Anthropic’s IPO could be the biggest ever. Your AI budget helps explain why

An early Anthropic investor, speaking anonymously, predicts an IPO price of about $1.8 trillion and a post-IPO value of $2.3 trillion to $2.4 trillion, despite Anthropic not yet having filed its S-1; the forecast is below the company’s reported $2 trillion target at pricing. The article says about 80% of Anthropic revenue comes from corporate customers, versus roughly half of OpenAI’s sales. The investor expects AI prices to decline as open models gain ground, while rising usage could still push total enterprise spending higher; the valuation outlook remains his view, not a public-market verdict.

Analysis

The investable signal is not the rumored private-market valuation; it is the possibility that enterprise AI shifts from scarce, high-priced capacity to cheaper, heavily used inference. If unit costs fall faster than usage rises, enterprise adoption can broaden while value accrues to customers and software distributors rather than model providers. If usage outpaces price declines, aggregate budgets and infrastructure demand can still grow—but falling prices may weaken providers’ ability to convert usage into durable margins. The key metric is therefore revenue and gross profit per unit of inference, not user growth alone.

The IPO comparison with SpaceX is a weak valuation anchor: first-day market capitalization says little about sustainable earnings, float, or subsequent dilution. The forecast comes from an unnamed early backer, not disclosed financials or a filed S-1. A multi-trillion-dollar valuation would make execution and eventual public-market liquidity unusually consequential; any gap between adoption and monetization could drive sharp multiple compression.

Near term, there is no direct public-equity catalyst here, and the article provides no basis for a trade in SPCX. Over 1–3 months, watch for an S-1, audited revenue quality, customer concentration, gross-margin trends, and evidence that enterprise usage converts to paid, retained workloads. Over 6–18 months, the structural test is whether cheaper models expand total spend without commoditizing providers. The thesis weakens if inference prices decline while usage growth or retention disappoints, or if disclosed margins fail to support the valuation narrative.

Key Decisions for Investors

• Do not use the reported valuation forecast as a reason to buy SPCX: it is a comparison, not evidence of a change in SpaceX fundamentals.
• Keep Anthropic exposure on watch rather than treating the investor’s estimate as price discovery. Reassess only after an S-1 provides audited growth, gross-margin, customer-concentration, dilution, and share-supply details.
• For public AI exposure, favor companies with demonstrable ability to capture enterprise workflow spend over businesses whose economics depend primarily on maintaining high inference prices; verify monetization and margin trends before sizing.
• Set a falsification trigger: if enterprise usage accelerates but provider revenue per unit and gross profit fall persistently, reduce exposure to model-provider monetization claims; if paid workload growth offsets price declines and margins stabilize, the adoption thesis gains credibility.