The current surge in artificial intelligence (AI) investment is being likened to the dot-com boom of the late 1990s and the subsequent e-commerce expansion of the 2000s. While these comparisons hold some merit, they vary significantly in nature. The internet bubble was characterized by speculative capital chasing unsustainable business models, whereas e-commerce involved substantial losses over years to cultivate consumer habits, develop logistics networks, and build trust before achieving profitability. Today’s AI landscape appears to occupy a middle ground: underscored by real demand and utility, yet backed by strong balance sheets and abundant free cash flows.
The dot-com boom spanned from 1995 to 2000, marked by a disconnect between internet valuations and business fundamentals. The Nasdaq reached its peak in March 2000 and subsequently fell nearly 78% by 2002. However, while valuations suffered, the internet itself remained robust, merely needing time for viable business models to materialize.
The subsequent phase manifested through e-commerce, exemplified by Amazon, founded in 1994 and listed in 1997, which took nearly nine years to post a full-year profit. During this period, it invested in warehouses, fulfillment centers, delivery systems, and customer acquisition, all while consumers gradually adapted to online payments and delayed deliveries. This is the closest parallel to AI today.
Currently, consumers and businesses are forming new behaviors around AI technologies such as conversational search, software copilots, AI-assisted coding, automated workflows, and productivity tools. Just as e-commerce required nearly a decade to become mainstream, AI may need a similarly extended period before monetization aligns with its investment influx.
One of the distinct differences between the AI wave and the dot-com era lies in funding sources. In the 1990s, many speculative companies relied on external capital markets. Today, leading AI investors are profitable platform companies generating significant internal cash flows.
For instance, Alphabet generated approximately $165 billion in operating cash flow in 2025 and held about $127 billion in cash and securities, producing over $70 billion in free cash flow despite substantial capital expenditures. Meta Platforms reported nearly $116 billion in operating cash flow and maintained over $80 billion in cash and investments, yielding more than $40 billion in free cash flow. Amazon also reported about $140 billion in operating cash flow, holding around $123 billion in liquidity, although its free cash flow saw a sharp decline due to increased infrastructure spending.
This difference in funding is crucial. If AI had to rely predominantly on venture capital or debt, the investment surge would likely falter as market rates rose or sentiment soured. Conversely, the current expansion is financed by companies that have created large cash engines through ventures in search advertising, cloud computing, digital commerce, and social media. Essentially, the excess returns from past platform successes are being reinvested into AI infrastructure, allowing for a prolonged period of economic irrationality, as the backers can afford to be patient.
The pivotal question surrounding AI’s trajectory is not its reality—clearly established—but whether current expenditures yield acceptable returns. The likelihood of this being evident within the next 12 months is low, with clearer insights expected over a three- to seven-year horizon.
In the next two to three years, markets may be tolerant of robust spending if growth in usage remains solid. AI assistants, cloud inference demand, coding copilots, and enterprise adoption can support optimistic sentiments. However, smaller AI startups with insufficient differentiation may struggle first, while larger players continue their capital expenditures.
By years three to five, investors will begin pressing for monetization, evaluating whether AI contributes to search revenue, cloud margins, software pricing power, ad conversion rates, and labor productivity. If revenue growth lags behind costs, skepticism may rise.
In years five to seven, the future of AI will hinge on whether it becomes an embedded infrastructure akin to cloud computing and e-commerce logistics; otherwise, the industry may face a significant capital expenditure reset.
In conclusion, today’s AI landscape can be viewed not as a traditional bubble but rather as a cash-funded cycle focused on habit formation. Similar to the early e-commerce phase, it may take years of upfront losses or diminished returns before user behaviors stabilize. While valuations might surpass rational limits, leading to the exit of certain players, the major funders today are not fragile startups but some of the most financially robust companies in history. This distinction suggests that the AI investment cycle could persist longer than market predictions, possibly lasting five to seven years before the definitive winners and losers emerge.
(Karan Taurani is EVP at Elara Capital)
Published on April 27, 2026.
