The chickens are coming home to roost for Big Tech’s AI spending spree.
Over $1 trillion in market value has evaporated from technology stocks in the past week, triggered by a seemingly innocuous Friday announcement from Anthropic about new plugins for its Claude AI assistant. What started as a routine product launch morphed into a full-blown market reckoning over whether artificial intelligence will cannibalize the very software companies betting billions on building it.
The selloff wasn’t just brutal. It was surgical. Software stocks bore the brunt of the damage, with an iShares software ETF plunging nearly $1 trillion over seven trading days. Microsoft, Nvidia, Oracle, Meta, Amazon, and Alphabet all saw their shares decline as investors confronted an uncomfortable truth: The companies pouring $660 billion into AI infrastructure this year might be funding their own obsolescence.
The Spark That Lit The Fire
Anthropic’s crime, if you can call it that, was releasing industry-specific plugins for Claude Cowork on January 30. The additions covered legal work, finance, product marketing, and data analysis. On paper, nothing earth-shattering. In practice, a four-paragraph press release that moved markets like an earthquake.
Thomson Reuters collapsed 15.83% on Tuesday in its worst single-day drop on record. LegalZoom cratered 19.68%. RELX, the London-based parent of LexisNexis, shed 14%. FactSet, MSCI, and a parade of enterprise software darlings followed them down.
By Tuesday’s close, $285 billion had been wiped from software, financial services, and asset management stocks. Asset managers with exposure to software companies—Apollo, Ares, Blackstone, Blue Owl, Carlyle, KKR—tumbled between 3% and 11% as credit concerns spread.
The message from markets was clear: If Anthropic can build legal workflow tools in-house, what’s stopping them from gutting the entire software-as-a-service ecosystem that built Salesforce, Bloomberg, and every other enterprise giant?
Not Just Software In The Crosshairs
While software took the hardest hit, the contagion spread to Big Tech itself. These are the companies investors believed were AI winners, not casualties.
Salesforce is down 26% in 2026, making it the second-worst performer in the Dow Jones Industrial Average. CEO Marc Benioff told Fortune last year the company won’t hire additional software engineers, customer service agents, or lawyers because of AI tools. The strategy isn’t protecting the stock.
Oracle’s relationship with OpenAI has become a liability rather than a badge of honor. The company announced plans to raise $50 billion in debt and equity to fund AI infrastructure, then felt compelled to publicly reassure markets it’s “highly confident in OpenAI’s ability to raise funds and meet its commitments.” The stock fell 2.79% that day. When you have to issue bank-run language about your biggest customer, you’ve already lost.
Microsoft shares dipped despite beating earnings estimates. The problem? CFO Amy Hood said capital expenditure growth would accelerate in fiscal 2026 after previously signaling it would slow. Investors who thought peak spending was approaching got whiplash instead.
The equal-weight S&P 500 hit a record high this week. That tells you everything you need to know about where the damage is concentrated: at the top of the market cap ladder.
The Math That Doesn’t Add Up
Big Tech’s AI spending projections would make a venture capitalist blush. Morgan Stanley estimates these companies will spend $3 trillion on AI infrastructure through 2028. Bank of America warns that capital expenditures will consume 94% of operating cash flows minus dividends and share buybacks in 2025 and 2026, up from 76% in 2024.
That’s before you account for the debt. Meta and Oracle alone issued $75 billion in bonds and loans in September and October 2025 to fund AI data center buildouts. Oracle is carrying roughly $100 billion in debt while betting its future on OpenAI’s ability to keep raising money and paying bills.
The circular financing arrangements have become impossible to ignore. OpenAI committed $300 billion to Oracle over five years for computing power. Oracle’s market value soared by nearly one-third of a trillion dollars on the announcement. OpenAI’s valuation nearly doubled from $300 billion to $500 billion in less than a year. Meanwhile, CNBC reported Oracle expects to “lose considerable sums of money” on data center rentals to OpenAI, including a $100 million loss last quarter.
Nvidia, AMD, Broadcom, and the chip makers provide financing to AI companies, who use that money to buy more chips from the same vendors. OpenAI enters deals with CoreWeave worth tens of billions, renting chip capacity in exchange for CoreWeave stock, which OpenAI could theoretically use to pay those same rental fees.
It’s vendor financing all the way down, and the last time this playbook ran at scale was during the dot-com bubble. Ask anyone who held WorldCom bonds how that ended.
The Bubble Question No One Wants To Answer
Is this an AI bubble? Wall Street remains split.
Goldman Sachs and Morgan Stanley analysts have largely dismissed bubble concerns, pointing to strong fundamentals and modest valuations compared to the dot-com era. Nvidia CEO Jensen Huang called fears that AI will replace software “illogical” and insisted “time will prove itself.”
JPMorgan’s Mark Murphy described it as an “illogical leap” to believe a single plugin from an AI model would “replace every layer of mission-critical enterprise software.”
But the counterarguments are piling up faster than data center construction permits.
A Nanda report from MIT’s Media Lab found that despite $30 billion to $40 billion in enterprise investment into generative AI, 95% of organizations are getting zero return. The Bank of England warned last year about growing risks of global market correction due to overvaluation of leading AI firms, specifically citing OpenAI’s valuation tripling from $157 billion to $500 billion.
Anthropic CEO Dario Amodei, whose company inadvertently triggered this week’s selloff, has been blunt about disruption. He told Axios in June that AI could displace half of all entry-level white-collar jobs in the next one to five years.
Michael Burry, the investor who shorted the housing market in 2008, is now betting against Nvidia. His thesis centers on circular deals and accounting gymnastics. “True end demand is ridiculously small,” Burry wrote on X. “Almost all customers are funded by their dealers.”
What Happens Next
Two scenarios emerge from the wreckage.
In the optimistic case, this is a healthy correction. AI continues advancing, enterprise software adapts rather than dies, and the massive infrastructure buildout eventually generates returns that justify the spending. Software stocks rebound as companies demonstrate they can coexist with AI tools rather than be replaced by them.
Wedbush Securities articulated this view in a research note, arguing enterprises won’t “completely overhaul tens of billions of dollars of prior software infrastructure investments to migrate over to Anthropic, OpenAI, and others.”
The pessimistic case looks more like 2000. Spending continues accelerating while revenue growth disappoints. Debt loads become unsustainable. A major AI player—perhaps one overly dependent on circular financing—stumbles. The dominoes fall faster than anyone expects. Software companies discover their moats were shallower than advertised.
James St. Aubin, chief investment officer at Ocean Park Asset Management, offered the most sobering take: “My biggest fear is that this is a canary in the coal mine for the labor market.”
Because if AI can replace paralegals, junior associates, customer service agents, and entry-level analysts, the software companies aren’t the only ones who should be worried. The workers are next.
The Bottom Line
Big Tech spent the past three years convincing investors that AI would create unprecedented value. The infrastructure buildout—$660 billion this year alone—was framed as a land grab, a race to establish dominance in the most transformative technology since the internet.
This week’s selloff suggests investors are starting to ask harder questions. Not whether AI is transformative. That’s obvious. But whether the companies spending trillions to build it will capture the value, or whether they’re building the railroads while someone else gets rich selling tickets.
Software stocks lost nearly $1 trillion in value because Anthropic released plugins that automate legal and finance workflows. If that doesn’t make you question the sustainability of software-as-a-service pricing models, you’re not paying attention.
The AI boom continues. The AI bubble debate just got a lot more serious.
