Summary of Key Points
Three AI-related companies—SpaceX, OpenAI, and Anthropic—with valuations approaching or exceeding one trillion US dollars are about to go public in large numbers, which could draw a significant amount of liquidity from global capital markets. The U.S. stock market has become dominated by the AI sector, with concentrated giants and conflicting opinions among investors (some are optimistic about the returns on AI infrastructure investments, while others are skeptical about the high valuations). Although these companies have high valuations, they are all currently losing money and rely on upstream suppliers of essential components such as chips and servers for their profitability. The downstream application layer has yet to demonstrate genuine demand. There are no Chinese companies with valuations in the trillion-dollar range, and while China is pursuing a path of domestic substitution for AI, it also faces challenges in verifying market demand. Ultimately, all these companies must answer the question: “Can AI truly create new demand?”
1. Can the Market Accommate the Public Offerings of These Three “Trillion-Dollar Giants”?
SpaceX (targeting a valuation of $1.75–2 trillion), OpenAI ($1 trillion), and Anthropic (nearly $1 trillion) plan to go public around 2026, with a combined valuation of over $3.6 trillion. This is staggering:
- More than the total IPO value during the Internet bubble in 2000: The combined market value of all 2,600 companies that went public that year was less than this amount.
- The impact on market liquidity could be significant: Based on typical IPO share sales (15%-25%), these companies would draw between $400 billion and $500 billion from the market, which is equivalent to using the total funds raised in U.S. IPOs over the past decade in just one quarter.
There are concerns that these “giants” could deplete global capital pools, so they may have to issue shares at very low sales ratios (3%-8%) to even get listed, but this would still squeeze the financing opportunities for other companies.
2. The U.S. Stock Market: A Dominated by AI
The total market value of the U.S. stock market has surpassed $75 trillion, but the composition is imbalanced: A few large companies, such as Apple and Microsoft, account for nearly 40% of the S&P 500’s value. In 2025, the S&P’s growth of 45% came from these seven companies, while the remaining 493 stocks saw little increase. The market is dominated by the AI sector.
Wall Street is divided into two camps:
- Optimists (JPMorgan Chase, Morgan Stanley): Believe that investment in AI infrastructure (chips, servers, etc.) is just beginning and will require $3–5 trillion over the next few years, creating huge demand.
- Pessimists (Goldman Sachs): Argue that valuations are too high (the S&P forward P/E ratio has reached 22 times, near its peak in 2021), and profits have not kept up with expectations. They fear that once growth slows, valuations will collapse.
The core issue is whether the money invested in AI can be recouped.
3. Behind the Trillion-Dollar Valuations: High Losses and Reliance on Upstream Suppliers
The financial health of these companies is somewhat questionable:
- SpaceX: Expected to lose $4.9 billion in 2025, with most of the loss coming from its xAI business; only its Starlink satellite internet service is profitable.
- OpenAI: Generated $570 million in revenue in the first quarter but incurred a loss of $1.22 for every dollar earned, and it may not turn a profit until 2029.
- Anthropic: Appears to be profitable, but this is due to aggressive expansion (signing agreements for 10 GW of computing power); its profitability is more of a temporary phenomenon.
Their high valuations are sustained because the upstream AI components (chips, servers, etc.) are generating substantial profits:
- NVIDIA reported an operating profit of $53.5 billion in one quarter, selling complete AI solutions.
- Dell’s AI server sales increased by 757%, with new orders worth $24.4 billion and cash flow of $4.1 billion—this reflects the real revenue from these upstream suppliers.
The problem lies in the downstream application layer: Apart from AI programming, there are few scalable, profitable AI applications. The savings in labor costs are often offset by increased technology costs, leaving net profits uncertain.
4. China’s AI Industry: Domestic Substitution and Similar Challenges
China’s approach to AI differs from that of the U.S., but the core issues remain the same:
- No trillion-dollar companies: Tencent (with a valuation of nearly 5 trillion RMB) is the largest Chinese AI company, but its U.S. equivalent would be around $70 billion.
- Market value is driven by cash flow: Bank stocks (ICBC, ABC) are performing well due to high dividends, while tech stocks (Tencent, Alibaba) benefit from revaluation of their AI assets.
- Domestic substitution strategy: Companies like Cambricon and摩尔线程 have a combined market value of only 4% of NVIDIA’s, yet their P/E ratios are higher (70 times vs. NVIDIA’s 40 times). Their success relies on expectations driven by government support and investments from large companies like BAT.
Whether it’s the U.S. model of “closed-loop applications” or China’s approach to domestic substitution, both need to reach a point where users are willing to pay for AI services. The company that succeeds in doing this will transform its vision into reality.
5. Three Tips for Chinese Business Leaders:
- Distinguish between Technology Leadership and Valuation-Driven Growth: Dell’s success is based on practical technology, while the high valuations of these companies are largely driven by speculation. Stay close to opportunities for technological transformation and avoid inflated bubbles.
- Focus on Cash Flow: As funds flow into AI, financing in non-AI sectors may dry up, so maintaining healthy cash reserves is crucial. Don’t abandon profitable businesses just because of inflated valuations.
- Create New Demand Rather than Simply Replace Old Labor: While AI can replace some jobs, creating new demand (similar to how the internet transformed industries like social media and e-commerce) is key to long-term success. The real winners will be those who can create new markets, not just those who write code quickly.
The race to dominate the AI industry has begun, but the winners won’t be the loudest proponents; rather, they will be those who know where to go. The real test lies in the actual growth of market demand.