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

The article compares three prospective mega-IPOs, highlighting SpaceX at a $2 trillion valuation versus 2025 revenue of $18 billion, Anthropic at roughly $10.9 billion in quarterly revenue annualizing to about $45 billion, and OpenAI at an estimated $25 billion annualized revenue but a reported operating margin of -122%. It argues Anthropic currently has the strongest revenue growth and profitability profile, while OpenAI is lagging on revenue and remains highly unprofitable. The piece is largely valuation commentary rather than a fresh catalyst, so direct market impact is limited.

Analysis

The market is still pricing this as an AI-versus-space narrative, but the more important split is between distribution and cost of computation. GOOGL is the quiet beneficiary because it is one of the few platforms large enough to both fund frontier-model demand and absorb the capex/traffic risk if external vendors become more expensive or unreliable; every dollar spent on third-party inference is a dollar not spent across its own stack. NVDA benefits in the near term from any arms race in training and inference, but the longer this turns into a multi-vendor procurement market, the more pricing power shifts from model labs to chip suppliers and hyperscale operators that can enforce volume commitments.

The biggest second-order loser is not the model labs, it is any business model predicated on “winner-take-most” consumer AI monetization without clear unit economics. OpenAI’s under-margin structure implies a financing treadmill: if revenue growth decelerates even modestly, the implied need for repeated capital raises becomes a control issue, not just a valuation issue. Anthropic’s better profitability profile is strategically valuable because it buys time, but it also makes the company more dependent on enterprise retention and less on explosive consumer virality; that usually compresses terminal multiples once the growth curve normalizes.

For INTC, the relevant angle is not direct competition with the startups, but the broader reindustrialization of AI infrastructure. If compute demand keeps expanding, any credible domestic silicon alternative or packaging/advanced-node capacity becomes more valuable, especially if geopolitics constrain supply chains or export rules tighten. That said, the timeline matters: the revenue inflection for non-NVDA beneficiaries is measured in quarters to years, while the tradeable move in private-market repricing can happen in days around IPO pricing and secondary-round marks.

Consensus is probably overestimating how much “headline revenue” converts into durable equity value at these stages. The market is likely underpricing cancellation optionality, customer concentration, and the risk that enterprise AI spend consolidates around the lowest-cost, most integrated stack rather than the fastest-growing standalone model. If the IPOs clear at extreme revenue multiples, the better setup may be to fade the froth in the least profitable name and own the picks-and-shovels that collect tolls regardless of which model wins.

Key Decisions for Investors

• Long GOOGL / short a basket of private AI exposure proxies via liquid tech beta if available; thesis is that hyperscale distribution and integrated inference economics outperform standalone model labs over the next 6-12 months.
• Buy NVDA on IPO-related AI enthusiasm pullbacks; use 3-6 month horizon and target strength on any confirmation of capex expansion, but trim if valuation extends without fresh order commentary.
• Avoid owning the lowest-margin AI lab at IPO if pricing implies a premium to Anthropic despite weaker unit economics; if forced, express it as a small position with strict stop-loss due to financing risk.
• Long INTC as a 12-24 month optionality trade on domestic AI infrastructure buildout, but only on weakness; treat it as a catalyst-driven position rather than a core fundamental compounder.
• Pair trade: long AI infrastructure beneficiaries, short the most crowded frontier-AI name into IPO windows; risk/reward favors fading narrative-driven multiple expansion when audited profitability is absent.