Morgan Stanley Says SpaceX at $100 Prices Its AI Unit at Zero

The line that stands out from Morgan Stanley's latest SpaceX note, reported by Bloomberg, is what the market would be saying about the company's AI business if the stock slid to $100. According to analyst Adam Jonas, at that price investors would be assigning essentially no value to the AI operation SpaceX absorbed when it merged with xAI earlier this year, and undervaluing the launch and connectivity businesses on top of that.

The setup matters. SpaceX went public in mid-June in a record $86 billion offering, surged nearly 50% in its first three sessions, then rolled over, touching a low of $110.85, about 18% below the IPO price. The next pressure point is August 6, when the first post-IPO lockup expires and up to 911.5 million shares become eligible to trade. Jonas's framing, as summarized by Yahoo Finance, is that many investors now see a drop to $100 as realistic once insiders can sell.

The analyst's own view is the opposite of that pessimism. He reiterated an Overweight rating and a $300 price target, describing SpaceX as "uniquely positioned across launch, connectivity, and AI," and argued the selloff has disconnected from largely unchanged fundamentals. Morgan Stanley's longer-arc case is that the combined SpaceX-xAI entity could generate roughly $190 billion in AI-related revenue by 2030, which is what makes the current price look like a discount to them, not a warning.

The honest caveat is that this is one bank's model talking. The $190 billion 2030 figure is a projection, not booked revenue, and the reporting does not spell out how Morgan Stanley separates the "AI" slice of enterprise value from launch and Starlink cash flows, or how much of the freshly unlocked float insiders actually intend to sell. Take the specifics as reported, not settled.

What is worth watching beyond SpaceX itself is the repricing signal. If a name this large can have its AI premium quietly zeroed out by a lockup calendar, every other AI-heavy secondary mark, from private compute plays to still-private model labs, gets a fresh reference point for what public investors will and will not pay for embedded AI optionality.