Just this past week, we’ve seen a massive “Capex (Capital Expenditure) Reality Check.” The “Big Four” (Alphabet, Amazon, Meta, and Microsoft) have announced staggering investment plans for 2026, totaling roughly $650 billion. There is a growing skepticism about how companies are using “AI” as a cover for traditional business moves. Amazon alone just spooked the market by announcing a $200 billion spend on AI infrastructure and robotics for this year.

This month has been brutal for the “AI at all costs” narrative. The $650 Billion Bet: Amazon’s stock recently tumbled 10% in a single night because, despite making money, investors are terrified that $200 billion spent on “AI infrastructure” is a bridge to nowhere. The fear is that we are building massive power plants for a city where nobody has moved in yet.

“AI Washing” vs. Real Productivity

Many firms have been blaming layoffs on “AI efficiencies.” However, analysts are calling this “AI Washing.” They argue that in many cases, companies are actually cutting staff due to high interest rates or pandemic-era overhiring, and simply using the AI narrative to keep their stock prices from tanking. Experts (like those at Forrester) point out that actually replacing a human worker with a mature AI application takes 18–24 months of “heavy lifting” that many companies haven’t actually done yet.

“Anthropic Shock” & Disruption Fears

While the “bubble” might refer to stock prices, the disruption is very real. This month, shares in traditional software and data services (like Pearson and RELX) dropped significantly. Why? Startups like Anthropic launched agentic tools that can automate specific, high-value tasks like contract legal review or tax strategy faster than legacy software. This is creating a “hollowed out” feeling in the software sector. Investors are rotating out of “old tech” that AI might replace and into the “picks and shovels” (like Nvidia/chips) or the AI-native companies.

Comparison to the Dot-com Bubble

The parallels to the year 2000 are being drawn daily:

• Then: Companies added “.com” to their names to double their market cap.
• Now: Companies are adding “AI-powered” to everything.
• The Difference: Unlike 1999, the giants leading this charge (Microsoft, Alphabet) are actually highly profitable and have mountains of cash. The “bubble” isn’t necessarily that the tech is fake it’s that the valuation assumes AI will double global productivity in a decade, which is a very high bar to clear.

The bubble is going to burst and the cracks are visible. Investors are pouring money into it, but people are not using it, so the industry is going to shift from “Hype” to “Utility.” The cracks aren’t just visible; they are starting to affect the biggest balance sheets in the world. While 90% of companies say they are “using” AI, the actual usage is often just employees playing with chatbots or minor coding help.

Gartner’s 2026 analysis shows Generative AI is officially entering the “Trough of Disillusionment.” The novelty has worn off, and businesses are finding that “hallucinations” and high subscription costs make it hard to justify replacing human workflows entirely.

“AI Washing” as a Smokescreen

The industry is shifting, and one of the ways it’s hiding the “bursting bubble” is through AI Washing. Companies like Salesforce and Amazon have cut thousands of jobs this month, claiming it’s for “AI efficiency.” Analysts are calling foul. Many of these companies are actually trimming fat from pandemic-era overhiring or reacting to high interest rates, and they are simply using “AI” as a buzzword to make the layoffs look “strategic” to the stock market instead of “desperate.”

The Shift to “Deterministic” Tech

Because the “General AI” bubble is leaking, we are seeing a massive pivot this month: Companies like Salesforce are actually moving away from pure generative models (the ones that talk like humans) and toward “deterministic” automation. They want tools that do one specific thing (like looking up a bank routing number) with 100% accuracy, rather than a bot that “thinks” it knows the answer.

What this means for data websites

In a “bubble burst” scenario, Data is King. The speculative money for “AI that can write poetry” is drying up, but the need for verified, structured data actually increases. AI agents can’t function without the “ground truth” data that niche sites provide. The industry is moving from “AI as a God” to “AI as a Utility.” If it doesn’t save money or make money this quarter, it’s getting cut.

In many ways, this is starting to look like 2008, but with a different “poison” in the system. Back in 2008, the “trash” was hidden in housing debt (subprime mortgages). In 2026, the “trash” is unproductive AI infrastructure debt.

Just like banks in 2007 couldn’t stop lending, tech giants today can’t stop spending. It’s not just cash; a lot of this is fueled by private credit. Morgan Stanley recently estimated that debt for data centers could exceed $1 trillion by 2028. If the AI revenue doesn’t show up, that debt becomes “toxic” just like the bad mortgages of 2008.

There’s a growing fear of “circular funding,” where Big Tech companies invest in AI startups, who then use that same money to buy cloud services back from the Big Tech companies. This inflates revenue numbers without any actual outside customers using the tech.

In 2008, the government could bail out banks. In 2026, the problem is deeper:

• We are hitting a wall with electricity. Data centers are already eating up 40% of the power in cities like Frankfurt. We can’t just “print” more electricity like we printed money in 2008.
• People aren’t using it. This is creating a “ghost economy” where billions are spent on chips and servers that sit idle because the average person doesn’t need a chatbot to write their emails every single day.
• We’ve seen over 82,000 tech layoffs in just the first six weeks of 2026. Companies are cutting human staff to pay for their AI “gambles,” which reduces consumer spending power a classic recipe for a recession.

And when the “hype” bubble pops, only the utilities survive.

In 2008, the banks that survived were the ones with real deposits and boring, stable assets. Same now: the boring, stable asset banks will survive.

Regarding the internet, in 2026 the verified, “ground truth” data will survive. If the bubble bursts, the “dreamers” go bankrupt, but the “mapmakers” stay in business.

In this scenario, the ones who poured billions into AI won’t survive. The “bubble” is big enough that after the crash, banks will stop lending money all over the world to businesses, and we will see a huge depression and inflation rising from the ground up.