AI-related stocks have lost $1.8 trillion in market value since October 29, 2025, marking the sharpest pullback in the sector's rapid rise.
Nvidia alone wiped out nearly $493 billion, with Meta, Microsoft, and Broadcom contributing hundreds of billions more in losses.
Signs of Strain in Late 2025
High valuations now face scrutiny as capex surges ahead of monetization.
Tech giants committed trillions to data centers and AI infrastructure in 2025—Amazon $100 billion, Microsoft $80 billion, Google $75 billion, and Meta planning over $600 billion in the coming years.
Yet revenue growth lags, prompting executives to voice concerns.
Jarek Kutylowski of DeepL noted exaggerated valuations and bubble signs on the horizon.
Picsart's Hovhannes Avoyan highlighted companies raising at tremendous levels without meaningful revenue, calling it vibe-based investing.
Historical Parallels and Current Risks
The surge mirrors past bubbles, with AI capturing a disproportionate share of market gains.
Analysts compare it to the dot-com era, where hype inflated assets before sharp corrections.
Deutsche Bank labeled summer 2025 as when AI turned ugly, warning of unsustainable growth.
Bain & Company estimates the industry needs $2 trillion in annual revenue by 2030 to justify scaling, far above current trajectories.
OpenAI reported $4.3 billion in income but $13.5 billion in losses in the first half of 2025.
Arguments for Sustainability
Demand for AI chips and infrastructure remains strong, with no widespread dark GPUs or unused capacity.
Companies continue beating earnings expectations in some quarters, and adoption in enterprise tools grows steadily.
Nvidia dismissed bubble fears, projecting massive business ahead.
Brookings Institution suggests monitoring adoption rates, product pricing, and competition as key indicators rather than declaring an imminent burst.
What Happens Next
A full burst could trigger broader market contraction, job losses in tech, and economic ripple effects, given AI's role in recent GDP growth via data centers.
Corrections often follow overvaluation, but timing remains uncertain.
Investors should watch earnings reports, revenue per watt in data centers, and real-world adoption metrics.
The current pullback may represent a healthy reset or the start of a larger unwinding—evidence points to heightened risks in late 2025.
Understanding these dynamics helps separate transformative potential from speculative excess in AI's evolution.
