Author: Wang Chuan from Silicon Valley , investguru

1/ Many times, anxiety is a stress response in the amygdala of the brain . To alleviate anxiety, simply lecturing someone with comforting phrases like "Don't be anxious, anxiety won't help" is insufficient. This doesn't change the underlying mechanism behind the stress response. The truly effective method is to have the anxious person slowly write down the things that cause their anxiety. The process of writing transfers control of the brain from the amygdala, which is primarily focused on alertness and tension, to the prefrontal cortex, which is focused on rational thinking. The more detailed the writing and the deeper the analysis, the more thorough the transfer of control, and the anxiety naturally decreases. Even if you can't finish writing in one go, just starting to write and knowing roughly how long it will take to organize your thoughts will begin to reduce anxiety. Simply relying on motivational platitudes without forcing yourself to write slowly won't truly complete this transfer of control and will prevent you from fundamentally changing your anxiety.

One worrying aspect of May 2026 was the across-the-board surge in semiconductor stocks. The renowned Philadelphia Semiconductor Index rose 150% compared to a year ago. Stocks of various semiconductor memory companies saw widespread gains, with SNDK's closing price on May 8th at $1562, 38 times higher than a year prior. While I have no investment in semiconductor companies, I will attempt to analyze the underlying logic behind these stock price increases in this article and share my insights with readers.

3/ This story begins with the rapidly rising valuations of various AI startups. Many speculators on the periphery, who only care about stock price fluctuations, lack some basic common sense:

• First, the vast majority of AI startups are suffering heavy losses, but many people completely confuse the two basic concepts of revenue and profit (some even do it intentionally), and they never learn from this mistake, spreading misinformation.
• Second, when a company says its ARR (Annual Recurring Revenue) has reached $100 million, it doesn't mean its total revenue over the past year reached $100 million. At most, it only means its revenue for the most recent month reached $8.33 million. If you're being presumptuous and misunderstanding, that's your own problem.
• Third, a company's silence can reveal more information than it actually says. If it reports ARR but doesn't mention profitability, it implies significant losses. Six months ago, it claimed its ARR reached 100 million, but it has been silent for a long time, likely indicating shrinking revenue and even greater losses.

4/ Take Anthropic (hereinafter referred to as Anth), a currently booming large-scale model company, as an example. At the end of April, some media outlets reported that the company's ARR (Average Revenue per Hour) was approaching $40 billion. In other words, the company's revenue in April alone was close to $3.3 billion (400 divided by 12). In March, the company's ARR was less than $30 billion, meaning that its revenue in March was less than $2.5 billion. In a public document released in March, Anth's CFO disclosed that its cumulative revenue from its founding in 2021 to March 2026 exceeded $5 billion. In other words, Anth's cumulative revenue (to reiterate, not profit) over the past five years since its founding is approximately $10.8 billion.

5/ There's also the issue of revenue definition here. A large proportion of Anth's revenue comes from Google Cloud services and AWS. Cloud service providers take at least 20% of what customers pay, and then give the rest to Anth. Therefore, Anth's advertised ARR needs to be discounted by another 20%. But we won't dwell on that here, because Anth is still losing a lot of money every month.

6/ How much are the losses? Several different AIs were asked, and based on limited publicly available information, assuming monthly revenue reaches $3.3 billion, estimates for monthly losses ranged from $1.1 billion to $1.7 billion. This is the fundamental reason why Anth continues to seek external funding. As of the end of April, Anth had raised a total of $72.3 billion in funding over the past five years, but its cumulative revenue was only $10.8 billion, yet it still needs to raise more capital. Those who only follow news headlines mistakenly believe that Anth has already made a fortune, but they don't understand the immense pressure on management to constantly seek new external funding to survive.

7/Anth is not without competition. Its president recently admitted in a public statement that other US companies' AI models are only one to three months behind Anth's, while Chinese companies' AI models are six to twelve months behind. Product features from different companies are constantly evolving and changing rapidly, making it difficult to determine the ultimate winner based on performance over one or two years. For example, Cursor was the most popular programming aid among AI developers from 2023 to early 2025, and the company was extremely popular. However, Anth's Claude Code, launched in the first half of 2025, can automatically generate large amounts of code, allowing many non-programmers to participate in programming. This is 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 user experience and cost-effectiveness are inferior to OpenAI's CodeX in some aspects, and therefore switched to the latter. For many users, the switching cost is not high. In addition to OpenAI, SpaceX's Xai also began working closely with cursor last month to develop a competitor to Claude Code.

8/ Investors bullish on Anth often cite a period when Claude Code's profit margin for inference work reached 70%. However, this argument ignores two key points: first, the enormous cost Anth incurs while continuously training new models; and second, it assumes no competition and no pressure to lower prices. This is clearly unrealistic. When Anth raised funds in February, its valuation was $380 billion. Just three months later, it sought new funding with a valuation exceeding $900 billion. Recently, it reportedly reached a valuation of $1.2 trillion on a private equity secondary market. (Some podcasts have even proclaimed a long-term valuation of $5 trillion). This is a company with only five years of history, accumulated revenue (not profit!) of $10.5 billion, yet still losing billions of dollars every month. Just because of its rapid growth in the past year, some believe it can last forever. Raising $720, losing money for five years to generate $105 in revenue, and then cashing out at a $10,000 valuation in the capital market—no wonder so many entrepreneurs aspire to this! 😊

9/ Berkshire Hathaway (BRK), owned by Warren Buffett, is projected to have $370 billion in revenue, $45.9 billion in operating cash flow, $66.9 billion in after-tax profit, and nearly $400 billion in cash on hand, yet its market capitalization is only about $1 trillion. From another perspective, BRK's revenue in ten days equals Anth's total historical revenue. However, some investors are willing to believe that Anth, which still heavily relies on external investment for survival, is valued higher than BRK. The immense courage required to build such a belief is truly remarkable.

The valuation growth of 10/Anth and OpenAI is crucial to the surge in capital expenditure (capex) across the entire AI industry and the subsequent boom in the storage sector. Looking back at the developments of the past few years, this logical chain is very clear. To find out what happens next, stay tuned.