AI Boom and Internet Bubble: Comparisons and Insights Spanning 25 Years

experienced its largest single - day market value decline of 17%, with a market value evaporation of approximately \(600 billion, setting the highest record for the single - day market value loss of a single company in the history of the US stock market. The trigger for this plunge was the release of a low - cost, open - source large model, DeepSeek - R1, by the Chinese company DeepSeek. In 2023, NVIDIA’s stock soared all the way, increasing by 239%; in 2024, it rose again by 171%. The rocket - like rise in its stock price has raised doubts among some investors about whether the valuation has reached its peak. Satya Nadella, the CEO of Microsoft, a major supporter of OpenAI, clearly expressed his concern in a recent interview: "Current AI companies will eventually collapse and disappear if they cannot bring about real GDP growth and if there is no real demand to support the products they develop." Is the current frenzy in the AI market and the soaring stock prices based on the true fundamentals of companies or on irrational exuberance? This epic single - day sharp decline is more reminiscent of the bubble in the Internet era 25 years ago. When the NASDAQ index climbed to its all - time high of 5048 in March 2000, Wall Street never anticipated that the Internet revolution symbolized by ".com" would evaporate \)6.5 trillion in market value within two years, causing Amazon’s stock price to plummet by 90% and Sina’s stock price to drop by more than 98%. Today, as a representative of Chinese companies, DeepSeek’s potential in technological breakthroughs and cost control has not only intensified global technological competition but also triggered deep - seated reflections on the valuation bubble in the AI industry.
Is a Capital Bubble Inevitable in Technological Revolutions? And Which Companies Can Cross the Cycle After the Bubble Bursts?
Although 25 years apart, these two major waves show profound similarities and differences in market scale and the survival logic of enterprises. This article aims to deeply analyze the inevitability and importance of capital bubbles in technological revolutions by comparing the Internet bubble 25 years ago with the current AI investment boom, and to explore whether the AI era faces a similar risk of decline, as well as what corporate characteristics can help institutions cross the cycle. With the rapid rise of Chinese companies in the global AI competition and their continuous expansion of global influence, this technological competition is shifting from the monopolistic pattern of Silicon Valley unicorns to a diversified competition ecosystem that is very different from the Internet era.
The ".com" Era
On August 9, 1995, Netscape went public through an IPO with an EBITDA margin of - 28.11% and a return on equity (ROE) of - 26.14%. On that day, its stock price soared from the offering price of \(28 to \)75 and finally closed at $58.25 per share. As the first Internet company to bring the "unprofitable but high - growth" model to the mainstream through an IPO, this landmark event is considered one of the beginnings of the Internet bubble. It broke the traditional valuation system centered on profitability, making investors start to pay attention to user growth, technological leadership, and market potential. This trend led to a large number of unprofitable start - ups flooding into the capital market in the late 1990s, ultimately forming a bubble.

From 1995 to 2000, the NASDAQ index skyrocketed by 573% in five years, during which representative companies such as Amazon, AOL, Netscape, eBay, Yahoo, and Google were born. At the same time, more than a thousand Internet companies, including Sina, Tencent, and Alibaba, emerged in China. On March 10, 2000, the NASDAQ index reached its then - all - time high of 5048.62 points, but by October 9, 2002, it dropped to 1114.11 points, which means that the NASDAQ index fell by approximately 78% from its peak to the trough. During the continuous decline for more than two years, nearly 5,000 Internet companies in the United States went bankrupt.
The Three Giants That Fell in the Bubble
Yahoo, one of the most representative giants in the early days of the Internet, its rise and fall reflect the typical trajectory of the first - generation Internet companies. As a leading portal at that time, Yahoo innovated the advertising business model by pioneering the hierarchical directory navigation system - becoming the first to transplant the traditional media advertising model to the Internet. Its advertising revenue accounted for 90% of its revenue in 1999. However, the company failed to continuously update its business model, adhered to a single advertising revenue structure for a long time, failed to effectively expand new businesses (such as e - commerce and social networking), and also lost to Google in search technology due to a lack of technological moats and failure to invest in key technological development. Its strategic conservatism made it fail to successfully transform in the Web 2.0 era (user - generated content) and missed major opportunities to acquire Google, Facebook, etc. These factors led to Yahoo's inability to maintain its competitiveness after the Internet bubble burst. From its peak market value of \(125 billion in 2000, it plummeted to about \)1 billion within two years, and was finally acquired by Verizon for \(4.8 billion in 2016, which marked its symbolic end.
AOL was the first to launch a \)19.95 monthly Internet access service in 1996, opening the "always - on" Internet era. The company implemented a marketing strategy of "free trial + CD bombing" from 1993 - by mailing hundreds of millions of installation CDs to potential users, its user base surged from 200,000 in 1993 to a total of 34 million registered users in 2000. However, the company stubbornly continued the dial - up subscription model in the later period and ignored the increase in broadband penetration, ultimately resulting in a large - scale loss of users. The merger with Time Warner is considered one of the most failed mergers in history: AOL was valued at approximately \(182 billion when it announced its merger with Time Warner in 2000. But after the merger, due to intensified cultural conflicts, combined with neglect of product technology transformation (especially in the broadband field), management chaos, and failed resource integration, the synergistic effect was not achieved. These problems led to the loss of core talents, a decline in advertising revenue, and financial pressure after the Internet bubble burst, and its market value shrank significantly. Finally, in 2015, AOL was acquired by Verizon for about \)4.4 billion.

