Think SpaceX's IPO Is Too Expensive? 2 Recent Developments Could Change That

SpaceX is being discussed for a potential IPO at a $1.77 trillion valuation, implying about 95x 2025 sales on $18.7 billion of revenue. New Anthropic and Google compute deals could add roughly $26 billion of annual revenue, lowering the implied multiple to about 35x forward sales and materially improving the valuation case. The article remains cautious overall, but the deals create a significant new AI revenue stream that could reshape expectations for the company.

Analysis

The market is likely underappreciating how quickly these contracts re-rate the implied revenue mix and reduce the dependence on the more cyclical launch business. The key second-order effect is not just higher top-line visibility for the private market story, but a more financeable asset base: reserved compute capacity can support better debt capacity, tighter IPO pricing ranges, and a lower effective discount rate for future capital raises. That matters because the multiple compression from 95x to ~35x sales is not just arithmetic — it changes who can own the stock at listing and whether the deal can clear with institutional support.

For Alphabet, the deal is more interesting as a scarcity play than as an earnings driver. It signals that hyperscalers may be forced to pay up for third-party GPU capacity when internal deployment lags demand, which reinforces the premium on firms that can secure power, chips, and rack density ahead of time. Nvidia benefits indirectly through utilization and ASP support, but the real strategic winner is the compute landlord layer that can monetize idle capacity; that said, this also raises the risk that future AI spend gets reallocated away from new GPU orders and toward rental-like contracts, which could flatten incremental hardware demand over a 6-12 month horizon.

The contrarian concern is that the new revenue headline may be masking concentration risk and non-recurring demand. If these contracts are primarily a way to absorb excess capacity, renewal risk rises after the initial term, and the market could eventually mark the revenue at a lower quality multiple than standard recurring software revenue. For Intel, the implication is mostly negative/neutral: if external compute supply remains plentiful enough to route around in-house infrastructure, it reinforces how far behind legacy chipmakers are in the AI stack and how hard it will be to reclaim pricing power in the next 1-2 years.

Key Decisions for Investors

• Long GOOGL into the IPO window over 3-6 months; the market should reward evidence that Alphabet is securing scarce AI infrastructure through off-balance-sheet capacity rather than only capex-heavy buildout. Risk/reward: modest upside to earnings visibility with limited multiple downside if the contract is framed as strategic, not dilutive.
• Add NVDA on weakness over the next 1-3 months; higher utilization across the ecosystem should support GPU pricing and extend the AI capex cycle. Risk: if large customers pivot from buying GPUs to renting capacity, near-term unit growth could decelerate even as ecosystem demand stays strong.
• Avoid or underweight INTC for 6-12 months; this setup highlights the value chain’s preference for integrated AI compute environments over legacy semiconductor capacity. The thesis is negative not because of direct exposure, but because the market may keep awarding the scarce asset premium elsewhere.
• If SpaceX lists, consider a tactical post-lockup approach rather than buying the IPO outright; the first tradable quarter should be about proving contract durability and margin structure. Better entry likely comes after the market separates headline sales from contract quality.
• Pair trade: long NVDA / short a basket of slower-moving legacy hardware names over 3-6 months; the contract structure reinforces demand for premium chips and penalizes firms without pricing power in AI infrastructure. Use tight risk controls because sentiment can swing quickly if AI capex decelerates.