Fierce market debate over the "AI bubble"; Deutsche Bank advises investors: Don’t try to "time the market," long-term holding is the best strategy.

Whether the AI investment boom has become a bubble has become a focal point in the market, but the latest research from Deutsche Bank shows that discussions on the "AI bubble" itself have already cooled down. The bank suggests abandoning market timing strategies and sticking to long-term holdings to achieve the best returns.

Tech giants are spending hundreds of billions of dollars to build AI infrastructure, and this unprecedented scale of investment has raised market concerns about bubble risks. From OpenAI’s $500 billion Stargate plan to Meta’s commitment to invest hundreds of billions in data centers, capital expenditures in AI have reached staggering levels.

Bain & Company predicts that by 2030, AI companies will need $2 trillion in annual revenue to support required computing power, but actual revenue may fall short by $800 billion. This huge gap intensifies market doubts about the sustainability of the current AI investment boom.

According to a previous article by Jiemen, Amazon founder Jeff Bezos recently said that the current AI investment is a "good bubble" and that, even if it bursts, it will provide long-term benefits for society. Meanwhile, Goldman Sachs CEO David Solomon warned that much of the capital flowing into AI might not yield the expected returns.

In its latest research report, Deutsche Bank believes that although there are bubble risks in the market, trying to time the market is extremely difficult, and a long-term investment strategy is more viable. Historical data demonstrates that missing the best trading days significantly reduces investment returns.

Aggressive Investments by Tech Giants Trigger Bubble Concerns

Since the beginning of this year, tech giants have frequently announced AI infrastructure investments in the hundreds of billions of dollars.

In January, OpenAI CEO Sam Altman announced the $500 billion “Stargate” AI infrastructure plan. Subsequently, Meta’s Mark Zuckerberg promised to invest hundreds of billions in building data centers, and Altman even stated that OpenAI expects to spend "trillions of dollars" on AI infrastructure.

The construction of AI infrastructure is generating unprecedented financing arrangements. Nvidia has agreed to invest up to $100 billion in OpenAI's data center construction, prompting analysts to question whether the chipmaker is supporting customers to maintain demand for its products.

OpenAI is also considering debt financing rather than continuing to rely on partners like Microsoft and Oracle. According to tech media The Information, OpenAI expects to burn through $115 billion in cash by 2029.

Other tech giants are increasingly relying on debt to support their unprecedented spending.

Meta secured $26 billion in financing to build data center campuses in Louisiana, while JPMorgan Chase and Mitsubishi UFJ Financial Group led over $22 billion in loans to Vantage Data Centers.

Some unproven companies are also trying to ride the data center gold rush.

Amsterdam-based cloud service provider Nebius, spun off from Russian internet giant Yandex, signed a $19.4 billion infrastructure deal with Microsoft.

UK data center firm Nscale, which previously focused on cryptocurrency mining, is now working with Nvidia, OpenAI, and Microsoft to build data centers in Europe.

A study by MIT found that 95% of organizations have not seen any return on their AI investments.

Meanwhile, Harvard and Stanford researchers pointed out that "work junk" created by employees using AI could cost large organizations millions of dollars in lost productivity annually.

Technological Development Encounters Bottlenecks

The AI technology itself also faces challenges. For years, developers like OpenAI and Anthropic have relied on the "scaling law," believing that more computing power, data, and larger models would bring leaps in AI capabilities. However, the returns from these efforts have diminished over the past year.

In August, OpenAI CEO Sam Altman admitted, "We are still missing some very important things" to achieve artificial general intelligence. The much anticipated GPT-5 model release received a lukewarm response and failed to meet previous hype.

In January this year, the release of DeepSeek's low-cost AI model triggered a trillion-dollar sell-off in tech stocks, with Nvidia plummeting 17% in a single day. Although stock prices later rebounded, this event highlighted the potential risks of AI investment.

In addition, the large-scale construction of AI industry data centers also faces real restrictions from the stressed national power grid. The greatly increased electricity consumption may be hindered by infrastructure limitations.

Nevertheless, industry leaders in AI remain optimistic. Altman acknowledges that investors may be "overexcited," but insists that AI is "the most important thing in a long time." Zuckerberg stated that an AI bubble "very likely" exists, but he is more worried about insufficient investment.

The Market Intensely Debates the AI Bubble

At present, the AI investment bubble is attracting increasing debate.

David Einhorn, founder of hedge fund Greenlight Capital, stated:

"The numbers being thrown around are so extreme, it’s really hard to fathom. I’m sure it’s not zero, but there’s a reasonable chance that a lot of capital will be destroyed in this cycle."

OpenAI chairman and CEO of $10 billion AI startup Sierra, Bret Taylor, believes there are many similarities to the internet bubble.

"AI changing the economy is true. I think it’s going to create tremendous economic value in the future, just like the internet. I also think we’re in a bubble, and a lot of people are going to lose a lot of money."

According to Jiemen, at a recent event, Bezos said that the current AI boom should be seen as an "industrial bubble" rather than a "financial bubble." He explained that even if an industrial bubble bursts, it leaves valuable legacy behind—just as fiber optic cable investments from the internet bubble era laid the foundation for later development.

Bezos recalled that during the internet bubble, Amazon’s stock price fell from $113 to $6, but the company’s business remained strong. He emphasized, "AI is real, and it will change every industry."

Goldman Sachs CEO Solomon adopts a more cautious approach. Although he recognizes AI’s potential to boost productivity, he warns that the large sums invested in AI may ultimately "fail to deliver returns."

Solomon said he is uncertain whether a bubble has already formed and expects a market correction over the next 12-24 months.

Deutsche Bank's latest research shows that online search volume for "AI bubble" has plunged from its peak on August 21 to 15%. The bank's analysts note that this "bubble of AI bubble discussions" burst reflects typical patterns seen in previous bubbles.

Deutsche Bank’s natural language analysis found that concern in the English-language media about AI investments fell from 7.3 out of 10 in the last week of August to the current 5.1. Discussion on Reddit follows the same trend.

Deutsche Bank Warns Against Market Timing, Recommends Long-Term Holding

Deutsche Bank stresses that identifying a bubble is almost impossible because nobody can agree on the precise definition of "asset prices significantly higher than intrinsic value." Historical experience shows that bubbles are not linear processes and generally play out over several rounds of ups and downs.

Citing the internet bubble as an example, Deutsche Bank notes that in the five years before the Nasdaq peaked on March 10, 2000, there were seven pullbacks of more than 10%. More importantly, even after bubble discussions became common, the market could keep rising for a long time.

On November 19, 1998, when the Nasdaq index was just under 2,000, Michael Murphy of Murphy Investment Management warned of a "serious bubble"—yet the market climbed for another 16 months before finally peaking above 5,000.

The core advice from Deutsche Bank is to avoid trying to time the market. Their data shows that if you invested $10,000 at the start of 1996 and held until June this year, it would be worth over $170,000. But missing just the 10 best trading days would cut returns in half; missing the 20 best trading days would reduce returns to a quarter.

Even more crucially, the best and worst trading days often occur close to each other. Of the 10 best trading days between 1996 and this June, 5 were within a week of the 10 worst days. This shows that precise market timing is extremely difficult.

Deutsche Bank concludes that the market can remain "irrational" longer than investors can remain solvent. The bank recommends investors adopt a long-term holding strategy to earn the risk premium necessary to compensate for equity investment risks.

Risk Warning and DisclaimerThe market carries risks, investment needs caution. This article does not constitute personal investment advice and does not take into account individual users' investment objectives, financial situation, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article are suitable for their particular situation. If you choose to invest accordingly, you do so at your own risk.