Author:BIT Securities

The U.S. stock market is about to witness another historical moment.

According to media reports and rumors from the secondary market, AI company Anthropic is expected to officially go public as early as October, with a market valuation exceeding $2 trillion (This figure has not been officially confirmed by the company and is subject to the disclosures in the official prospectus). If this number is realized, it will set a new record for the largest IPO globally—surpassing the previous record set by SpaceX.

A company that was valued at only about $965 billion during its financing in May 2026, due to the rapid growth of its Claude series models, has had its trading price in the secondary market soar to its current heights. Yet, before this IPO has landed, an old question has been reignited: Has the AI investment boom entered the realm of bubble territory?

1. Bulls hold a strong hand

First, let’s examine where the supporters derive their confidence—Anthropic's recently disclosed second-quarter operational data was stronger than expected.

According to the market-reported operational data (not yet officially disclosed by the company; specifics should refer to the prospectus or official financial report), Anthropic's single-quarter revenue reached $11.5 billion, up from only $787 million in the same period last year, representing a year-over-year growth of 1,361%; this year’s first quarter was $4.7 billion, essentially doubling within three months. On the profitability front, the company has crossed a significant threshold: it has started making a profit, with the gross margin of its inference business jumping from 38% a year ago to over 70%.

The skepticism that “AI companies only invest without producing output” appears weak in light of these figures. Consequently, the narrative from the bulls is straightforward: the high valuation purchases the speed of AI commercialization and the extensive penetration of large models in the enterprise market.

2. But the quality of gross margin deserves a question mark

Why can the profit margin rise so quickly? Breaking down the cost structure reveals that the answer does not lie within Anthropic itself, but among its chip suppliers: Google's TPU and custom chips from Broadcom.

Currently, the landscape of computing power is as follows—during the training phase, NVIDIA's monopoly remains unshaken; however, expenditures in the inference phase are increasingly shifting toward custom chips, and the larger the business scale, the faster this shift is occurring. Anthropic's gross margin rise is largely benefiting from the reduction in inference costs rather than an improvement in pricing power.

This directly impacts an upcoming observation window: NVIDIA's financial report on August 26, where the revenue growth from the inference business will serve as the most intuitive scale for measuring the progress of this "inference migration."

3. The bear's perspective: Overvalued and liquidity siphoning

Next, let’s look at what the skeptics are worried about.

First is the pricing method. A $2 trillion valuation corresponds to expected revenues for 2028—today's price tag is affixed to two years down the line's report card. Such pricing places almost harsh demands on growth rates: if revenue growth slides from ten times a year to merely one or two times a year, the existing valuation framework will lose its support.

Second is liquidity. The largest IPO in history means pulling away hundreds of billions of funds from the secondary market at once; the more troubling issue is that OpenAI is also in line behind. By the time Anthropic officially rings the bell, there may not be enough money in the market to go around.

Therefore, two things are worth monitoring: the growth rate of inference revenue in NVIDIA's financial report; and whether the revenue metrics disclosed when Anthropic formally submits its prospectus align with the data circulating in the secondary market—discrepancies in metrics often signal a change in the narrative.

4. Stalemate between bulls and bears? Bank of America proposes a "double bet" strategy

Whether AI has entered bubble territory is a contention between bulls and bears. According to publicly available reports, Bank of America's chief strategist Michael Hartnett in his latest research report suggests that there is no need to pick sides between bulls and bears; instead, consider concurrently allocating two types of assets (specific strategy details should be verified against Bank of America's official research report; transcription may involve information loss).

His recommendation is to simultaneously go long on leading AI tech stocks and on forgotten assets that have been out of the spotlight for years, capturing gains from the final peak phase of nominal GDP bubbles while shorting AI-related bonds.

This judgment is supported by a set of data. Bank of America’s Bull & Bear Indicator fell from 9.7 to 9.3, still lingering in the extreme bullish zone, with the "sell" signal remaining intact—yet, since this signal lit up in May, global stock markets have been climbing. The flow of funds is also diverging: gold and commodities continue to receive structural inflows, tech stocks have just experienced the largest weekly outflow in seven weeks, while private clients’ equity holdings have risen to historic peaks.

Hartnett cites the patterns of historical bubbles, noting that when bubbles are near their peak, emerging markets or severely oversold cyclical assets often benefit from spillover effects; this time, the consumer sector is most likely to take up the baton. The rationale for shorting AI bonds is equally clear: over $1 trillion in AI capital expenditure paired with negative cash flow means a flood of related corporate bond issuances is imminent, and under pressure from supply, prices are unlikely to improve.

The core of this strategy is encapsulated in the phrase: in the most frenzied last stages of a bubble, returns are often the richest—but be wary: you cannot place all your chips on one side.

Important Notice: BIT Securities operates through its group company BIT (formerly Matrixport) under the main entity Matrix Gelephu Pte. Ltd., regulated by the Financial Services Office of Grlep, Bhutan, providing eligible clients with access to U.S. stock trading services. Related trades are executed and cleared through U.S. registered broker-dealers. This article is written and provided by an external contributor. The valuation, financial data, and listing timeline of Anthropic mentioned herein are compiled from publicly available reports and market rumors and have not been officially confirmed; specific details are subject to the company's officially disclosed documents. The opinions, analyses, and judgments expressed herein represent the views of the author only and do not reflect the official position of BIT or BIT Research; BIT also does not guarantee the accuracy, completeness, or timeliness of related content. The third-party institutional viewpoints referenced in this article (including those of Bank of America Securities strategists) represent only their own or the analysts’ personal opinions and do not reflect BIT’s stance, nor do they guarantee accuracy. This article does not constitute investment advice, an offer, or an invitation to offer. Valuation of unlisted companies and volatility of AI-related assets are high, past growth rates do not guarantee future performance, and investors should make independent assessments and seek professional advice.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。