions are clearly greater. Unlike the EU AI Act, which focuses on risk classification, the New York City proposal places more emphasis on enforcement mechanisms — the termination switch to resolve “what to do if something goes wrong”, third party verification to resolve “who will check before listing”, and whistleblower bonuses to resolve “who will expose internal issues that regulators cannot see.” The combination of these three forms a closed loop from prevention to detection to accountability.

The intersection of capital feast and compliance reef

SpaceX is currently in the midst of an intense tug-of-war between capital narratives and regulatory pressure. On the same day that news of the hearing broke, Space Capital released a report predicting that the amount of investment in the global space economy is expected to break the annual mark of 100 billion US dollars for the first time in 2026, driven by SpaceX's record IPOs, defense spending, and space infrastructure needs.

This forecast comes from investment agency Space Capital's latest analysis of third-quarter spending. In addition to quantifying the scale of investment, the report also indicates that the number of space companies that have reached a “unicorn” valuation is reaching a record high.

“We all thought 2021 was an insanely hot market,” Space Capital CEO Chad Anderson said, referring to the SPAC deal frenzy at the time. “But compared to the current situation, it was just a fleeting flash.”

SpaceX landed on the NASDAQ on June 12, with an issue price of 135 US dollars and raised about 75 billion US dollars. The market capitalization surpassed 2 trillion US dollars on the first day of listing, and then climbed further to about 2.2 trillion US dollars.

But the other side of capital inflows is the accumulation of risk. At the same time, Anderson warned that the business models of some space companies are based on uncertain variables, such as excessive reliance on the launch capability of the SpaceX Starship. “This is a major risk.” The same logic applies to SpaceX itself — the regulatory risk of its AI division is becoming an underpriced variable in the valuation.

SpaceX has clearly disclosed in its prospectus that xAI faces “numerous regulatory risks” involving the EU AI Act, the California Frontier AI Act, and New York State related legislation. The regulatory storm that Grok triggered in early 2026 is even more direct: the UK Communications Authority and the European Commission have launched formal investigations, and the UK privacy regulator has since launched an independent investigation. Thirty-five state attorneys general sent a joint letter requesting XAI to take action to prevent Grok from generating child sexual abuse material. At the federal level, some Pentagon agencies have used the Grok model to a limited extent, extending regulatory disputes from “consumer protection” to “national security procurement.”

Highlights from Friday's hearing

Friday's hearing will be the first time SpaceX will publicly answer AI security questions under oath conditions. Lawmakers are expected to ask three core questions: How do companies assess the catastrophic risks of AI? What mechanisms prevent companies from continuing to publish models when security reviews fail? If the AI system causes serious damage, who is responsible?

References have already been given in last week's hearings — representatives of the four major AI companies failed to provide quantitative answers to catastrophic risks, and generally avoided the promise of “whether to automatically put the model on hold if the safety review fails.” Whether SpaceX's response will be different is worth watching.