Ahead of its IPO, Anthropic’s Daniela Amodei shrugs off doubts about AI’s returns

Anthropic said it filed confidentially for an IPO after its $65 billion fundraise at a $965 billion valuation was described as greatly oversubscribed, underscoring strong investor demand for AI exposure. The company also said annualized revenue crossed $47 billion in May, up from roughly $9 billion at the end of 2025, though management flagged heavy capital needs for model training and inference. The article highlights continued AI sector momentum, with Anthropic leaning on public markets and external compute partnerships rather than building its own data centers.

Analysis

Anthropic’s IPO path is less about liquidity and more about converting private-market scarcity into a durable financing machine. The second-order implication is that frontier AI is moving from a venture-style capitalization model to a quasi-utility model: the winners will be firms that can repeatedly raise at scale, while smaller model labs face a widening cost-of-capital gap. That should compress the strategic window for mid-tier AI startups that cannot credibly fund multi-year inference and training commitments.

For public equities, the more important signal is not the IPO itself but the implied validation of continued enterprise AI spend, which is still the core debate. If customers start treating AI as a variable-cost productivity layer rather than a discretionary software line item, revenue durability improves across the stack; if not, the market will quickly reprice the growth assumptions embedded in high-multiple software and mobility names with heavy AI messaging. UBER is the cleanest canary: its negative data tag fits the risk that AI spend may not yet be translating into measurable margin expansion, making any slowdown in corporate AI budgets disproportionately painful for “AI-enabled efficiency” stories.

The compute decision is also telling. By not internalizing capex, Anthropic is preserving flexibility and avoiding stranded-asset risk, but it is also effectively outsourcing balance-sheet intensity to infrastructure partners. That favors hyperscalers, colocation providers, and networked infrastructure suppliers with the cheapest incremental power and financing, while leaving model providers more exposed to margin compression if utilization stalls. The xAI capacity deal suggests compute is already becoming a liquid market, which should lower barriers for well-capitalized rivals and intensify price competition over the next 6-18 months.

Contrarian read: the market may be overconfident that private demand directly translates into public-market upside for the entire AI complex. The tighter trade is to own picks-and-shovels and short the most narrative-driven beneficiaries of AI spend where monetization is still unproven, especially if enterprise IT budgets soften into year-end. The key catalyst to watch is whether hyperscaler capex guides stay elevated over the next two earnings seasons; if they roll over, the AI trade becomes much more selective.

Key Decisions for Investors

• Short UBER vs long MSFT or AMZN over the next 1-2 quarters: express the view that AI monetization is more proven in platform infrastructure than in mobility efficiency spending; tight stop if UBER shows margin expansion from AI initiatives.
• Add to infrastructure beneficiaries exposed to AI compute demand (e.g., AMAT, MU, ANET) on any post-IPO enthusiasm in frontier AI names; 6-12 month horizon, with upside tied to continued capex intensity and downside cushioned by diversified end markets.
• Pair trade: long hyperscaler/cloud infrastructure basket vs short high-multiple AI software names with limited current revenue proof; target 10-15% relative outperformance if enterprise AI ROI scrutiny rises over the next two earnings seasons.
• Buy 3-6 month downside protection on AI-rich software/index exposure via QQQ or SMH put spreads into upcoming earnings, as a hedge against any guide-down in AI-related capex or corporate spend.
• Avoid chasing private-market AI secondary exposure after the latest valuation print; wait for the IPO filing window to reveal dilution, lockup overhang, and any evidence of slowing net retention before adding risk.