Netscape launched the first commercial browser, Navigator, in 1994, once accounting for 80% of the market share, and its market value soared to $2.9 billion on the first day of listing in 1995. However, after being acquired by AOL in 1998, due to internal management problems and cultural conflicts, its core team left one after another, and the company's technological development came to a standstill. Although the open - source strategy gave birth to the Mozilla project, the company failed to respond to market competition in a timely manner. Microsoft quickly captured market share by bundling the IE browser into the operating system and adopting a free strategy, making Netscape's business model difficult to support long - term development. Although Netscape no longer exists as a company, it promoted the development of the early Internet and laid an important foundation for modern web technology.
Enterprises in the Post - bubble Era
During the Internet bubble period (1995 - 2003), although these representative Internet companies experienced a cycle of rapid rise and bubble burst, and each declined or even went bankrupt for different reasons, overall, the excessive optimism of the capital market about emerging technologies and the failure of traditional valuation models were the key factors. The negative - profit model made it difficult for traditional financial indicators (such as cash flow and price - to - earnings ratio) to effectively evaluate Internet companies, and instead relied on the "eyeball economy" and user growth. Coupled with weak corporate culture management, the fanaticism of the capital market, and the FOMO (fear of missing out) psychology, venture capital and the secondary market fell into irrational expansion. If a company lags behind in technology and lacks a sustainable commercial profit model, it is difficult to survive during the market bubble adjustment stage. However, there are also some companies that have become a few survivors and grown stronger, such as Amazon in the United States. In addition, some emerging companies not only avoided the bubble but also took advantage of the reshuffle period brought about by the industry adjustment to overtake on a curve, such as Google, Facebook, Tencent, and Alibaba.
Although Amazon's stock price plummeted by 90% during the Internet bubble, its founder, Jeff Bezos, always emphasized long - term innovation rather than short - term profits. Through diversified business layout, he invested funds in the construction of the logistics network and technological R & D. In 2006, Amazon launched the AWS cloud computing service, opening up a new growth point and gradually becoming the leader in the global cloud service. During the bubble burst, Amazon accumulated capital reserves through multiple financings, ensuring that it had sufficient cash flow to maintain operations, which provided key support for the company to survive the industry trough.
Google did not go public until 2004, so it successfully avoided the speculative frenzy in 1999. During this period, the company used venture capital to develop steadily, without blindly pursuing short - term expansion goals, but focused on breaking through search technology. After 2000, with the PageRank algorithm, Google replaced Yahoo and established its monopoly position in the search field. At the same time, the company innovated its advertising model by launching AdWords (in 2000) and AdSense (in 2003), bringing stable and substantial revenue to the company. In the later stage, the dual strategies of developing cloud computing and AI constructed a complete value chain from traffic acquisition to technological empowerment.
The Core Elements of Enterprises That Cross the Cycle in the "AI" Era

Looking back at the evolution of the Internet from prosperity to bubble and then to rebirth 25 years ago, it completely follows the Gartner curve of "technology emergence - inflated expectations - bubble burst - recovery and maturity". This historical cycle reveals the core survival elements of "survivors" in technological revolutions: not only the need for continuous technological breakthroughs and sustainable profit models, but also the support of a capital chain resistant to capital fluctuations, and a mature and stable governance decision - making system is its core support. Enterprises with these characteristics can not only cross the cycle and achieve long - term prosperity but also form the key elements for enterprises to break through and win in the red - ocean competition of the Internet.
The Valuation Dilemma of Start - ups from 0 to 1 Still Exists
During the Internet period, a large number of start - ups achieved technological breakthroughs from 0 to 1, but the exit mechanism for enterprises in this period was dominated by IPOs. For example, in 1999, the total venture capital in the United States reached \(54 billion, of which 62% flowed to unprofitable enterprises - a typical case was Sequoia Capital, which achieved excess returns by betting on projects such as Google and PayPal and relying on IPO exits. At that time, the number of enterprise listings showed explosive growth, and in 2000 alone, 442 companies completed IPOs on the NASDAQ. However, due to the lack of stable profit data as an anchor for valuation and the difficulty of quantifying the value of enterprises brought about by technological innovation, the valuation logic of these companies deviated significantly from traditional financial models (such as the discounted cash flow method). This contradiction made it difficult for IPO pricing to truly reflect the core value of enterprises, laying hidden dangers for the bubble. When the bubble burst, not only benchmark enterprises such as Netscape and Yahoo fell, but also WorldCom, the second - largest long - distance telephone company in the United States at that time, went bankrupt due to blind expansion leading to the deterioration of its financial structure. These cases reveal the systematic deviation between market value and the true value of enterprises, and this is the core mechanism of bubble generation: the decoupling of capital pricing from fundamentals.
The current AI era faces a similar dilemma. Although the technological revolution also gives birth to breakthrough innovations from 0 to 1, the valuation of AI enterprises still lacks a mature framework for reference. Different from the path of relying on IPO exits during the Internet period, the current stage tends to private financing and M & A integration (such as Microsoft's acquisition of Nuance and Google's acquisition of DeepMind). However, the essential contradiction of enterprise value assessment has not changed - the split between the long - term income potential created by technological innovation and short - term financial performance is still the core problem of capital pricing. This uncertainty not only drives the AI investment boom but also may repeat the historical valuation bubble.
As of February 2025, the weekly active user number of ChatGPT has reached 400 million, a 33% increase compared to 300 million users in December last year. The company is expected to achieve revenue of \)11 billion this year. The valuation of OpenAI has increased from \(150 billion in 2024 to \)340 billion in the latest round of financing. However, according to media analysis of OpenAI's financial documents, the company may lose \(14 billion in 2025 and is expected to become profitable in 2029, with revenue reaching \)100 billion at that time. During the period from 2023 to 2028, the company is expected to accumulate a total loss of \(44 billion.

