AI stocks compared to the dot-com bubble: CAPE stands at 38, with concentration exceeding levels seen in 2000, yet the companies are genuinely profitable.

      TL;DR

      The Shiller CAPE ratio is currently between 38 and 40, the second-highest in 155 years, surpassed only by the dot-com era peak of 44.19. Additionally, the top 10 companies in the S&P 500 account for almost 50% more market concentration than during the dot-com phase. However, AI firms are significantly more profitable than those in the past, with Nvidia alone reporting $120 billion in net income, and the tech sector currently trades at 30 times forward earnings compared to 50 times in 2000 at the market's peak. The outcome hinges on whether annual hyperscaler capital expenditures of $660-690 billion will yield returns that validate the investment, a determination that remains pending until the infrastructure cycle demonstrates results.

      The Shiller cyclically adjusted price-to-earnings ratio (CAPE) for the S&P 500 fluctuates around 38 to 40, depending on the day. In the past 155 years, the CAPE has only reached a higher level once: in March 2000, peaking at 44.19 just before the Nasdaq began a drop that would see it lose 78% of its value over the subsequent two and a half years. Currently, the ten largest firms in the S&P 500 represent 36% to 40% of the index's overall market value, nearly 50% higher than the 27% concentration during the dot-com peak. According to Deutsche Bank’s recent survey of fund managers, 57% of institutional investors see a potential AI valuation crash as the predominant risk to the markets. Jeremy Grantham, co-founder of GMO and known for predicting both the dot-com and housing bubbles, claims there is "slim to none" possibility that the current AI surge will not lead to a collapse. These statistics fuel comparisons to 2000, though they remain incomplete on their own.

      The cause for concern

      The structural similarities between the current AI stock surge and the dot-com bubble are significant. Market dominance has surpassed dot-com levels considerably. The performance of the Nasdaq-100 is primarily influenced by a few companies whose valuations are based on anticipated AI revenue that has yet to materialize on the level the market expects. The capital expenditures of hyperscalers like Microsoft, Google, Amazon, and Meta are projected to reach $660 billion to $690 billion by 2026, marking the largest corporate investment initiative in history outside of wartime. This expenditure is partially financed by downsizing human workforce in favor of AI infrastructure development: Meta and Microsoft reportedly laid off up to 23,000 employees while simultaneously pledging record capital spending, effectively reallocating funds from payroll to data center construction.

      Bank of America’s Savita Subramanian forecasts a year-end target of 7,100 for the S&P 500, with a bearish scenario of 5,500, anticipating a compression of multiples as earnings growth slows in the latter half of 2026. The Motley Fool has identified four indicators that it associates with bubble conditions: retail investor exuberance, concentrated speculative capital, detachment of valuations from fundamental metrics, and a compelling narrative that makes skepticism seem intellectually dubious. All four are evident today. OpenAI's $852 billion valuation reflects a company that has yet to make a profit, priced roughly twice that of Coca-Cola, a company that has generated profits since the 1890s. Accel's $5 billion AI fund, the largest in venture capital history, exemplifies the influx of capital into AI at the private market level. The public and private markets are interlinked: venture-backed AI companies secure funding at high valuations, while publicly traded AI companies increase spending to outpace their competitors, driving both valuations and capital expenditures upward.

      The basis for optimism

      A key distinction between 2000 and 2026 is the matter of profitability. During the dot-com peak, tech companies collectively were losing capital. Cisco had a P/E ratio of 200, and Pets.com recorded no earnings. The entire speculation hinged on future revenues from an internet economy that, while tangible, was years away from generating the cash flows anticipated by the market. In contrast, the companies fueling the AI rally in 2026 are some of the most profitable in history. Nvidia reported net income exceeding $120 billion for fiscal 2026, with its forward P/E ratio around 41—high, but not as excessive as Cisco's 200. The tech sector's overall forward P/E is approximately 30, down from 50 at the dot-com peak. Apple, Microsoft, Alphabet, Amazon, and Meta collectively generated $350 billion in free cash flow in their most recent fiscal years. These are not speculative ventures burning through venture capital but rather cash-generating entities reinvesting at historically high rates.

      Analyst John Higgins from Capital Economics presents a nuanced viewpoint, differentiating between a "stock bubble" and a "fundamental bubble." He posits that