Is the Trillion-Dollar AI Investment in Silicon Valley All Just a Bubble?

The technology industry in the United States is undergoing a profound transformation. Tech giants are gradually shifting from independent competition to cluster - style cooperation, especially in the field of artificial intelligence.

Although top artificial intelligence companies in Silicon Valley are still competing with each other, to some extent, they are engaging in large - scale cooperation in increasingly complex ways. Some investors view this change as ecosystem building, while others call it a "scam."

Roughly estimated, as a key player in connecting the artificial intelligence supply chain, OpenAI has completed transactions worth up to $1 trillion this year, including a $270 billion computing power deal with AMD and a $500 billion computing power deal with NVIDIA.

The continuous announcement of large - scale transactions has led the market to continue chasing technology company stocks in an environment of increasing political risks. However, this has made Wall Street analysts increasingly worried. A rise or fall of tens of billions of dollars in market value within just a few minutes often indicates an extremely unhealthy market, which inevitably reminds people of the dot - com bubble.

Reciprocal Transactions

Looking back at the transactions between OpenAI, AMD, and NVIDIA, the reciprocal model where Company A provides funds to Company B to buy Company A's products undoubtedly creates some new risks.

Even though the industry has created transaction values up to trillions of dollars, researchers still cannot clearly explain how artificial intelligence models will be implemented, and the financial mechanisms created behind the industry are becoming increasingly difficult for investors to understand and track.

This is extremely similar to the bursting of the dot - com bubble and the 2008 financial crisis, when both disasters created opaque and unconventional financing mechanisms.

Brian Mulberry, a client portfolio manager at Zacks Investment Management, said that the failure of any single transaction could have a worrying domino effect.

Different from historical crises, technology stocks currently account for an increasingly large proportion of the U.S. stock market. Currently, top technology stocks account for about 35% of the S&P 500 index, while in 1999, this proportion was less than 15%.

Meanwhile, since the launch of ChatGPT in October 2022, the S&P index has soared by 90%. Three - quarters of the increase comes from artificial intelligence companies including the "Magnificent Seven," while the remaining 493 companies have only risen by 25%, highlighting the high concentration of the upward trend.

What makes some investors even more uneasy is that OpenAI has increasingly become the core of the large - scale cluster in the U.S. artificial intelligence industry. It has intricate connections with chip - manufacturing enterprises, data center manufacturers, and investment groups, and can be said to be an important pillar supporting the artificial intelligence industry, which in turn is the core pillar supporting the U.S. economy.

This means that once OpenAI fails or investors lose confidence in the company, the entire artificial intelligence industry may be severely impacted, and other sectors of the U.S. economy will inevitably suffer collateral damage.

The Elephant in the Room

Michael O'Rourke, the chief market strategist at Jonestrading, said that the market currently behaves as if it believes that everyone who trades with OpenAI will be a winner. However, OpenAI is a company with negative cash flow, so it has nothing to lose by signing these transactions.

The stock price movement of Oracle this Tuesday may be a warning sign. After news spread in September that it had signed a $300 billion cloud - service agreement with OpenAI, Oracle's stock price rose by more than 35% in a single day.

But on Tuesday, news about the company's cloud - business sales data suddenly cooled the market. According to internal documents, the gross profit margin of its cloud - business sales to NVIDIA was only 14%, far lower than its overall gross profit margin of about 70%. The company's stock price dropped by more than 7% during Tuesday's trading session, reminding investors to pay attention to the fact that the stock price increases of some companies are out of line with their fundamentals.

Analysts also have many questions about OpenAI's new cooperation with AMD. Timothy Arcuri, an analyst at UBS, pointed out in a report this week that the sixth and final batch of stock warrants provided by AMD to OpenAI will only be triggered when AMD's market value reaches $1 trillion. At that time, the value of OpenAI's shares in AMD will be about $100 billion.

But Arcuri believes that it is more likely that OpenAI will sell its shares in AMD during this period to pay AMD's bills. Therefore, in essence, this is a plan for AMD to provide funds for this customer to purchase products.

In other words, OpenAI helps AMD drive up the stock price, and it will be retail and institutional investors who ultimately pay for OpenAI's purchase of AMD chips. In this process, OpenAI hardly has to pay any price, while AMD may gain a market share of up to 30%.

Another Elephant

In contrast, NVIDIA has started to cast a wide net using its already very strong financial strength. According to PitchBook data, NVIDIA participated in 52 different artificial - intelligence company venture - capital investments in 2024, and 50 of them were completed by the end of September this year. Its equity investment in xAI is the latest example.

Some people believe that this model is an important reason for the artificial - intelligence industry bubble. Orlando Bravo, the co - founder of Thoma Bravo, believes that a key change in the market now is that large companies with healthy balance sheets are providing financing for artificial - intelligence enterprises.

In his view, investors should not value a company with an annual recurring revenue of $50 million at $10 billion. Such valuations should at least match free cash flow of $1 billion to double investors' investments, and the process is quite difficult.

On the other hand, companies' unrestrained capital injections into other companies have also "fattened" a large number of enterprises. Lisa Shalett, the chief investment officer of Morgan Stanley Wealth Management, pointed out that her team originally thought that only the stocks of 7 to 10 companies in artificial intelligence would be significantly driven up, but now the number has expanded to about 40.

Shalett therefore warns that if the momentum of generative - artificial - intelligence capital expenditure weakens, the final situation will not be optimistic. She also pointed out that NVIDIA's large amount of capital may be an important signal that the business cycle is approaching some kind of end, that is, extending the financing stage, and it is actually buying its own customers.

Optimistic Forecasts

Although reciprocal transactions have made many analysts extremely nervous, some people are still optimistic. Vivek Arya, a senior analyst at Bank of America, pointed out that NVIDIA's investment in OpenAI is a strategic decision based on performance and competitive demand, rather than a speculative frenzy.

He also emphasized that NVIDIA is not bearing all the funds in the trillion - dollar transactions. In fact, the company's total investment in the past 12 months was less than $8 billion.

He believes that in the future, there will be four or five large - scale ecosystems in Silicon Valley. NVIDIA will not invest in all of them, but only in one, the most potentially disruptive one, OpenAI. However, the news that NVIDIA announced an investment in xAI on Wednesday may challenge this view.

David Solomon, the CEO of Goldman Sachs, said that huge expenditures do not determine prosperity or depression. Capital investment without output in the end is no different from other investments. People don't know what the final result will be.

This article is from the WeChat official account "Venture Capital Daily." Author: Ma Lan. Republished by 36Kr with permission.