Anthropic could raise $100bn, ten times Europe’s largest listing in decades

Anthropic is reportedly considering a listing as early as October that could raise around $100bn and imply a valuation of at least $2tn, potentially surpassing SpaceX’s fundraising record. The Financial Times notes that executives haven’t set a valuation target even privately, but the deal is now being modeled with specific figures. If realized at this scale, it would be a major sentiment and capital-markets catalyst for AI-focused private markets.

Analysis

A prospective mega-valuation for a frontier model lab is less about the company itself than about what public investors are being forced to underwrite: a multi-year compute arms race, not a one-cycle product story. That is structurally supportive for semis, networking, power, and data-center infrastructure names because private capital is signaling that training/inference demand is still outrunning supply discipline. The flip side is that public AI software names with weak monetization paths may face multiple compression as capital rotates toward the few players with scale, distribution, and balance-sheet access.

Second-order, this kind of listing can widen the moat for hyperscalers and incumbent cloud platforms while making it harder for smaller, compute-hungry startups to finance loss-making roadmaps. If the raise price is accepted, it is a green light for adjacent vendors selling picks-and-shovels to both model labs and cloud operators; if the deal has to be marked down or delayed, it will be read as the first real sign that private AI capital is becoming less forgiving. The main near-term risk is that the market treats the headline as validation and overbids the whole AI complex before any of the economics are public.

The contrarian view is that this is more a liquidity event than a fundamental milestone: a giant private valuation can coexist with weak public-market returns if burn rates stay high and monetization lags. Over 1-3 months, watch for evidence of capex re-acceleration from MSFT/AMZN/GOOGL and for any commentary on inference economics; over 6-18 months, the key falsifier is a downward revision to the listing range or a prolonged delay, which would signal the private AI bid is peaking rather than expanding.

Key Decisions for Investors

• Buy SMH on any post-headline dip; use a 1-3 month horizon and treat a sharp pullback in NVDA/AVGO after the deal chatter as a better entry than chasing strength.
• Long ANET or VRT vs. short a basket of high-multiple, low-revenue AI software names over the next 1-3 months; thesis is that infrastructure captures the monetizable spend while application names face valuation scrutiny.
• Add MSFT and AMZN as relative winners if the listing proceeds at the top of the expected range; they own the distribution layer and should capture follow-on demand even if the lab economics stay opaque.
• Set an alert for any pricing below the rumored level or a delay beyond October; that would be the trigger to fade AI beta via a short SMH hedge for 2-6 weeks.
• No direct trade in the name itself until filing terms are public; if a public proxy eventually appears, look to sell the first lockup-driven rally rather than buy the debut pop.