How to Buy Pre-IPO Shares in Anthropic and OpenAI (And Why You Might Not Want To)

OpenAI (valued at $852B) and Anthropic (valued at $965B) filed confidential SEC S-1s (May 22 and June 1) and could go public in late 2026/early 2027. The article warns that only a 5%-10% float could spark an initial pop, but secondary-share and SPV structures may be voided by the companies, potentially rendering pre-IPO shares worthless; it also flags ETF/proxy-fund NAV risks where hype can push prices above NAV before shares dump at the IPO. Overall, steep valuations plus net losses and intense AI competition are framed as likely catalysts for post-IPO repricing.

Analysis

The near-term winners are the toll collectors, not the underlying AI names: secondary venues like FRGE and proxy wrappers like HOOD-sponsored SPVs can see a short-lived surge in flows as retail chases scarcity. The problem is durability—if the issuer can void transactions or restrict transfers, the “asset” is really a contingent claim on future access, so fee revenue and NAV quality diverge fast. That makes any premium in VCX/ARKVX-type vehicles fragile once the IPO calendar becomes visible and the market can finally buy the real shares.

The second-order loser is the late-stage private-mark valuation stack across adjacent AI software names. If public markets eventually force OpenAI/Anthropic to trade on negative earnings power and capital intensity, it becomes harder to justify rich marks for other private AI application companies; meanwhile, infrastructure beneficiaries with tangible capex linkage remain the cleaner expression of the theme. In other words, the article is more bearish on synthetic pre-IPO access than on AI itself.

The main risk to the short-the-proxy thesis is timing: retail hype can keep premiums elevated for many months, especially if IPO rumors intensify into 2026. The thesis breaks if these funds continue to trade close to NAV after filings, or if issuers explicitly legitimize transferability. The better catalyst window is after the first credible pricing range or lockup discussion, when investors are forced to choose between opaque wrappers and direct public equity.

Key Decisions for Investors

• Fade retail proxy vehicles on strength: short VCX on any premium expansion versus reported NAV; target 1-3 month mean reversion, with invalidation if premium persists above ~10% for multiple weeks.
• Use FRGE as a tactical sell-the-rip name rather than a long-term compounder; any revenue pop from secondary volume is likely to be episodic and legally fragile, not recurring.
• Avoid or underweight HOOD as an indirect beneficiary of private-AI access products; the SPV angle adds reputational/regulatory risk without a clearly defensible, durable economic moat.
• Relative-value expression: long NVDA (or broader AI infrastructure basket) / short VCX or other proxy-fund exposure, since the former monetizes actual capex while the latter monetizes sentiment.
• Watch item, not recommendation: if OpenAI/Anthropic file updated terms that constrain secondary transfers, treat that as a bearish catalyst for secondary-market and SPV-related names within days.