SpaceX, Anthropic, or OpenAI: Which IPO Is the Better Buy?

SpaceX is described as a $2 trillion valuation story with about $18 billion in 2025 revenue and Starlink generating $11.4 billion last year, but it still posted a $2.6 billion operating loss. Anthropic is the strongest fundamental performer in the article, with expected quarterly revenue of $10.9 billion and annualized revenue near $45 billion, while reportedly already profitable. OpenAI trails on revenue at an annualized $25 billion and is described as highly unprofitable, with a reported operating margin of -122%.

Analysis

The market is starting to price a very different AI stack than the one most investors own. If Anthropic is scaling enterprise monetization faster than OpenAI, the incremental beneficiary is less the consumer-facing model layer and more the infrastructure and distribution rails: cloud, chips, networking, and enterprise software that gets embedded into workflows. That supports GOOGL indirectly, because Alphabet is simultaneously a customer, competitor, and distribution gatekeeper; the more valuable enterprise AI becomes, the more pricing power shifts toward whoever controls default access and compute orchestration.

The bigger second-order issue is capital intensity. A $100B revenue outcome for these startups is not just a software story; it implies enormous GPU, power, and data-center demand over multiple years. That creates a durable demand floor for the picks-and-shovels beneficiaries, but it also raises the probability that public-market investors overpay for headline growth while underestimating gross margin compression, revenue-recognition differences, and future dilution from repeated capital raises.

The contrarian take is that the “winner” in the private market may be the one with the clearest path to profitable scale, not the one with the largest user base or the flashiest consumer narrative. OpenAI’s consumer mindshare is an asset, but unless monetization closes the gap quickly, it risks becoming a distribution giant with inferior unit economics. Meanwhile, SpaceX’s AI/data-center angle is still mostly optionality; the market is paying today for a multi-decade platform thesis that has execution risk at both the launch and orbital compute layers.

For public markets, the near-term trade is not to chase the IPOs but to own the infrastructure beneficiaries and fade the hype cycle if valuations detach from realized cash generation. The key reversal catalyst is a slowdown in enterprise AI spend or evidence that model training/inference efficiency improves faster than expected, which would reduce the compute intensity behind the narrative over the next 6-18 months.

Key Decisions for Investors

• Go long GOOGL on a 3-12 month horizon as a relative winner from AI distribution plus cloud monetization; use pullbacks rather than strength, with the thesis that incremental enterprise AI spend accrues to the hyperscalers before it accrues to private model labs.
• Buy a basket of AI infrastructure beneficiaries versus private-AI hype exposure: long GOOGL / short avoid-or-underweight high-multiple AI entrants if/when they list, as the risk/reward favors cash-generative platforms over unseasoned top-line growth.
• Initiate a pair trade: long AI compute/power beneficiaries, short duration-sensitive growth names that are priced for perpetual hypergrowth; if AI revenue growth normalizes, multiple compression will hit the highest-duration assets first.
• If the IPOs price at extreme enterprise-value-to-sales multiples, wait 30-60 days post-debut before initiating any long exposure; the better entry is likely after lockup/first-quarter audit scrutiny when revenue-quality questions surface.
• Monitor any announcement cadence from GOOGL around enterprise AI packaging and pricing; if monetization accelerates, add on confirmation because the market may re-rate Alphabet as an AI cash-flow compounder rather than a defensive mega-cap.