Author: Silicon Valley Wang Chuan, investguru

1/ A person's anxiety is often a stress response from the brain's amygdala. To relieve anxiety, one cannot simply say, "Do not be anxious; anxiety is useless," and other comforting words. This does not change the underlying mechanisms behind the stress response. A truly effective method is to have the anxious person gradually write down the things and reasons causing their anxiety on paper. The writing process transfers control from the amygdala, which only knows tension and alertness, to the rational thinking prefrontal cortex. The more detailed they write and the deeper the analysis, the more complete the transfer of control, and the anxiety naturally decreases. Even if they cannot finish writing in one sitting, as long as they start writing and have an idea of how long it will take to sort out the logic, the anxiety will begin to decrease. If one does not force themselves to write it down gradually, but only drinks mental chicken soup, they will not truly complete this transfer of control, and thus cannot fundamentally change the anxiety.

2/ One anxious matter in May 2026 is the overall surge in stock prices across the semiconductor industry. The famous Philadelphia Semiconductor Index has risen by 150% compared to a year ago. Various semiconductor storage companies have also seen their stocks rise significantly, with SNDK's closing price on May 8th at $1562, which is 38 times that of a year ago. The author does not invest in semiconductor companies but attempts to dissect the underlying logic behind these stock increases and share it for discussion with readers.

3/ This story starts with the rapid increase in valuation of various AI startup companies. Many speculators, who only care about stock price fluctuations, lack some basic common sense:

• First, the vast majority of AI startups are severely loss-making, yet many confuse income and profit—some even do it intentionally—and repeatedly propagate misinformation.
• Second, when a company states that its ARR (Annual Recurring Revenue) has reached $100 million, it does not mean that its total revenue for the past year is $100 million; at most, it means that its revenue for the most recent month is $8.33 million. If you mistakenly take this to heart, the misunderstanding is your own issue.
• Third, a company may reveal little, but the amount of information it does disclose is often more telling than what is explicitly stated. If it reports an ARR but does not mention profitability, it implies serious losses. If it said six months ago that its ARR was $100 million but has not communicated since, it likely means that its income has shrunk and losses have worsened.

4/ Taking the currently popular large model company Anthropic (hereinafter referred to as Anth) as an example, information disclosed by some media at the end of April indicated that the company's ARR was nearing $40 billion. In other words, the company's revenue for the month of April was close to $3.3 billion ($40 billion divided by 12). However, in March, the company's ARR was still below $30 billion, meaning that March's revenue was below $2.5 billion. Anth's CFO disclosed in a public document in March that from its founding in 2021 to March 2026, it had accumulated revenue exceeding $5 billion. In other words, Anth's total cumulative revenue over the past five years since its founding (emphasizing again, not profit) is about $10.8 billion.

5/ There is also an issue regarding the definition of revenue. A significant portion of Anth's revenue comes from Google Cloud Services and AWS. Cloud service providers take at least a 20% commission on the customer's payments, leaving the remainder for Anth. Therefore, the ARR that Anth promotes externally needs to be discounted by another 20%. However, we will not delve into that here, as Anth continues to lose a lot of money each month.

6/ How much are the losses? After asking several different AI sources and based on limited public information, it's speculated that if the monthly revenue reaches $3.3 billion, monthly losses range from $1.1 billion to $1.7 billion. This is also the fundamental reason why Anth continues to seek external financing. As of the end of April, Anth has accumulated $72.3 billion in financing over five years, but its cumulative revenue is only $10.8 billion, and it still needs to continue financing. Those who only read the headlines may mistakenly believe that Anth has already made a fortune, not realizing the immense pressure on management to constantly seek new external funds to sustain operations.

7/ Anth is not without competition; its president recently acknowledged in public statements that AI models from other American companies lag behind Anth by just one to three months, while Chinese companies' AI models lag by six to twelve months. The functionality of various company products is rapidly changing, making it very difficult to judge who will be the ultimate winner based solely on a year or two of performance. For instance, Cursor was the most popular programming aid among AI developers from 2023 to early 2025, and the company became quite hot. However, the Claude Code launched by Anth in the first half of 2025 can automatically generate large amounts of code, allowing many who were not originally programmers to participate in programming. This has been the core driving force behind Anth's rapid rise in revenue and valuation over the past year. Recently, some programmers have complained that Claude Code's experience and cost-effectiveness in certain areas do not compare well with OpenAI's Codex, leading them to switch to the latter. For many users, the switching cost is not high. Besides OpenAI, SpaceX's Xai also began close cooperation with Cursor last month to develop a competitor to Claude Code.

8/ Investors bullish on Anth would cite scenarios from certain periods, claiming that Claude Code's profit margin when doing inference work is as high as 70%. However, this assertion ignores the enormous cost of continuously training new models for Anth and assumes that competitors do not exist, without price pressures. This clearly does not align with reality. When Anth raised funds in February, its valuation was $380 billion. Just three months later, it began seeking new financing, with a valuation proposed to be over $900 billion. Recently, during trading on a secondary market for private equity, a valuation as high as $12 trillion reportedly emerged. (Some podcasters even mentioned a long-term valuation of $50 trillion). This is a company with only five years of history, accumulated revenue (not profit!) of $10.5 billion, yet continues to lose billions each month, and merely due to rapid growth over the past year, some believe it has a sustainable future. Financing $720 billion and operating at a loss for five years with $10.5 billion in business, and can cash out at a $10 trillion valuation in the capital market—this kind of wonderful situation is no wonder so many entrepreneurs yearn for it! 😊

9/ Berkshire Hathaway (hereinafter referred to as BRK), under Buffett's leadership, has a projected income of $370 billion in 2025, operating cash flow of $45.9 billion, after-tax profit of $66.9 billion, close to $400 billion in cash on hand, and a market value of only about $1 trillion. From another perspective, BRK's income over ten days is equal to Anth's total accumulated income to date, yet some investors are willing to believe that the still-highly-dependent-on-external-investment Anth has a higher valuation than BRK. The immense courage required to build such a belief is astonishing.

10/ The valuation growth of Anth and OpenAI is crucial for the massive increase in capital expenditures (capex) in the entire AI industry, as well as the subsequent prosperity of the storage industry. Looking back at the developments over the past few years, this logical chain is very clear. To know what happens next, stay tuned for the next breakdown.

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。