The three hard-tech moonshots fueling SpaceX’s unbelievable IPO

SpaceX is set to launch its IPO, with investors showing significant interest despite the company's current financial losses and the inherent risks of large IPOs. The offering is reportedly deeply oversubscribed, indicating strong market confidence in Elon Musk's ventures, whose erratic online behavior typically doesn't deter tech investors.

The core of SpaceX's pre-IPO strategy revolves around an ambitious AI business focused on orbital data centers. This vision, which emerged in the last 18 months, aims to unify Musk's conglomerate and requires three major engineering feats: a reusable rocket, a new American chip foundry, and unprecedented speed in satellite manufacturing.

Financial analysts, including Morningstar and NYU Professor Aswath Damodaran, have presented more conservative valuations for SpaceX, significantly lower than the company's internal assessment of $1.8 trillion. Morningstar valued the company at $825 billion, while Damodaran suggested $1.2 trillion. The discrepancy largely stems from integrating a high-risk AI business with SpaceX's more stable space launch and satellite internet operations.

SpaceX's AI strategy, as outlined in its S-1 market analysis, targets the enterprise AI market, envisioning its models powering coding tools and equipping digital agents for white-collar labor. The company projects a total market of $22.7 trillion for this sector. However, this focus appears to contradict recent deals to sell compute capacity to rivals like Anthropic and Google, suggesting a potential shift towards becoming a compute provider.

The challenge for SpaceX lies in balancing its role as a compute provider versus a model-builder. Industry experts generally believe large-scale space data centers are a decade away, but Musk asserts they are much closer. He highlights SpaceX's unique capability to cost-effectively place mass into orbit, build solar panels, and manufacture chips, essential for developing these data centers. While Musk aims for an annualized rate of a gigawatt per year by the end of next year in space AI compute, this goal necessitates a production rate of 6,666 satellites annually, significantly higher than current rates, from facilities yet to be built.

Further complicating matters are the development of Terafab, the company's proposed chip foundry, and the reliance on Starship for economical orbital deployment. Chip foundries are complex, multibillion-dollar projects that can take a decade to complete. The rapid reusability of Starship remains crucial for the economic viability of launching such a large number of satellites and chips into orbit.