OpenAI is in a strategic expansion period of sacrificing profits for hegemony. Whether its profit model is sustainable depends on its ability to maintain a technological generation gap and the speed of forming a commercial closed - loop. If it can break through the computing power shackles through self - developed hardware and establish a profit - sharing system in vertical fields such as healthcare and education, it may achieve a transformation path from loss to a trillion - dollar market value; however, the valuation of OpenAI is still based on the prediction of future cash flows, and the risks caused by such uncertainties cannot be ignored. Once it fails to maintain technological updates and a commercial profit model and the capital chain breaks, it is not excluded that it will become a specimen of an AI bubble under the catalysis of capital.
As a dark horse in the industry that emerged in early 2025, DeepSeek has had a significant impact on the market. Its subversion stems from algorithm optimization and innovation - reducing the pre - training cost to less than 1/10 of the training cost of industry - equivalent performance models, redefining AI economics. The market has a sharp divergence in the valuation of DeepSeek: the prediction range is from \)1 billion to $20 billion, and some even believe that its value is at least half of that of OpenAI. This huge difference confirms the common problem of the ambiguity of the valuation system for innovative enterprises. Currently, its core advantage lies in the support of its parent company, Huafang Quant (a leading quantitative hedge fund in China) - the latter had reserved more than 10,000 NVIDIA GPU computing power clusters as early as 2021, enabling it to bypass the bottleneck of relying on external financing for traditional enterprises and focus on long - term technological R & D. Currently, the focus of risk is not the pressure on the capital chain but how to maintain the speed of technological iteration, build exclusive barriers to improve developer retention rates, and the sustainability of the open - source business model. The high uncertainty of these variables is the key constraint suppressing the valuation consensus.
Will the AI Market Decline? The Historical Mirror of the Time Window
The Involvement of Chinese Enterprises Sparks Valuation Thinking in AI

Although both the Internet period and the AI period show the characteristics of capital fanaticism, and there are commonalities in the valuation dilemmas of start - ups, there are still differences between the two. The way Chinese enterprises participate is causing structural changes: in the Internet era, Chinese enterprises had a core dependence on the technology ecosystem dominated by the United States, resulting in a lack of discourse power in technological standards and market rules; while in the AI era, Chinese enterprises represented by DeepSeek and Tongyi Qianwen, through engineering innovation and breakthroughs in vertical application fields, not only challenge the traditional closed - source model but also directly impact the business model centered on "pay - for - results". For example, ChatGPT relies on a subscription system and API fees (such as a $20 - per - month premium service), while DeepSeek adopts a free strategy, forcing competitors to reduce prices or adjust their business models. This "cost - performance revolution" may compress the overall profit margin of the industry and trigger short - term valuation shocks. This "cost - performance revolution" not only impacts the existing order but also gives birth to a new ecosystem. The reduction in technological barriers is attracting more participants to quickly enter the AI vertical field, expanding the application scope and prolonging the commercialization cycle, thus making the market adjustment tend to be more moderate. In the long run, the intensifying competition triggered by enterprises such as DeepSeek is essentially a dynamic game between technological democratization and monopoly interests - by promoting technological inclusiveness and productivity progress, it injects long - term momentum into economic development.
The Inevitability of the Decline
During the Internet bubble period, the tightening of liquidity and the negative - profit model were the main risks. In 2001, the net profit of NASDAQ - listed companies plummeted by 89%, completely exposing the vulnerability of unprofitable enterprises in the ebb of capital. The challenges in the current AI market are more complex: computing power bottlenecks (such as the shortage of high - end NVIDIA chips), the lag in the commercialization and profitability of applications (most AI enterprises have not yet found a sustainable monetization path), and geopolitical technological decoupling (such as China - US chip control) - which form a significant contrast with the growth environment driven by global technological dividends during the Internet period. The superposition of such risks and the failure