What Happens When The AI Bubble Bursts

Image Credit: Generated with Gemini

Big Tech announced $600 billion in artificial intelligence infrastructure spending for 2026 in the first week of February, and markets responded by erasing $1 trillion in software and data analytics stock value within days (Economic Times, 2026). 

Amazon revealed plans to spend $200 billion on AI capital expenditure, $50 billion more than analysts expected, and its stock dropped 7% (BBC, 2026). Alphabet said capital spending could double this year, sending its shares down 8% before recovering (Economic Times, 2026). Meta tripled its debt load in a single month to finance AI infrastructure (FX Empire, 2025). The frantic build out has drawn comparisons to the dot-com bubble, and investors are no longer rewarding spending for spending's sake.

The recent selloff follows a pattern that played out in 1999 and 1929, when speculative valuations pulled forward years of projected earnings without pricing in the risk that those returns might never materialize (Business Insider, 2024). Goldman Sachs notes that $19 trillion in market capitalization is running ahead of demonstrable economic impact, and the bank cites five danger signals that mirror the late 1990s tech boom, including peaking investment, falling profits, rising debt, Federal Reserve rate cuts, and widening credit spreads. An MIT study found that 95% of companies see zero return on their generative AI investments despite collectively spending $30 billion to $40 billion. Bain projects that AI will need to generate $2 trillion in annual revenue by 2030 just to justify current infrastructure spending, which is more than the combined revenues of America's largest tech firms in 2024 (FX Empire, 2025).

The most immediate pressure landed on software and data analytics firms. The S&P North American Software Index posted a 15% decline in January, its worst month since October 2008, and February selling accelerated after Anthropic released free plug-in software tools that threatened existing business models (Fintool, 2026). Salesforce fell 26%, ServiceNow dropped double digits, and Thomson Reuters experienced a record one day plunge (BBC, 2026). Traders coined the term SaaSpocalypse to describe approximately $300 billion in market value that evaporated from software companies in 48 hours (Forbes, 2026). Private credit firms with exposure to technology loans also cratered, with Blue Owl, TPG, Ares Management, etc., all falling by double digits as investors worried about defaults in the software sector (Fintool, 2026).

The comparison to the dot-com era shows both similarities and differences. The Nasdaq Composite peaked at a price to earnings ratio of 60.1 in March 2000, while today's Nasdaq trades at 26.4 times earnings. Dot-com firms often had no sales, making price to sales ratios undefined, whereas current AI leaders trade on strong growth with actual revenue. However, AI draws 53% to 58% of all venture capital globally, a level of concentration that exceeds the late 1990s and indicates bubble dynamics. Valuations per employee at AI startups range from $400 million to $1.2 billion, a metric with no dot-com parallel that signals intense overheating (Intuition Labs, 2026).

If the bubble bursts, the correction will likely unfold as a rolling reset similar to 2000 through 2002 rather than a sudden collapse. The weakest players fail first and drag parts of the system with them while Big Tech survives due to diversified revenue streams and strong balance sheets. The casualties will be the unicorns valued at hundreds of billions without paths to profitability, the developers relying on off balance sheet financing structures, and the investors who believed the hype without verifying the financial fundamentals. Goldman Sachs estimates that AI capital expenditure now accounts for 0.8% of GDP, still nearly half the 1.5% reached during comparable tech booms over the past 150 years, which suggests the spending could continue if demand holds (Investing.com, 2026).

The timing depends on whether companies can keep funding losses and lenders keep extending credit long enough for the sector to grow into its valuations. Ruchir Sharma, chairman of Rockefeller International, argued that the burst of all bubbles stems from the same factor, higher interest rates, and predicted the AI bubble may burst in 2026 once rising inflation forces the Federal Reserve to reverse course (36Kr, 2026). Whether that happens this year or stretches into 2027 will determine if today's infrastructure spending looks prescient or reckless in hindsight.