Why Anthropic May Let Early Investors Sell Shares, Unlike SpaceX and Cerebras

Anthropic has confidentially filed for an IPO and is considering letting existing shareholders sell shares in the IPO while extending lockups beyond the typical 180 days to delay when more pre-IPO shares hit the market. In a parallel setup, SpaceX’s first lockup expiration (Aug 2026) more than doubled its public float, illustrating how Anthropic’s secondary-vs-primary mix and longer lockups could materially affect near-term float and trading supply. The reported IPO size could exceed SpaceX’s ~$86B, but cash to Anthropic will depend on how much is primary versus secondary sales.

Analysis

This is less a pure growth-financing event than a liquidity-clearing mechanism, and that distinction matters for how the market should price the AI supply chain. If the deal skews secondary, the “IPO value” headline can look enormous while the company itself captures much less fresh capital, which limits any immediate incremental demand signal for compute vendors like NVDA, ARM, and SMCI. In that case, the main beneficiaries are existing holders and secondary desks, not the operating ecosystem that investors may reflexively bid on.

The first-order float effect is bullish for aftermarket stability if the company constrains the primary issuance, but the second-order effect is a delayed supply overhang: longer lockups push selling pressure out 6-18 months instead of removing it. That tends to compress the window where momentum buyers can rely on scarcity, and it increases the odds of a sharp reversal when staged unlocks begin. The risk is not day-one disappointment; it is that the market over-anchors on a high valuation multiple before understanding how little of the deal actually funds growth.

Contrarian view: consensus may be treating this as validation of the whole AI trade, when the cleaner read is that late-stage private holders want monetization before public-market discipline arrives. If the prospectus shows meaningful insider/founder liquidity or limited primary capital, the move in AI proxies is probably overdone. If, however, primary proceeds are large enough to fund incremental compute capacity, that would be the real bullish signal for semis and rack-scale infrastructure.

Near term, this is an alert rather than a clean trade until the filing discloses the primary/secondary mix and lockup terms. The best risk/reward setup is to fade any broad AI sympathy rally only if the document confirms a secondary-heavy structure; otherwise, stay flexible and treat the deal as a sentiment event, not a fundamentals event.

Key Decisions for Investors

• Set a filing-day alert on the primary vs. secondary split: if secondary exceeds ~50% and lockups extend beyond 180 days, fade any AI sympathy rally via short NVDA or SOXX into strength; target a 2-5% pullback over 1-3 weeks, stop on confirmation of a large primary raise.
• Do not chase the IPO headline itself until the prospectus is public; the key catalyst is cash to the company, not valuation. If proceeds are mostly secondary, treat the event as neutral-to-slightly negative for NVDA/ARM/SMCI demand expectations.
• If the deal is primary-heavy and explicitly funds new compute capacity, rotate to long NVDA against a less direct AI-beta basket only after pricing, not before; upside would come from a revised capex signal, not from the listing itself.
• Watch for post-pricing lockup overhang: if the first filing implies staged releases, prepare to short the broader AI enthusiasm trade 6-12 months later when the supply overhang becomes visible; this is the higher-conviction timing window than the IPO itself.